diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,31866 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 5306,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2500000000000002e-07,
+      "loss": 1.3344,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5000000000000004e-07,
+      "loss": 1.2693,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.75e-07,
+      "loss": 1.3083,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 1.2993,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.25e-07,
+      "loss": 1.32,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.5e-07,
+      "loss": 1.2237,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.75e-07,
+      "loss": 1.3457,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 1.247,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.125e-06,
+      "loss": 1.2766,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.25e-06,
+      "loss": 0.2229,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3750000000000002e-06,
+      "loss": 1.2177,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-06,
+      "loss": 1.2416,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6250000000000001e-06,
+      "loss": 1.2047,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.75e-06,
+      "loss": 1.1305,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8750000000000003e-06,
+      "loss": 1.1855,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.1187,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.125e-06,
+      "loss": 1.1283,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.25e-06,
+      "loss": 0.1834,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.375e-06,
+      "loss": 1.0852,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-06,
+      "loss": 1.1729,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6250000000000003e-06,
+      "loss": 1.08,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7500000000000004e-06,
+      "loss": 1.0779,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.875e-06,
+      "loss": 1.0951,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3e-06,
+      "loss": 1.0405,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.125e-06,
+      "loss": 1.0444,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2500000000000002e-06,
+      "loss": 1.0825,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3750000000000003e-06,
+      "loss": 1.083,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5e-06,
+      "loss": 0.9669,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.625e-06,
+      "loss": 1.0429,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 0.948,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.875e-06,
+      "loss": 1.0339,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.0494,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.125e-06,
+      "loss": 0.978,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.25e-06,
+      "loss": 0.9871,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3750000000000005e-06,
+      "loss": 0.9504,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5e-06,
+      "loss": 1.0076,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.625000000000001e-06,
+      "loss": 0.9915,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.75e-06,
+      "loss": 0.9957,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.875e-06,
+      "loss": 0.9736,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 1.0025,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.125e-06,
+      "loss": 0.9763,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2500000000000006e-06,
+      "loss": 0.976,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.375e-06,
+      "loss": 0.9514,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 0.9477,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.625e-06,
+      "loss": 0.9801,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.75e-06,
+      "loss": 1.0223,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.8750000000000005e-06,
+      "loss": 0.9759,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6e-06,
+      "loss": 0.9503,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.125000000000001e-06,
+      "loss": 0.9659,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.25e-06,
+      "loss": 0.9168,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.375e-06,
+      "loss": 0.9025,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 0.9699,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.625e-06,
+      "loss": 0.9611,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.750000000000001e-06,
+      "loss": 0.9034,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.875e-06,
+      "loss": 0.9611,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7e-06,
+      "loss": 0.938,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.125e-06,
+      "loss": 0.9205,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.25e-06,
+      "loss": 0.9268,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.375000000000001e-06,
+      "loss": 0.9389,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.9056,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.625e-06,
+      "loss": 0.9276,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.75e-06,
+      "loss": 0.913,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.875e-06,
+      "loss": 0.9638,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.8983,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.125000000000001e-06,
+      "loss": 0.9533,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.25e-06,
+      "loss": 0.9103,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.375e-06,
+      "loss": 0.9937,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.5e-06,
+      "loss": 0.9226,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.625000000000001e-06,
+      "loss": 0.9068,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.750000000000001e-06,
+      "loss": 0.2447,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.875e-06,
+      "loss": 0.954,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9e-06,
+      "loss": 0.9096,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.125e-06,
+      "loss": 0.9494,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.250000000000001e-06,
+      "loss": 0.9324,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.375000000000001e-06,
+      "loss": 0.9203,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.5e-06,
+      "loss": 0.942,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.625e-06,
+      "loss": 0.9506,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.75e-06,
+      "loss": 0.9414,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.875000000000001e-06,
+      "loss": 0.9368,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.9464,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0125e-05,
+      "loss": 0.9248,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.025e-05,
+      "loss": 0.9263,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0375000000000001e-05,
+      "loss": 0.8898,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0500000000000001e-05,
+      "loss": 0.2277,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0625e-05,
+      "loss": 0.9155,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.075e-05,
+      "loss": 0.9338,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0875e-05,
+      "loss": 0.9169,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.8661,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1125000000000001e-05,
+      "loss": 0.8867,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.125e-05,
+      "loss": 0.9037,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1375e-05,
+      "loss": 0.8906,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.15e-05,
+      "loss": 0.966,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1625000000000001e-05,
+      "loss": 0.975,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1750000000000001e-05,
+      "loss": 0.913,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1875e-05,
+      "loss": 0.9903,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2e-05,
+      "loss": 0.8911,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2125e-05,
+      "loss": 0.893,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2250000000000001e-05,
+      "loss": 0.8975,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2375000000000001e-05,
+      "loss": 0.9076,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.25e-05,
+      "loss": 0.9319,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2625e-05,
+      "loss": 0.8899,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.275e-05,
+      "loss": 0.9079,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2875000000000001e-05,
+      "loss": 0.9581,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.9046,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3125e-05,
+      "loss": 0.9159,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.325e-05,
+      "loss": 0.9759,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3375e-05,
+      "loss": 0.9298,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 0.9358,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3625e-05,
+      "loss": 0.9426,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.375e-05,
+      "loss": 0.2321,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3875e-05,
+      "loss": 0.9424,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4e-05,
+      "loss": 0.9285,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4125000000000003e-05,
+      "loss": 0.9197,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.425e-05,
+      "loss": 0.916,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4375e-05,
+      "loss": 0.8597,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.45e-05,
+      "loss": 0.8471,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4625e-05,
+      "loss": 0.9421,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4750000000000003e-05,
+      "loss": 0.914,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4875000000000002e-05,
+      "loss": 0.876,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.8788,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5125e-05,
+      "loss": 0.9019,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.525e-05,
+      "loss": 0.9216,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5375e-05,
+      "loss": 0.9021,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.55e-05,
+      "loss": 0.9242,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5625e-05,
+      "loss": 0.8936,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.575e-05,
+      "loss": 0.9434,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5875e-05,
+      "loss": 0.2238,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.9027,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6125000000000002e-05,
+      "loss": 0.9169,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6250000000000002e-05,
+      "loss": 0.8991,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6375e-05,
+      "loss": 0.232,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.65e-05,
+      "loss": 0.9042,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6625e-05,
+      "loss": 0.8947,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.675e-05,
+      "loss": 0.9206,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6875e-05,
+      "loss": 0.9598,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7e-05,
+      "loss": 0.9742,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7125e-05,
+      "loss": 0.9011,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7250000000000003e-05,
+      "loss": 0.9198,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7375000000000002e-05,
+      "loss": 0.9037,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 0.9184,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7625e-05,
+      "loss": 0.8825,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.775e-05,
+      "loss": 0.8871,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7875e-05,
+      "loss": 0.915,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8e-05,
+      "loss": 0.9384,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8125e-05,
+      "loss": 0.9369,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.825e-05,
+      "loss": 0.854,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8375e-05,
+      "loss": 0.9309,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8500000000000002e-05,
+      "loss": 0.2524,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8625000000000002e-05,
+      "loss": 0.9405,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 0.8603,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8875e-05,
+      "loss": 0.8927,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9e-05,
+      "loss": 0.8601,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9125000000000004e-05,
+      "loss": 0.8548,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.925e-05,
+      "loss": 0.8418,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9375e-05,
+      "loss": 0.9133,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.95e-05,
+      "loss": 0.8605,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9625e-05,
+      "loss": 0.8701,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9750000000000002e-05,
+      "loss": 0.8924,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9875000000000002e-05,
+      "loss": 0.9211,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.9168,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999998136496676e-05,
+      "loss": 0.8919,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999254598739e-05,
+      "loss": 0.9235,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999983228474224e-05,
+      "loss": 0.8728,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999970183960658e-05,
+      "loss": 0.8985,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999995341245155e-05,
+      "loss": 0.8898,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999993291395315e-05,
+      "loss": 0.8828,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99999086884731e-05,
+      "loss": 0.8526,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999880736020426e-05,
+      "loss": 0.9414,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999984905660555e-05,
+      "loss": 0.8968,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999813650240275e-05,
+      "loss": 0.9504,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999774516937797e-05,
+      "loss": 0.8609,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999731656712703e-05,
+      "loss": 0.868,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999685069580968e-05,
+      "loss": 0.9249,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999634755559952e-05,
+      "loss": 0.9037,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999958071466841e-05,
+      "loss": 0.9328,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999952294692648e-05,
+      "loss": 0.8687,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999461452355696e-05,
+      "loss": 0.9297,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999396230978972e-05,
+      "loss": 0.8485,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999327282820622e-05,
+      "loss": 0.8731,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999254607906338e-05,
+      "loss": 0.9085,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999917820626321e-05,
+      "loss": 0.899,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999909807791971e-05,
+      "loss": 0.2573,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99990142229057e-05,
+      "loss": 0.8916,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999892664125244e-05,
+      "loss": 0.8944,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999883533299256e-05,
+      "loss": 0.9031,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998740298160105e-05,
+      "loss": 0.9163,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998641536790485e-05,
+      "loss": 0.8934,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999853904892051e-05,
+      "loss": 0.9506,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998432834588375e-05,
+      "loss": 0.9203,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998322893833677e-05,
+      "loss": 0.8874,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999820922669738e-05,
+      "loss": 0.8928,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999809183322185e-05,
+      "loss": 0.887,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997970713450842e-05,
+      "loss": 0.2473,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997845867429494e-05,
+      "loss": 0.9094,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999771729520434e-05,
+      "loss": 0.8981,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99975849968233e-05,
+      "loss": 0.8145,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997448972335674e-05,
+      "loss": 0.9251,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997309221792166e-05,
+      "loss": 0.8287,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999716574524486e-05,
+      "loss": 0.8516,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997018542747228e-05,
+      "loss": 0.9325,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999686761435413e-05,
+      "loss": 0.9288,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999671296012182e-05,
+      "loss": 0.9109,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999655458010794e-05,
+      "loss": 0.9337,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996392474371514e-05,
+      "loss": 0.8562,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996226642972964e-05,
+      "loss": 0.9265,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996057085974094e-05,
+      "loss": 0.8488,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999588380343809e-05,
+      "loss": 0.8641,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995706795429544e-05,
+      "loss": 0.9637,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995526062014423e-05,
+      "loss": 0.9707,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995341603260087e-05,
+      "loss": 0.8914,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999515341923529e-05,
+      "loss": 0.896,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994961510010156e-05,
+      "loss": 0.8977,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994765875656218e-05,
+      "loss": 0.9057,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999456651624639e-05,
+      "loss": 0.899,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999436343185497e-05,
+      "loss": 0.8783,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994156622557644e-05,
+      "loss": 0.8765,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99939460884315e-05,
+      "loss": 0.918,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993731829554994e-05,
+      "loss": 0.896,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993513846007986e-05,
+      "loss": 0.8906,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999329213787172e-05,
+      "loss": 0.8879,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993066705228824e-05,
+      "loss": 0.9306,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992837548163315e-05,
+      "loss": 0.8579,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992604666760606e-05,
+      "loss": 0.8803,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992368061107485e-05,
+      "loss": 0.893,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999212773129214e-05,
+      "loss": 0.8859,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999188367740414e-05,
+      "loss": 0.8594,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991635899534444e-05,
+      "loss": 0.9002,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991384397775403e-05,
+      "loss": 0.953,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991129172220743e-05,
+      "loss": 0.9088,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990870222965597e-05,
+      "loss": 0.891,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990607550106466e-05,
+      "loss": 0.8579,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990341153741257e-05,
+      "loss": 0.9384,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999007103396925e-05,
+      "loss": 0.8816,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989797190891124e-05,
+      "loss": 0.8746,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989519624608934e-05,
+      "loss": 0.9418,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989238335226132e-05,
+      "loss": 0.8395,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998895332284756e-05,
+      "loss": 0.8999,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988664587579432e-05,
+      "loss": 0.8834,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988372129529373e-05,
+      "loss": 0.9156,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988075948806366e-05,
+      "loss": 0.9147,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987776045520814e-05,
+      "loss": 0.9036,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998747241978448e-05,
+      "loss": 0.8863,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998716507171053e-05,
+      "loss": 0.9032,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986854001413513e-05,
+      "loss": 0.9072,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998653920900936e-05,
+      "loss": 0.9164,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986220694615403e-05,
+      "loss": 0.876,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985898458350347e-05,
+      "loss": 0.921,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998557250033429e-05,
+      "loss": 0.9808,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985242820688716e-05,
+      "loss": 0.8919,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984909419536503e-05,
+      "loss": 0.9218,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99845722970019e-05,
+      "loss": 0.9422,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998423145321056e-05,
+      "loss": 0.8872,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983886888289515e-05,
+      "loss": 0.952,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983538602367182e-05,
+      "loss": 0.8851,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998318659557337e-05,
+      "loss": 0.8823,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982830868039266e-05,
+      "loss": 0.8834,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998247141989746e-05,
+      "loss": 0.8185,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998210825128191e-05,
+      "loss": 0.2558,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981741362327973e-05,
+      "loss": 0.8899,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998137075317239e-05,
+      "loss": 0.918,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998099642395328e-05,
+      "loss": 0.9113,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980618374810167e-05,
+      "loss": 0.8773,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980236605883942e-05,
+      "loss": 0.9242,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979851117316893e-05,
+      "loss": 0.8491,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997946190925269e-05,
+      "loss": 0.2319,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99790689818364e-05,
+      "loss": 0.8384,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978672335214453e-05,
+      "loss": 0.9365,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997827196953469e-05,
+      "loss": 0.8937,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977867884946322e-05,
+      "loss": 0.8542,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977460081599956e-05,
+      "loss": 0.8984,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977048559647577e-05,
+      "loss": 0.9161,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997663331924256e-05,
+      "loss": 0.9148,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976214360539664e-05,
+      "loss": 0.901,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975791683695042e-05,
+      "loss": 0.2557,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997536528886622e-05,
+      "loss": 0.8845,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997493517621212e-05,
+      "loss": 0.8857,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974501345893035e-05,
+      "loss": 0.9066,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974063798070666e-05,
+      "loss": 0.8743,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997362253290808e-05,
+      "loss": 0.8897,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973177550569743e-05,
+      "loss": 0.9237,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972728851221494e-05,
+      "loss": 0.881,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972276435030565e-05,
+      "loss": 0.9179,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971820302165576e-05,
+      "loss": 0.89,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971360452796523e-05,
+      "loss": 0.8944,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997089688709479e-05,
+      "loss": 0.9129,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997042960523316e-05,
+      "loss": 0.874,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969958607385777e-05,
+      "loss": 0.8779,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996948389372819e-05,
+      "loss": 0.8867,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969005464437318e-05,
+      "loss": 0.8433,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968523319691474e-05,
+      "loss": 0.8531,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996803745967036e-05,
+      "loss": 0.8427,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967547884555053e-05,
+      "loss": 0.9238,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967054594528012e-05,
+      "loss": 0.8796,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966557589773095e-05,
+      "loss": 0.9103,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996605687047553e-05,
+      "loss": 0.8955,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996555243682194e-05,
+      "loss": 0.8976,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965044289000323e-05,
+      "loss": 0.8627,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996453242720007e-05,
+      "loss": 0.8448,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996401685161195e-05,
+      "loss": 0.9106,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996349756242812e-05,
+      "loss": 0.899,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962974559842117e-05,
+      "loss": 0.9267,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962447844048866e-05,
+      "loss": 0.9494,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961917415244674e-05,
+      "loss": 0.8689,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961383273627236e-05,
+      "loss": 0.8991,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996084541939562e-05,
+      "loss": 0.9208,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996030385275029e-05,
+      "loss": 0.8753,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959758573893082e-05,
+      "loss": 0.8619,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995920958302723e-05,
+      "loss": 0.9445,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958656880357336e-05,
+      "loss": 0.933,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99581004660894e-05,
+      "loss": 0.9246,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995754034043079e-05,
+      "loss": 0.8938,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956976503590276e-05,
+      "loss": 0.9116,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956408955777988e-05,
+      "loss": 0.8788,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955837697205454e-05,
+      "loss": 0.8804,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955262728085592e-05,
+      "loss": 0.8665,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995468404863268e-05,
+      "loss": 0.9103,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954101659062406e-05,
+      "loss": 0.2252,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953515559591818e-05,
+      "loss": 0.919,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952925750439355e-05,
+      "loss": 0.8729,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952332231824843e-05,
+      "loss": 0.8634,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951735003969484e-05,
+      "loss": 0.8926,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995113406709587e-05,
+      "loss": 0.8449,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950529421427966e-05,
+      "loss": 0.2589,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949921067191127e-05,
+      "loss": 0.9232,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949309004612084e-05,
+      "loss": 0.8963,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994869323391895e-05,
+      "loss": 0.9035,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948073755341237e-05,
+      "loss": 0.8836,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994745056910981e-05,
+      "loss": 0.8855,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946823675456937e-05,
+      "loss": 0.9329,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994619307461626e-05,
+      "loss": 0.8936,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994555876682281e-05,
+      "loss": 0.8974,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944920752312984e-05,
+      "loss": 0.857,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944279031324577e-05,
+      "loss": 0.8804,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994363360409676e-05,
+      "loss": 0.9135,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942984470870083e-05,
+      "loss": 0.2417,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942331631886476e-05,
+      "loss": 0.8341,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941675087389254e-05,
+      "loss": 0.9131,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994101483762311e-05,
+      "loss": 0.8606,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994035088283412e-05,
+      "loss": 0.2489,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939683223269745e-05,
+      "loss": 0.9088,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939011859178817e-05,
+      "loss": 0.9024,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938336790811556e-05,
+      "loss": 0.8926,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993765801841956e-05,
+      "loss": 0.8427,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936975542255806e-05,
+      "loss": 0.9072,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9936289362574656e-05,
+      "loss": 0.9152,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993559947963185e-05,
+      "loss": 0.2468,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934905893684505e-05,
+      "loss": 0.9614,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934208604991126e-05,
+      "loss": 0.8397,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933507613811584e-05,
+      "loss": 0.8645,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932802920407148e-05,
+      "loss": 0.8887,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932094525040452e-05,
+      "loss": 0.875,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931382427975517e-05,
+      "loss": 0.8734,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930666629477745e-05,
+      "loss": 0.8147,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992994712981391e-05,
+      "loss": 0.822,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929223929252174e-05,
+      "loss": 0.8094,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928497028062072e-05,
+      "loss": 0.8682,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992776642651452e-05,
+      "loss": 0.9388,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927032124881816e-05,
+      "loss": 0.9078,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992629412343763e-05,
+      "loss": 0.8911,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992555242245702e-05,
+      "loss": 0.822,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992480702221642e-05,
+      "loss": 0.9128,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992405792299364e-05,
+      "loss": 0.8859,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923305125067864e-05,
+      "loss": 0.8177,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922548628719664e-05,
+      "loss": 0.8728,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992178843423099e-05,
+      "loss": 0.8641,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921024541885164e-05,
+      "loss": 0.8992,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992025695196689e-05,
+      "loss": 0.2155,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919485664762252e-05,
+      "loss": 0.2492,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918710680558698e-05,
+      "loss": 0.8852,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991793199964508e-05,
+      "loss": 0.8401,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917149622311603e-05,
+      "loss": 0.9252,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916363548849862e-05,
+      "loss": 0.8704,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991557377955283e-05,
+      "loss": 0.8542,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991478031471485e-05,
+      "loss": 0.905,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913983154631654e-05,
+      "loss": 0.924,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913182299600333e-05,
+      "loss": 0.8262,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912377749919373e-05,
+      "loss": 0.8787,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911569505888635e-05,
+      "loss": 0.8362,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991075756780934e-05,
+      "loss": 0.9101,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909941935984108e-05,
+      "loss": 0.8845,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909122610716927e-05,
+      "loss": 0.8747,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908299592313146e-05,
+      "loss": 0.9073,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990747288107952e-05,
+      "loss": 0.8098,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990664247732416e-05,
+      "loss": 0.8744,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990580838135655e-05,
+      "loss": 0.8604,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904970593487568e-05,
+      "loss": 0.9436,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904129114029457e-05,
+      "loss": 0.8717,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903283943295833e-05,
+      "loss": 0.9098,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990243508160169e-05,
+      "loss": 0.8773,
+      "step": 389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901582529263404e-05,
+      "loss": 0.8977,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900726286598722e-05,
+      "loss": 0.8402,
+      "step": 391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989986635392677e-05,
+      "loss": 0.9526,
+      "step": 392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899002731568034e-05,
+      "loss": 0.845,
+      "step": 393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898135419844394e-05,
+      "loss": 0.2628,
+      "step": 394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897264419079097e-05,
+      "loss": 0.8123,
+      "step": 395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896389729596764e-05,
+      "loss": 0.921,
+      "step": 396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895511351723394e-05,
+      "loss": 0.9187,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989462928578636e-05,
+      "loss": 0.8325,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893743532114405e-05,
+      "loss": 0.8929,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892854091037652e-05,
+      "loss": 0.8765,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98919609628876e-05,
+      "loss": 0.9113,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891064147997108e-05,
+      "loss": 0.862,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989016364670043e-05,
+      "loss": 0.8343,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988925945933318e-05,
+      "loss": 0.9141,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988835158623235e-05,
+      "loss": 0.8537,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887440027736305e-05,
+      "loss": 0.8426,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988652478418478e-05,
+      "loss": 0.9011,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885605855918887e-05,
+      "loss": 0.7925,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884683243281117e-05,
+      "loss": 0.8004,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988375694661532e-05,
+      "loss": 0.8938,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882826966266736e-05,
+      "loss": 0.8475,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881893302581966e-05,
+      "loss": 0.8769,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880955955908985e-05,
+      "loss": 0.2819,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880014926597145e-05,
+      "loss": 0.8385,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879070214997168e-05,
+      "loss": 0.9061,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878121821461144e-05,
+      "loss": 0.8681,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877169746342545e-05,
+      "loss": 0.8713,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876213989996215e-05,
+      "loss": 0.8718,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875254552778353e-05,
+      "loss": 0.8665,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987429143504655e-05,
+      "loss": 0.9037,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987332463715976e-05,
+      "loss": 0.8754,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872354159478307e-05,
+      "loss": 0.8623,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871380002363885e-05,
+      "loss": 0.918,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987040216617957e-05,
+      "loss": 0.9099,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869420651289804e-05,
+      "loss": 0.8995,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986843545806039e-05,
+      "loss": 0.8785,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867446586858515e-05,
+      "loss": 0.8696,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866454038052733e-05,
+      "loss": 0.9196,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986545781201296e-05,
+      "loss": 0.2458,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98644579091105e-05,
+      "loss": 0.2533,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986345432971801e-05,
+      "loss": 0.9088,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862447074209532e-05,
+      "loss": 0.8642,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861436142960464e-05,
+      "loss": 0.8388,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860421536347586e-05,
+      "loss": 0.8039,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859403254749038e-05,
+      "loss": 0.8761,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858381298544336e-05,
+      "loss": 0.9051,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857355668114365e-05,
+      "loss": 0.9565,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856326363841376e-05,
+      "loss": 0.8231,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855293386108995e-05,
+      "loss": 0.8999,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854256735302207e-05,
+      "loss": 0.8564,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985321641180738e-05,
+      "loss": 0.8575,
+      "step": 441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852172416012236e-05,
+      "loss": 0.9039,
+      "step": 442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985112474830588e-05,
+      "loss": 0.8906,
+      "step": 443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850073409078778e-05,
+      "loss": 0.8792,
+      "step": 444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849018398722757e-05,
+      "loss": 0.2859,
+      "step": 445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847959717631024e-05,
+      "loss": 0.9235,
+      "step": 446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846897366198154e-05,
+      "loss": 0.8303,
+      "step": 447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845831344820083e-05,
+      "loss": 0.2616,
+      "step": 448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844761653894118e-05,
+      "loss": 0.2195,
+      "step": 449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843688293818936e-05,
+      "loss": 0.9319,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984261126499457e-05,
+      "loss": 0.8496,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984153056782244e-05,
+      "loss": 0.2269,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840446202705318e-05,
+      "loss": 0.885,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983935817004735e-05,
+      "loss": 0.8966,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838266470254044e-05,
+      "loss": 0.8974,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837171103732277e-05,
+      "loss": 0.913,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836072070890293e-05,
+      "loss": 0.8707,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834969372137706e-05,
+      "loss": 0.9081,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983386300788549e-05,
+      "loss": 0.9292,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832752978545983e-05,
+      "loss": 0.2537,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831639284532897e-05,
+      "loss": 0.9224,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983052192626131e-05,
+      "loss": 0.9201,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982940090414766e-05,
+      "loss": 0.8983,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828276218609753e-05,
+      "loss": 0.9431,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982714787006676e-05,
+      "loss": 0.87,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826015858939212e-05,
+      "loss": 0.852,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982488018564902e-05,
+      "loss": 0.8276,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823740850619445e-05,
+      "loss": 0.8518,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822597854275114e-05,
+      "loss": 0.2394,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821451197042028e-05,
+      "loss": 0.8537,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820300879347546e-05,
+      "loss": 0.8849,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819146901620393e-05,
+      "loss": 0.891,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817989264290653e-05,
+      "loss": 0.868,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816827967789783e-05,
+      "loss": 0.8684,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815663012550595e-05,
+      "loss": 0.9068,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814494399007272e-05,
+      "loss": 0.8841,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813322127595355e-05,
+      "loss": 0.8415,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812146198751753e-05,
+      "loss": 0.2645,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810966612914732e-05,
+      "loss": 0.8413,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980978337052393e-05,
+      "loss": 0.9065,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808596472020332e-05,
+      "loss": 0.9244,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807405917846304e-05,
+      "loss": 0.8685,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806211708445564e-05,
+      "loss": 0.9049,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805013844263193e-05,
+      "loss": 0.8868,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980381232574564e-05,
+      "loss": 0.8213,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980260715334071e-05,
+      "loss": 0.8675,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801398327497572e-05,
+      "loss": 0.9054,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800185848666755e-05,
+      "loss": 0.9236,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979896971730015e-05,
+      "loss": 0.9122,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797749933851008e-05,
+      "loss": 0.868,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979652649877395e-05,
+      "loss": 0.8649,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795299412524948e-05,
+      "loss": 0.8699,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794068675561332e-05,
+      "loss": 0.9292,
+      "step": 493
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792834288341807e-05,
+      "loss": 0.861,
+      "step": 494
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791596251326427e-05,
+      "loss": 0.8827,
+      "step": 495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790354564976607e-05,
+      "loss": 0.8874,
+      "step": 496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789109229755127e-05,
+      "loss": 0.8208,
+      "step": 497
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978786024612612e-05,
+      "loss": 0.255,
+      "step": 498
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786607614555086e-05,
+      "loss": 0.8547,
+      "step": 499
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785351335508888e-05,
+      "loss": 0.9165,
+      "step": 500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784091409455728e-05,
+      "loss": 0.8459,
+      "step": 501
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782827836865194e-05,
+      "loss": 0.8876,
+      "step": 502
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781560618208214e-05,
+      "loss": 0.879,
+      "step": 503
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780289753957085e-05,
+      "loss": 0.8578,
+      "step": 504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779015244585452e-05,
+      "loss": 0.8592,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777737090568334e-05,
+      "loss": 0.8983,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776455292382092e-05,
+      "loss": 0.8798,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775169850504463e-05,
+      "loss": 0.8615,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773880765414522e-05,
+      "loss": 0.8974,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772588037592716e-05,
+      "loss": 0.8844,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771291667520845e-05,
+      "loss": 0.8983,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976999165568207e-05,
+      "loss": 0.246,
+      "step": 512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768688002560902e-05,
+      "loss": 0.8766,
+      "step": 513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976738070864322e-05,
+      "loss": 0.8843,
+      "step": 514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766069774416242e-05,
+      "loss": 0.8592,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764755200368562e-05,
+      "loss": 0.914,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763436986990122e-05,
+      "loss": 0.9273,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762115134772218e-05,
+      "loss": 0.8503,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976078964420751e-05,
+      "loss": 0.9205,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759460515790003e-05,
+      "loss": 0.9224,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975812775001507e-05,
+      "loss": 0.8407,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756791347379432e-05,
+      "loss": 0.872,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755451308381166e-05,
+      "loss": 0.8617,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754107633519706e-05,
+      "loss": 0.8678,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752760323295843e-05,
+      "loss": 0.8458,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751409378211714e-05,
+      "loss": 0.9155,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750054798770816e-05,
+      "loss": 0.8518,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748696585478015e-05,
+      "loss": 0.8249,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747334738839504e-05,
+      "loss": 0.9106,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745969259362856e-05,
+      "loss": 0.8656,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744600147556972e-05,
+      "loss": 0.8855,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743227403932135e-05,
+      "loss": 0.8759,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974185102899996e-05,
+      "loss": 0.8768,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740471023273422e-05,
+      "loss": 0.8424,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973908738726685e-05,
+      "loss": 0.8113,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737700121495935e-05,
+      "loss": 0.9048,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97363092264777e-05,
+      "loss": 0.9098,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734914702730542e-05,
+      "loss": 0.8966,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733516550774192e-05,
+      "loss": 0.8306,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732114771129747e-05,
+      "loss": 0.8227,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973070936431965e-05,
+      "loss": 0.8294,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97293003308677e-05,
+      "loss": 0.8356,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972788767129904e-05,
+      "loss": 0.8473,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726471386140175e-05,
+      "loss": 0.87,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725051475918954e-05,
+      "loss": 0.8588,
+      "step": 545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972362794116457e-05,
+      "loss": 0.8752,
+      "step": 546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722200782407585e-05,
+      "loss": 0.8273,
+      "step": 547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97207700001799e-05,
+      "loss": 0.8888,
+      "step": 548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971933559501477e-05,
+      "loss": 0.288,
+      "step": 549
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971789756744679e-05,
+      "loss": 0.8644,
+      "step": 550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716455918011923e-05,
+      "loss": 0.8728,
+      "step": 551
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971501064724747e-05,
+      "loss": 0.9137,
+      "step": 552
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713561755692083e-05,
+      "loss": 0.8473,
+      "step": 553
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712109243885768e-05,
+      "loss": 0.9003,
+      "step": 554
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710653112369877e-05,
+      "loss": 0.2392,
+      "step": 555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709193361687106e-05,
+      "loss": 0.9118,
+      "step": 556
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707729992381508e-05,
+      "loss": 0.2535,
+      "step": 557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706263004998484e-05,
+      "loss": 0.8569,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704792400084777e-05,
+      "loss": 0.8725,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703318178188486e-05,
+      "loss": 0.9139,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701840339859052e-05,
+      "loss": 0.8808,
+      "step": 561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970035888564727e-05,
+      "loss": 0.8748,
+      "step": 562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698873816105272e-05,
+      "loss": 0.8199,
+      "step": 563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697385131786554e-05,
+      "loss": 0.9539,
+      "step": 564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969589283324594e-05,
+      "loss": 0.8722,
+      "step": 565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694396921039617e-05,
+      "loss": 0.918,
+      "step": 566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692897395725113e-05,
+      "loss": 0.8159,
+      "step": 567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691394257861298e-05,
+      "loss": 0.9155,
+      "step": 568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689887508008393e-05,
+      "loss": 0.8388,
+      "step": 569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968837714672797e-05,
+      "loss": 0.8991,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968686317458293e-05,
+      "loss": 0.8899,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685345592137544e-05,
+      "loss": 0.8671,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968382439995741e-05,
+      "loss": 0.8324,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968229959860948e-05,
+      "loss": 0.8776,
+      "step": 574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680771188662044e-05,
+      "loss": 0.8387,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679239170684745e-05,
+      "loss": 0.8478,
+      "step": 576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677703545248567e-05,
+      "loss": 0.8541,
+      "step": 577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676164312925837e-05,
+      "loss": 0.8622,
+      "step": 578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967462147429023e-05,
+      "loss": 0.8429,
+      "step": 579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967307502991676e-05,
+      "loss": 0.8494,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671524980381793e-05,
+      "loss": 0.9178,
+      "step": 581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669971326263028e-05,
+      "loss": 0.9029,
+      "step": 582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668414068139512e-05,
+      "loss": 0.8562,
+      "step": 583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666853206591642e-05,
+      "loss": 0.8912,
+      "step": 584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966528874220115e-05,
+      "loss": 0.8993,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966372067555111e-05,
+      "loss": 0.8995,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662149007225947e-05,
+      "loss": 0.9184,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966057373781142e-05,
+      "loss": 0.7985,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965899486789463e-05,
+      "loss": 0.8991,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657412398064023e-05,
+      "loss": 0.8202,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655826328909392e-05,
+      "loss": 0.8584,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654236661021863e-05,
+      "loss": 0.8042,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965264339499391e-05,
+      "loss": 0.8824,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651046531419335e-05,
+      "loss": 0.9056,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649446070893298e-05,
+      "loss": 0.8081,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964784201401229e-05,
+      "loss": 0.8426,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964623436137414e-05,
+      "loss": 0.2576,
+      "step": 597
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644623113578026e-05,
+      "loss": 0.8863,
+      "step": 598
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964300827122446e-05,
+      "loss": 0.8239,
+      "step": 599
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641389834915295e-05,
+      "loss": 0.9259,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639767805253726e-05,
+      "loss": 0.2581,
+      "step": 601
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963814218284428e-05,
+      "loss": 0.8846,
+      "step": 602
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9636512968292826e-05,
+      "loss": 0.86,
+      "step": 603
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634880162206578e-05,
+      "loss": 0.8625,
+      "step": 604
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963324376519408e-05,
+      "loss": 0.872,
+      "step": 605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963160377786522e-05,
+      "loss": 0.8374,
+      "step": 606
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629960200831228e-05,
+      "loss": 0.8583,
+      "step": 607
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962831303470466e-05,
+      "loss": 0.8573,
+      "step": 608
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626662280099412e-05,
+      "loss": 0.8581,
+      "step": 609
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9625007937630733e-05,
+      "loss": 0.8616,
+      "step": 610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9623350007915187e-05,
+      "loss": 0.8752,
+      "step": 611
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621688491570688e-05,
+      "loss": 0.9023,
+      "step": 612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620023389216493e-05,
+      "loss": 0.8844,
+      "step": 613
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618354701473174e-05,
+      "loss": 0.7966,
+      "step": 614
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616682428962654e-05,
+      "loss": 0.8424,
+      "step": 615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96150065723082e-05,
+      "loss": 0.7997,
+      "step": 616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613327132134397e-05,
+      "loss": 0.8109,
+      "step": 617
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961164410906717e-05,
+      "loss": 0.8461,
+      "step": 618
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609957503733793e-05,
+      "loss": 0.8474,
+      "step": 619
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608267316762858e-05,
+      "loss": 0.9178,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96065735487843e-05,
+      "loss": 0.8316,
+      "step": 621
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960487620042939e-05,
+      "loss": 0.8979,
+      "step": 622
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960317527233073e-05,
+      "loss": 0.8372,
+      "step": 623
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960147076512225e-05,
+      "loss": 0.8548,
+      "step": 624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959976267943923e-05,
+      "loss": 0.8453,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959805101591827e-05,
+      "loss": 0.8204,
+      "step": 626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959633577519731e-05,
+      "loss": 0.816,
+      "step": 627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594616957915626e-05,
+      "loss": 0.8239,
+      "step": 628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959289456471381e-05,
+      "loss": 0.8814,
+      "step": 629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959116859623381e-05,
+      "loss": 0.8853,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958943905311889e-05,
+      "loss": 0.8488,
+      "step": 631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587705936013656e-05,
+      "loss": 0.8681,
+      "step": 632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958596924556404e-05,
+      "loss": 0.8282,
+      "step": 633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584228982417308e-05,
+      "loss": 0.8677,
+      "step": 634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958248514722205e-05,
+      "loss": 0.8972,
+      "step": 635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958073774062821e-05,
+      "loss": 0.8805,
+      "step": 636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578986763287032e-05,
+      "loss": 0.8468,
+      "step": 637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577232215851117e-05,
+      "loss": 0.8716,
+      "step": 638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957547409897438e-05,
+      "loss": 0.8607,
+      "step": 639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957371241331208e-05,
+      "loss": 0.8554,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571947159520787e-05,
+      "loss": 0.8432,
+      "step": 641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570178338258417e-05,
+      "loss": 0.9105,
+      "step": 642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956840595018422e-05,
+      "loss": 0.9024,
+      "step": 643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566629995958754e-05,
+      "loss": 0.8233,
+      "step": 644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956485047624392e-05,
+      "loss": 0.8967,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563067391702952e-05,
+      "loss": 0.8874,
+      "step": 646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561280743000402e-05,
+      "loss": 0.799,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559490530802155e-05,
+      "loss": 0.8858,
+      "step": 648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955769675577543e-05,
+      "loss": 0.8828,
+      "step": 649
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555899418588762e-05,
+      "loss": 0.8851,
+      "step": 650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554098519912022e-05,
+      "loss": 0.8986,
+      "step": 651
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552294060416407e-05,
+      "loss": 0.8473,
+      "step": 652
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955048604077444e-05,
+      "loss": 0.857,
+      "step": 653
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954867446165997e-05,
+      "loss": 0.2764,
+      "step": 654
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954685932374817e-05,
+      "loss": 0.9554,
+      "step": 655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545040627715554e-05,
+      "loss": 0.8926,
+      "step": 656
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954321837423994e-05,
+      "loss": 0.8524,
+      "step": 657
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954139256400049e-05,
+      "loss": 0.2311,
+      "step": 658
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539563197677683e-05,
+      "loss": 0.8433,
+      "step": 659
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537730275953322e-05,
+      "loss": 0.8987,
+      "step": 660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535893799510538e-05,
+      "loss": 0.8433,
+      "step": 661
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534053769033794e-05,
+      "loss": 0.8772,
+      "step": 662
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9532210185208863e-05,
+      "loss": 0.9007,
+      "step": 663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9530363048722857e-05,
+      "loss": 0.8569,
+      "step": 664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9528512360264197e-05,
+      "loss": 0.8856,
+      "step": 665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526658120522643e-05,
+      "loss": 0.8041,
+      "step": 666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952480033018927e-05,
+      "loss": 0.8823,
+      "step": 667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522938989956472e-05,
+      "loss": 0.8724,
+      "step": 668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521074100517973e-05,
+      "loss": 0.8636,
+      "step": 669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519205662568827e-05,
+      "loss": 0.83,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517333676805394e-05,
+      "loss": 0.8487,
+      "step": 671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515458143925367e-05,
+      "loss": 0.9196,
+      "step": 672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513579064627755e-05,
+      "loss": 0.8024,
+      "step": 673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95116964396129e-05,
+      "loss": 0.8738,
+      "step": 674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509810269582447e-05,
+      "loss": 0.8474,
+      "step": 675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507920555239382e-05,
+      "loss": 0.8068,
+      "step": 676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506027297288e-05,
+      "loss": 0.2671,
+      "step": 677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504130496433915e-05,
+      "loss": 0.8758,
+      "step": 678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502230153384073e-05,
+      "loss": 0.8676,
+      "step": 679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500326268846727e-05,
+      "loss": 0.8558,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949841884353146e-05,
+      "loss": 0.8781,
+      "step": 681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949650787814917e-05,
+      "loss": 0.9212,
+      "step": 682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494593373412075e-05,
+      "loss": 0.2377,
+      "step": 683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9492675330033712e-05,
+      "loss": 0.8592,
+      "step": 684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9490753748728937e-05,
+      "loss": 0.8337,
+      "step": 685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9488828630213924e-05,
+      "loss": 0.9084,
+      "step": 686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486899975206165e-05,
+      "loss": 0.8755,
+      "step": 687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9484967784424477e-05,
+      "loss": 0.8437,
+      "step": 688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483032058588978e-05,
+      "loss": 0.8071,
+      "step": 689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9481092798421124e-05,
+      "loss": 0.8688,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479150004643673e-05,
+      "loss": 0.8313,
+      "step": 691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947720367798071e-05,
+      "loss": 0.8956,
+      "step": 692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947525381915763e-05,
+      "loss": 0.8307,
+      "step": 693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473300428901142e-05,
+      "loss": 0.8804,
+      "step": 694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9471343507939284e-05,
+      "loss": 0.9065,
+      "step": 695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469383057001398e-05,
+      "loss": 0.8897,
+      "step": 696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9467419076818144e-05,
+      "loss": 0.8843,
+      "step": 697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465451568121506e-05,
+      "loss": 0.8507,
+      "step": 698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9463480531644766e-05,
+      "loss": 0.8643,
+      "step": 699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946150596812253e-05,
+      "loss": 0.8325,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945952787829073e-05,
+      "loss": 0.8553,
+      "step": 701
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457546262886593e-05,
+      "loss": 0.9303,
+      "step": 702
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945556112264867e-05,
+      "loss": 0.8545,
+      "step": 703
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9453572458316825e-05,
+      "loss": 0.8893,
+      "step": 704
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9451580270632235e-05,
+      "loss": 0.8755,
+      "step": 705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449584560337388e-05,
+      "loss": 0.8393,
+      "step": 706
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447585328176087e-05,
+      "loss": 0.8653,
+      "step": 707
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445582574893448e-05,
+      "loss": 0.9201,
+      "step": 708
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94435763012359e-05,
+      "loss": 0.8828,
+      "step": 709
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9441566507951175e-05,
+      "loss": 0.8305,
+      "step": 710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9439553195788336e-05,
+      "loss": 0.8255,
+      "step": 711
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9437536365497736e-05,
+      "loss": 0.8521,
+      "step": 712
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435516017831054e-05,
+      "loss": 0.8265,
+      "step": 713
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943349215354127e-05,
+      "loss": 0.817,
+      "step": 714
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431464773382687e-05,
+      "loss": 0.8748,
+      "step": 715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9429433878110906e-05,
+      "loss": 0.8781,
+      "step": 716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9427399468482848e-05,
+      "loss": 0.8855,
+      "step": 717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942536154525673e-05,
+      "loss": 0.8805,
+      "step": 718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9423320109192094e-05,
+      "loss": 0.8571,
+      "step": 719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942127516104978e-05,
+      "loss": 0.8722,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9419226701591945e-05,
+      "loss": 0.8562,
+      "step": 721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9417174731582055e-05,
+      "loss": 0.82,
+      "step": 722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415119251784872e-05,
+      "loss": 0.9273,
+      "step": 723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941306026296648e-05,
+      "loss": 0.8439,
+      "step": 724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941099776589426e-05,
+      "loss": 0.862,
+      "step": 725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9408931761336918e-05,
+      "loss": 0.9001,
+      "step": 726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9406862250064443e-05,
+      "loss": 0.8647,
+      "step": 727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940478923284815e-05,
+      "loss": 0.8588,
+      "step": 728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940271271046065e-05,
+      "loss": 0.8793,
+      "step": 729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9400632683675867e-05,
+      "loss": 0.8231,
+      "step": 730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9398549153269028e-05,
+      "loss": 0.8768,
+      "step": 731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9396462120016663e-05,
+      "loss": 0.8344,
+      "step": 732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394371584696614e-05,
+      "loss": 0.8346,
+      "step": 733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9392277548088025e-05,
+      "loss": 0.8541,
+      "step": 734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9390180010971345e-05,
+      "loss": 0.894,
+      "step": 735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9388078974128325e-05,
+      "loss": 0.2909,
+      "step": 736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9385974438342023e-05,
+      "loss": 0.8749,
+      "step": 737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9383866404396805e-05,
+      "loss": 0.8774,
+      "step": 738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938175487307833e-05,
+      "loss": 0.881,
+      "step": 739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9379639845173574e-05,
+      "loss": 0.881,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377521321470806e-05,
+      "loss": 0.8917,
+      "step": 741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9375399302759602e-05,
+      "loss": 0.8565,
+      "step": 742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373273789830836e-05,
+      "loss": 0.9025,
+      "step": 743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9371144783476693e-05,
+      "loss": 0.8362,
+      "step": 744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9369012284490655e-05,
+      "loss": 0.8795,
+      "step": 745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366876293667505e-05,
+      "loss": 0.8714,
+      "step": 746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9364736811803325e-05,
+      "loss": 0.8627,
+      "step": 747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93625938396955e-05,
+      "loss": 0.8818,
+      "step": 748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936044737814273e-05,
+      "loss": 0.9009,
+      "step": 749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9358297427944988e-05,
+      "loss": 0.9208,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9356143989903567e-05,
+      "loss": 0.851,
+      "step": 751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9353987064821055e-05,
+      "loss": 0.8647,
+      "step": 752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9351826653501343e-05,
+      "loss": 0.8491,
+      "step": 753
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934966275674961e-05,
+      "loss": 0.8081,
+      "step": 754
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9347495375372348e-05,
+      "loss": 0.8976,
+      "step": 755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934532451017734e-05,
+      "loss": 0.881,
+      "step": 756
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934315016197367e-05,
+      "loss": 0.8268,
+      "step": 757
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9340972331571715e-05,
+      "loss": 0.9166,
+      "step": 758
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338791019783156e-05,
+      "loss": 0.8949,
+      "step": 759
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9336606227420967e-05,
+      "loss": 0.8385,
+      "step": 760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933441795529943e-05,
+      "loss": 0.8544,
+      "step": 761
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933222620423411e-05,
+      "loss": 0.8153,
+      "step": 762
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933003097504187e-05,
+      "loss": 0.8505,
+      "step": 763
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932783226854088e-05,
+      "loss": 0.9174,
+      "step": 764
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932563008555059e-05,
+      "loss": 0.8843,
+      "step": 765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932342442689177e-05,
+      "loss": 0.8812,
+      "step": 766
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9321215293386458e-05,
+      "loss": 0.8441,
+      "step": 767
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9319002685858e-05,
+      "loss": 0.8275,
+      "step": 768
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9316786605131047e-05,
+      "loss": 0.8507,
+      "step": 769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9314567052031522e-05,
+      "loss": 0.8756,
+      "step": 770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931234402738666e-05,
+      "loss": 0.8283,
+      "step": 771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9310117532024982e-05,
+      "loss": 0.8605,
+      "step": 772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307887566776308e-05,
+      "loss": 0.2566,
+      "step": 773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930565413247174e-05,
+      "loss": 0.8526,
+      "step": 774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303417229943684e-05,
+      "loss": 0.2479,
+      "step": 775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301176860025838e-05,
+      "loss": 0.8361,
+      "step": 776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9298933023553187e-05,
+      "loss": 0.8719,
+      "step": 777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929668572136201e-05,
+      "loss": 0.8609,
+      "step": 778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929443495428988e-05,
+      "loss": 0.8733,
+      "step": 779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292180723175656e-05,
+      "loss": 0.8388,
+      "step": 780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928992302885949e-05,
+      "loss": 0.8227,
+      "step": 781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928766187218283e-05,
+      "loss": 0.8684,
+      "step": 782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9285397253988415e-05,
+      "loss": 0.8843,
+      "step": 783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9283129175120265e-05,
+      "loss": 0.8837,
+      "step": 784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928085763642369e-05,
+      "loss": 0.8426,
+      "step": 785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92785826387453e-05,
+      "loss": 0.8869,
+      "step": 786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9276304182932985e-05,
+      "loss": 0.8041,
+      "step": 787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9274022269835927e-05,
+      "loss": 0.8785,
+      "step": 788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271736900304603e-05,
+      "loss": 0.8973,
+      "step": 789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9269448075190763e-05,
+      "loss": 0.7929,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926715579534746e-05,
+      "loss": 0.8511,
+      "step": 791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926486006162902e-05,
+      "loss": 0.8376,
+      "step": 792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9262560874891076e-05,
+      "loss": 0.8309,
+      "step": 793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926025823599053e-05,
+      "loss": 0.8882,
+      "step": 794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925795214578557e-05,
+      "loss": 0.7846,
+      "step": 795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9255642605135693e-05,
+      "loss": 0.9005,
+      "step": 796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9253329614901654e-05,
+      "loss": 0.8487,
+      "step": 797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925101317594551e-05,
+      "loss": 0.8123,
+      "step": 798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9248693289130602e-05,
+      "loss": 0.8748,
+      "step": 799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924636995532155e-05,
+      "loss": 0.9009,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9244043175384262e-05,
+      "loss": 0.828,
+      "step": 801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9241712950185933e-05,
+      "loss": 0.3067,
+      "step": 802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9239379280595036e-05,
+      "loss": 0.9048,
+      "step": 803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923704216748133e-05,
+      "loss": 0.8496,
+      "step": 804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9234701611715868e-05,
+      "loss": 0.8283,
+      "step": 805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9232357614170967e-05,
+      "loss": 0.8185,
+      "step": 806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923001017572024e-05,
+      "loss": 0.7944,
+      "step": 807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922765929723858e-05,
+      "loss": 0.8913,
+      "step": 808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9225304979602156e-05,
+      "loss": 0.889,
+      "step": 809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9222947223688432e-05,
+      "loss": 0.8176,
+      "step": 810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9220586030376135e-05,
+      "loss": 0.8915,
+      "step": 811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9218221400545295e-05,
+      "loss": 0.9048,
+      "step": 812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9215853335077203e-05,
+      "loss": 0.819,
+      "step": 813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921348183485444e-05,
+      "loss": 0.8549,
+      "step": 814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9211106900760863e-05,
+      "loss": 0.8183,
+      "step": 815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9208728533681617e-05,
+      "loss": 0.8504,
+      "step": 816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920634673450312e-05,
+      "loss": 0.8337,
+      "step": 817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920396150411307e-05,
+      "loss": 0.8673,
+      "step": 818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920157284340044e-05,
+      "loss": 0.8491,
+      "step": 819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919918075325549e-05,
+      "loss": 0.8269,
+      "step": 820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919678523456975e-05,
+      "loss": 0.8678,
+      "step": 821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9194386288236034e-05,
+      "loss": 0.8023,
+      "step": 822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9191983915148432e-05,
+      "loss": 0.8434,
+      "step": 823
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918957811620231e-05,
+      "loss": 0.885,
+      "step": 824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187168892294306e-05,
+      "loss": 0.8348,
+      "step": 825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918475624432234e-05,
+      "loss": 0.8235,
+      "step": 826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9182340173185614e-05,
+      "loss": 0.8419,
+      "step": 827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9179920679784595e-05,
+      "loss": 0.8578,
+      "step": 828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9177497765021028e-05,
+      "loss": 0.9235,
+      "step": 829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9175071429797937e-05,
+      "loss": 0.8269,
+      "step": 830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917264167501962e-05,
+      "loss": 0.8527,
+      "step": 831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9170208501591648e-05,
+      "loss": 0.2669,
+      "step": 832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9167771910420865e-05,
+      "loss": 0.9143,
+      "step": 833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916533190241539e-05,
+      "loss": 0.8402,
+      "step": 834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9162888478484612e-05,
+      "loss": 0.2568,
+      "step": 835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9160441639539205e-05,
+      "loss": 0.9012,
+      "step": 836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9157991386491102e-05,
+      "loss": 0.8584,
+      "step": 837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915553772025352e-05,
+      "loss": 0.8461,
+      "step": 838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9153080641740933e-05,
+      "loss": 0.8853,
+      "step": 839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9150620151869098e-05,
+      "loss": 0.8758,
+      "step": 840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9148156251555047e-05,
+      "loss": 0.86,
+      "step": 841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9145688941717074e-05,
+      "loss": 0.8725,
+      "step": 842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9143218223274743e-05,
+      "loss": 0.9147,
+      "step": 843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91407440971489e-05,
+      "loss": 0.9187,
+      "step": 844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913826656426165e-05,
+      "loss": 0.2832,
+      "step": 845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913578562553637e-05,
+      "loss": 0.904,
+      "step": 846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9133301281897706e-05,
+      "loss": 0.7753,
+      "step": 847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9130813534271577e-05,
+      "loss": 0.8725,
+      "step": 848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9128322383585167e-05,
+      "loss": 0.7928,
+      "step": 849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9125827830766933e-05,
+      "loss": 0.85,
+      "step": 850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912332987674659e-05,
+      "loss": 0.7951,
+      "step": 851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9120828522455134e-05,
+      "loss": 0.8862,
+      "step": 852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911832376882482e-05,
+      "loss": 0.8883,
+      "step": 853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9115815616789167e-05,
+      "loss": 0.8514,
+      "step": 854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911330406728297e-05,
+      "loss": 0.8253,
+      "step": 855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911078912124228e-05,
+      "loss": 0.2935,
+      "step": 856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9108270779604425e-05,
+      "loss": 0.8387,
+      "step": 857
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9105749043307988e-05,
+      "loss": 0.84,
+      "step": 858
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9103223913292825e-05,
+      "loss": 0.8803,
+      "step": 859
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910069539050005e-05,
+      "loss": 0.8058,
+      "step": 860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9098163475872046e-05,
+      "loss": 0.8881,
+      "step": 861
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909562817035246e-05,
+      "loss": 0.8385,
+      "step": 862
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9093089474886205e-05,
+      "loss": 0.8675,
+      "step": 863
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9090547390419452e-05,
+      "loss": 0.7723,
+      "step": 864
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9088001917899637e-05,
+      "loss": 0.8487,
+      "step": 865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9085453058275457e-05,
+      "loss": 0.8548,
+      "step": 866
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908290081249688e-05,
+      "loss": 0.8058,
+      "step": 867
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9080345181515126e-05,
+      "loss": 0.867,
+      "step": 868
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9077786166282682e-05,
+      "loss": 0.8395,
+      "step": 869
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907522376775329e-05,
+      "loss": 0.8324,
+      "step": 870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9072657986881963e-05,
+      "loss": 0.8491,
+      "step": 871
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9070088824624964e-05,
+      "loss": 0.8515,
+      "step": 872
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9067516281939826e-05,
+      "loss": 0.8735,
+      "step": 873
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906494035978534e-05,
+      "loss": 0.9049,
+      "step": 874
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9062361059121547e-05,
+      "loss": 0.8499,
+      "step": 875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9059778380909754e-05,
+      "loss": 0.8639,
+      "step": 876
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9057192326112532e-05,
+      "loss": 0.8356,
+      "step": 877
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9054602895693704e-05,
+      "loss": 0.8722,
+      "step": 878
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905201009061835e-05,
+      "loss": 0.9093,
+      "step": 879
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9049413911852814e-05,
+      "loss": 0.8611,
+      "step": 880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904681436036469e-05,
+      "loss": 0.8723,
+      "step": 881
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904421143712283e-05,
+      "loss": 0.8219,
+      "step": 882
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041605143097358e-05,
+      "loss": 0.8682,
+      "step": 883
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903899547925963e-05,
+      "loss": 0.9044,
+      "step": 884
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9036382446582265e-05,
+      "loss": 0.855,
+      "step": 885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9033766046039155e-05,
+      "loss": 0.8088,
+      "step": 886
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9031146278605427e-05,
+      "loss": 0.8815,
+      "step": 887
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9028523145257472e-05,
+      "loss": 0.8374,
+      "step": 888
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9025896646972936e-05,
+      "loss": 0.8858,
+      "step": 889
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9023266784730708e-05,
+      "loss": 0.8257,
+      "step": 890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902063355951095e-05,
+      "loss": 0.8609,
+      "step": 891
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9017996972295055e-05,
+      "loss": 0.7815,
+      "step": 892
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9015357024065693e-05,
+      "loss": 0.848,
+      "step": 893
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901271371580677e-05,
+      "loss": 0.8727,
+      "step": 894
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9010067048503444e-05,
+      "loss": 0.8813,
+      "step": 895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007417023142136e-05,
+      "loss": 0.8625,
+      "step": 896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9004763640710507e-05,
+      "loss": 0.8733,
+      "step": 897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900210690219748e-05,
+      "loss": 0.9152,
+      "step": 898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899944680859322e-05,
+      "loss": 0.2658,
+      "step": 899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996783360889145e-05,
+      "loss": 0.9146,
+      "step": 900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8994116560077923e-05,
+      "loss": 0.8744,
+      "step": 901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899144640715348e-05,
+      "loss": 0.8561,
+      "step": 902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8988772903110974e-05,
+      "loss": 0.875,
+      "step": 903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8986096048946826e-05,
+      "loss": 0.8849,
+      "step": 904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89834158456587e-05,
+      "loss": 0.7926,
+      "step": 905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8980732294245507e-05,
+      "loss": 0.8371,
+      "step": 906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897804539570742e-05,
+      "loss": 0.8847,
+      "step": 907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8975355151045837e-05,
+      "loss": 0.8701,
+      "step": 908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8972661561263413e-05,
+      "loss": 0.2615,
+      "step": 909
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8969964627364057e-05,
+      "loss": 0.9009,
+      "step": 910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8967264350352917e-05,
+      "loss": 0.254,
+      "step": 911
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8964560731236386e-05,
+      "loss": 0.8429,
+      "step": 912
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89618537710221e-05,
+      "loss": 0.2494,
+      "step": 913
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8959143470718957e-05,
+      "loss": 0.809,
+      "step": 914
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8956429831337077e-05,
+      "loss": 0.8319,
+      "step": 915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8953712853887843e-05,
+      "loss": 0.8181,
+      "step": 916
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8950992539383866e-05,
+      "loss": 0.8292,
+      "step": 917
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8948268888839016e-05,
+      "loss": 0.8441,
+      "step": 918
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8945541903268395e-05,
+      "loss": 0.8661,
+      "step": 919
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8942811583688355e-05,
+      "loss": 0.8597,
+      "step": 920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8940077931116487e-05,
+      "loss": 0.8392,
+      "step": 921
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8937340946571626e-05,
+      "loss": 0.8369,
+      "step": 922
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8934600631073844e-05,
+      "loss": 0.8265,
+      "step": 923
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8931856985644462e-05,
+      "loss": 0.8554,
+      "step": 924
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892911001130604e-05,
+      "loss": 0.9445,
+      "step": 925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8926359709082377e-05,
+      "loss": 0.8827,
+      "step": 926
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8923606079998507e-05,
+      "loss": 0.8482,
+      "step": 927
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8920849125080716e-05,
+      "loss": 0.8425,
+      "step": 928
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8918088845356518e-05,
+      "loss": 0.8928,
+      "step": 929
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8915325241854673e-05,
+      "loss": 0.8302,
+      "step": 930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8912558315605175e-05,
+      "loss": 0.7947,
+      "step": 931
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8909788067639264e-05,
+      "loss": 0.8784,
+      "step": 932
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8907014498989412e-05,
+      "loss": 0.2398,
+      "step": 933
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8904237610689334e-05,
+      "loss": 0.8168,
+      "step": 934
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890145740377397e-05,
+      "loss": 0.8677,
+      "step": 935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8898673879279507e-05,
+      "loss": 0.8798,
+      "step": 936
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8895887038243366e-05,
+      "loss": 0.8433,
+      "step": 937
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889309688170421e-05,
+      "loss": 0.8406,
+      "step": 938
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8890303410701928e-05,
+      "loss": 0.8594,
+      "step": 939
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8887506626277647e-05,
+      "loss": 0.8664,
+      "step": 940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888470652947373e-05,
+      "loss": 0.8674,
+      "step": 941
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888190312133378e-05,
+      "loss": 0.8091,
+      "step": 942
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8879096402902632e-05,
+      "loss": 0.8044,
+      "step": 943
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887628637522634e-05,
+      "loss": 0.8259,
+      "step": 944
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887347303935221e-05,
+      "loss": 0.8954,
+      "step": 945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8870656396328774e-05,
+      "loss": 0.8721,
+      "step": 946
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8867836447205797e-05,
+      "loss": 0.8528,
+      "step": 947
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8865013193034277e-05,
+      "loss": 0.8007,
+      "step": 948
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886218663486644e-05,
+      "loss": 0.8749,
+      "step": 949
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8859356773755745e-05,
+      "loss": 0.8746,
+      "step": 950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8856523610756885e-05,
+      "loss": 0.8967,
+      "step": 951
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8853687146925783e-05,
+      "loss": 0.2926,
+      "step": 952
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885084738331959e-05,
+      "loss": 0.8001,
+      "step": 953
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884800432099669e-05,
+      "loss": 0.8435,
+      "step": 954
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8845157961016686e-05,
+      "loss": 0.835,
+      "step": 955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8842308304440432e-05,
+      "loss": 0.2432,
+      "step": 956
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8839455352329983e-05,
+      "loss": 0.8349,
+      "step": 957
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8836599105748644e-05,
+      "loss": 0.8222,
+      "step": 958
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8833739565760938e-05,
+      "loss": 0.8561,
+      "step": 959
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8830876733432616e-05,
+      "loss": 0.284,
+      "step": 960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882801060983066e-05,
+      "loss": 0.8557,
+      "step": 961
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8825141196023278e-05,
+      "loss": 0.8756,
+      "step": 962
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8822268493079898e-05,
+      "loss": 0.8547,
+      "step": 963
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881939250207118e-05,
+      "loss": 0.8565,
+      "step": 964
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8816513224069008e-05,
+      "loss": 0.8626,
+      "step": 965
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881363066014649e-05,
+      "loss": 0.9184,
+      "step": 966
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8810744811377962e-05,
+      "loss": 0.8935,
+      "step": 967
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8807855678838977e-05,
+      "loss": 0.8622,
+      "step": 968
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8804963263606322e-05,
+      "loss": 0.8392,
+      "step": 969
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8802067566758e-05,
+      "loss": 0.8238,
+      "step": 970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8799168589373236e-05,
+      "loss": 0.83,
+      "step": 971
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8796266332532486e-05,
+      "loss": 0.8652,
+      "step": 972
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879336079731742e-05,
+      "loss": 0.8934,
+      "step": 973
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8790451984810932e-05,
+      "loss": 0.8738,
+      "step": 974
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878753989609714e-05,
+      "loss": 0.8527,
+      "step": 975
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878462453226138e-05,
+      "loss": 0.2508,
+      "step": 976
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8781705894390218e-05,
+      "loss": 0.8413,
+      "step": 977
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877878398357142e-05,
+      "loss": 0.894,
+      "step": 978
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8775858800893988e-05,
+      "loss": 0.809,
+      "step": 979
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8772930347448148e-05,
+      "loss": 0.8585,
+      "step": 980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.876999862432532e-05,
+      "loss": 0.8189,
+      "step": 981
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876706363261818e-05,
+      "loss": 0.8164,
+      "step": 982
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8764125373420584e-05,
+      "loss": 0.8416,
+      "step": 983
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876118384782763e-05,
+      "loss": 0.9213,
+      "step": 984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8758239056935626e-05,
+      "loss": 0.8566,
+      "step": 985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87552910018421e-05,
+      "loss": 0.849,
+      "step": 986
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8752339683645787e-05,
+      "loss": 0.8532,
+      "step": 987
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8749385103446653e-05,
+      "loss": 0.9199,
+      "step": 988
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874642726234587e-05,
+      "loss": 0.829,
+      "step": 989
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8743466161445823e-05,
+      "loss": 0.8235,
+      "step": 990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874050180185012e-05,
+      "loss": 0.8359,
+      "step": 991
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8737534184663582e-05,
+      "loss": 0.8611,
+      "step": 992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8734563310992237e-05,
+      "loss": 0.8613,
+      "step": 993
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8731589181943333e-05,
+      "loss": 0.8432,
+      "step": 994
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728611798625335e-05,
+      "loss": 0.9167,
+      "step": 995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8725631162147907e-05,
+      "loss": 0.2722,
+      "step": 996
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872264727362194e-05,
+      "loss": 0.8827,
+      "step": 997
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871966013415953e-05,
+      "loss": 0.8477,
+      "step": 998
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8716669744873986e-05,
+      "loss": 0.878,
+      "step": 999
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8713676106879825e-05,
+      "loss": 0.7984,
+      "step": 1000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8710679221292785e-05,
+      "loss": 0.8365,
+      "step": 1001
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87076790892298e-05,
+      "loss": 0.8697,
+      "step": 1002
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8704675711809025e-05,
+      "loss": 0.8872,
+      "step": 1003
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8701669090149818e-05,
+      "loss": 0.847,
+      "step": 1004
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8698659225372755e-05,
+      "loss": 0.8559,
+      "step": 1005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869564611859961e-05,
+      "loss": 0.8083,
+      "step": 1006
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8692629770953367e-05,
+      "loss": 0.9211,
+      "step": 1007
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8689610183558225e-05,
+      "loss": 0.856,
+      "step": 1008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8686587357539586e-05,
+      "loss": 0.812,
+      "step": 1009
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8683561294024057e-05,
+      "loss": 0.8143,
+      "step": 1010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8680531994139456e-05,
+      "loss": 0.8643,
+      "step": 1011
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8677499459014807e-05,
+      "loss": 0.8788,
+      "step": 1012
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8674463689780333e-05,
+      "loss": 0.8261,
+      "step": 1013
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8671424687567473e-05,
+      "loss": 0.8646,
+      "step": 1014
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8668382453508856e-05,
+      "loss": 0.8532,
+      "step": 1015
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8665336988738335e-05,
+      "loss": 0.857,
+      "step": 1016
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866228829439095e-05,
+      "loss": 0.7758,
+      "step": 1017
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8659236371602954e-05,
+      "loss": 0.8395,
+      "step": 1018
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8656181221511798e-05,
+      "loss": 0.905,
+      "step": 1019
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8653122845256146e-05,
+      "loss": 0.9128,
+      "step": 1020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865006124397585e-05,
+      "loss": 0.8523,
+      "step": 1021
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864699641881197e-05,
+      "loss": 0.8067,
+      "step": 1022
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8643928370906772e-05,
+      "loss": 0.8579,
+      "step": 1023
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8640857101403717e-05,
+      "loss": 0.8327,
+      "step": 1024
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863778261144747e-05,
+      "loss": 0.8346,
+      "step": 1025
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86347049021839e-05,
+      "loss": 0.8387,
+      "step": 1026
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8631623974760063e-05,
+      "loss": 0.2557,
+      "step": 1027
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862853983032423e-05,
+      "loss": 0.8503,
+      "step": 1028
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862545247002586e-05,
+      "loss": 0.7932,
+      "step": 1029
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8622361895015612e-05,
+      "loss": 0.9067,
+      "step": 1030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8619268106445352e-05,
+      "loss": 0.9265,
+      "step": 1031
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861617110546813e-05,
+      "loss": 0.8748,
+      "step": 1032
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8613070893238206e-05,
+      "loss": 0.7815,
+      "step": 1033
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860996747091103e-05,
+      "loss": 0.8431,
+      "step": 1034
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8606860839643242e-05,
+      "loss": 0.8579,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.86037510005927e-05,
+      "loss": 0.8579,
+      "step": 1036
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8600637954918426e-05,
+      "loss": 0.8545,
+      "step": 1037
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859752170378067e-05,
+      "loss": 0.8365,
+      "step": 1038
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859440224834085e-05,
+      "loss": 0.8019,
+      "step": 1039
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8591279589761596e-05,
+      "loss": 0.8362,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858815372920672e-05,
+      "loss": 0.9013,
+      "step": 1041
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8585024667841235e-05,
+      "loss": 0.8908,
+      "step": 1042
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8581892406831345e-05,
+      "loss": 0.8221,
+      "step": 1043
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8578756947344442e-05,
+      "loss": 0.8474,
+      "step": 1044
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857561829054912e-05,
+      "loss": 0.7729,
+      "step": 1045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8572476437615154e-05,
+      "loss": 0.8914,
+      "step": 1046
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8569331389713513e-05,
+      "loss": 0.835,
+      "step": 1047
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856618314801636e-05,
+      "loss": 0.8998,
+      "step": 1048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856303171369705e-05,
+      "loss": 0.8815,
+      "step": 1049
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855987708793012e-05,
+      "loss": 0.8762,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8556719271891305e-05,
+      "loss": 0.8421,
+      "step": 1051
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8553558266757522e-05,
+      "loss": 0.2431,
+      "step": 1052
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855039407370688e-05,
+      "loss": 0.8694,
+      "step": 1053
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854722669391868e-05,
+      "loss": 0.8403,
+      "step": 1054
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8544056128573405e-05,
+      "loss": 0.8365,
+      "step": 1055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8540882378852726e-05,
+      "loss": 0.8447,
+      "step": 1056
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8537705445939496e-05,
+      "loss": 0.8417,
+      "step": 1057
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8534525331017767e-05,
+      "loss": 0.8933,
+      "step": 1058
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8531342035272768e-05,
+      "loss": 0.9037,
+      "step": 1059
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8528155559890916e-05,
+      "loss": 0.874,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8524965906059814e-05,
+      "loss": 0.7919,
+      "step": 1061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8521773074968242e-05,
+      "loss": 0.8457,
+      "step": 1062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8518577067806176e-05,
+      "loss": 0.8796,
+      "step": 1063
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851537788576477e-05,
+      "loss": 0.8525,
+      "step": 1064
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851217553003636e-05,
+      "loss": 0.8557,
+      "step": 1065
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8508970001814463e-05,
+      "loss": 0.2371,
+      "step": 1066
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8505761302293785e-05,
+      "loss": 0.855,
+      "step": 1067
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850254943267021e-05,
+      "loss": 0.7953,
+      "step": 1068
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8499334394140806e-05,
+      "loss": 0.2538,
+      "step": 1069
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849611618790381e-05,
+      "loss": 0.8231,
+      "step": 1070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8492894815158668e-05,
+      "loss": 0.7742,
+      "step": 1071
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848967027710597e-05,
+      "loss": 0.8773,
+      "step": 1072
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848644257494751e-05,
+      "loss": 0.248,
+      "step": 1073
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848321170988626e-05,
+      "loss": 0.8244,
+      "step": 1074
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847997768312636e-05,
+      "loss": 0.8891,
+      "step": 1075
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847674049587313e-05,
+      "loss": 0.8654,
+      "step": 1076
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8473500149333076e-05,
+      "loss": 0.9004,
+      "step": 1077
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8470256644713884e-05,
+      "loss": 0.8338,
+      "step": 1078
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84670099832244e-05,
+      "loss": 0.8456,
+      "step": 1079
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8463760166074665e-05,
+      "loss": 0.8116,
+      "step": 1080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8460507194475878e-05,
+      "loss": 0.8633,
+      "step": 1081
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8457251069640433e-05,
+      "loss": 0.828,
+      "step": 1082
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8453991792781888e-05,
+      "loss": 0.8345,
+      "step": 1083
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8450729365114977e-05,
+      "loss": 0.8436,
+      "step": 1084
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8447463787855605e-05,
+      "loss": 0.8519,
+      "step": 1085
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844419506222086e-05,
+      "loss": 0.865,
+      "step": 1086
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8440923189428996e-05,
+      "loss": 0.8355,
+      "step": 1087
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8437648170699442e-05,
+      "loss": 0.8865,
+      "step": 1088
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8434370007252803e-05,
+      "loss": 0.8245,
+      "step": 1089
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8431088700310846e-05,
+      "loss": 0.8529,
+      "step": 1090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842780425109652e-05,
+      "loss": 0.8288,
+      "step": 1091
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8424516660833943e-05,
+      "loss": 0.8219,
+      "step": 1092
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8421225930748402e-05,
+      "loss": 0.2944,
+      "step": 1093
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841793206206635e-05,
+      "loss": 0.8995,
+      "step": 1094
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841463505601542e-05,
+      "loss": 0.811,
+      "step": 1095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8411334913824403e-05,
+      "loss": 0.8553,
+      "step": 1096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8408031636723268e-05,
+      "loss": 0.8155,
+      "step": 1097
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8404725225943147e-05,
+      "loss": 0.876,
+      "step": 1098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8401415682716342e-05,
+      "loss": 0.8574,
+      "step": 1099
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839810300827632e-05,
+      "loss": 0.8086,
+      "step": 1100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8394787203857716e-05,
+      "loss": 0.845,
+      "step": 1101
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8391468270696342e-05,
+      "loss": 0.8552,
+      "step": 1102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8388146210029155e-05,
+      "loss": 0.2677,
+      "step": 1103
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8384821023094294e-05,
+      "loss": 0.8905,
+      "step": 1104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838149271113106e-05,
+      "loss": 0.8347,
+      "step": 1105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8378161275379914e-05,
+      "loss": 0.8545,
+      "step": 1106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8374826717082484e-05,
+      "loss": 0.278,
+      "step": 1107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8371489037481565e-05,
+      "loss": 0.8698,
+      "step": 1108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8368148237821114e-05,
+      "loss": 0.8718,
+      "step": 1109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8364804319346243e-05,
+      "loss": 0.8407,
+      "step": 1110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8361457283303237e-05,
+      "loss": 0.8497,
+      "step": 1111
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835810713093954e-05,
+      "loss": 0.9121,
+      "step": 1112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8354753863503747e-05,
+      "loss": 0.8584,
+      "step": 1113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8351397482245635e-05,
+      "loss": 0.8299,
+      "step": 1114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834803798841612e-05,
+      "loss": 0.7488,
+      "step": 1115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8344675383267297e-05,
+      "loss": 0.8728,
+      "step": 1116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8341309668052403e-05,
+      "loss": 0.8152,
+      "step": 1117
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8337940844025847e-05,
+      "loss": 0.8536,
+      "step": 1118
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8334568912443188e-05,
+      "loss": 0.8437,
+      "step": 1119
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8331193874561148e-05,
+      "loss": 0.8091,
+      "step": 1120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8327815731637612e-05,
+      "loss": 0.8524,
+      "step": 1121
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8324434484931607e-05,
+      "loss": 0.8912,
+      "step": 1122
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8321050135703335e-05,
+      "loss": 0.8642,
+      "step": 1123
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8317662685214135e-05,
+      "loss": 0.7561,
+      "step": 1124
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8314272134726518e-05,
+      "loss": 0.8935,
+      "step": 1125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8310878485504148e-05,
+      "loss": 0.8795,
+      "step": 1126
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8307481738811832e-05,
+      "loss": 0.8572,
+      "step": 1127
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8304081895915545e-05,
+      "loss": 0.8446,
+      "step": 1128
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830067895808241e-05,
+      "loss": 0.8662,
+      "step": 1129
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8297272926580703e-05,
+      "loss": 0.8137,
+      "step": 1130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8293863802679856e-05,
+      "loss": 0.8356,
+      "step": 1131
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8290451587650452e-05,
+      "loss": 0.8271,
+      "step": 1132
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8287036282764224e-05,
+      "loss": 0.8103,
+      "step": 1133
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8283617889294056e-05,
+      "loss": 0.8447,
+      "step": 1134
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8280196408513993e-05,
+      "loss": 0.8455,
+      "step": 1135
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8276771841699216e-05,
+      "loss": 0.2715,
+      "step": 1136
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8273344190126066e-05,
+      "loss": 0.7848,
+      "step": 1137
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8269913455072034e-05,
+      "loss": 0.8207,
+      "step": 1138
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826647963781575e-05,
+      "loss": 0.859,
+      "step": 1139
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8263042739637006e-05,
+      "loss": 0.8451,
+      "step": 1140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8259602761816736e-05,
+      "loss": 0.8135,
+      "step": 1141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825615970563702e-05,
+      "loss": 0.8583,
+      "step": 1142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8252713572381085e-05,
+      "loss": 0.8551,
+      "step": 1143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8249264363333312e-05,
+      "loss": 0.8705,
+      "step": 1144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824581207977922e-05,
+      "loss": 0.8557,
+      "step": 1145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824235672300548e-05,
+      "loss": 0.8661,
+      "step": 1146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8238898294299907e-05,
+      "loss": 0.8491,
+      "step": 1147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8235436794951456e-05,
+      "loss": 0.8052,
+      "step": 1148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823197222625023e-05,
+      "loss": 0.8866,
+      "step": 1149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8228504589487478e-05,
+      "loss": 0.8864,
+      "step": 1150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822503388595559e-05,
+      "loss": 0.8773,
+      "step": 1151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8221560116948103e-05,
+      "loss": 0.8134,
+      "step": 1152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8218083283759684e-05,
+      "loss": 0.8853,
+      "step": 1153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821460338768616e-05,
+      "loss": 0.2976,
+      "step": 1154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8211120430024486e-05,
+      "loss": 0.888,
+      "step": 1155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8207634412072765e-05,
+      "loss": 0.8644,
+      "step": 1156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8204145335130236e-05,
+      "loss": 0.8995,
+      "step": 1157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820065320049728e-05,
+      "loss": 0.8171,
+      "step": 1158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8197158009475423e-05,
+      "loss": 0.8949,
+      "step": 1159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8193659763367318e-05,
+      "loss": 0.8102,
+      "step": 1160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8190158463476767e-05,
+      "loss": 0.8639,
+      "step": 1161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818665411110871e-05,
+      "loss": 0.8293,
+      "step": 1162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8183146707569218e-05,
+      "loss": 0.8543,
+      "step": 1163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8179636254165497e-05,
+      "loss": 0.8381,
+      "step": 1164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817612275220591e-05,
+      "loss": 0.7922,
+      "step": 1165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8172606202999922e-05,
+      "loss": 0.8241,
+      "step": 1166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8169086607858172e-05,
+      "loss": 0.8889,
+      "step": 1167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8165563968092404e-05,
+      "loss": 0.8425,
+      "step": 1168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816203828501551e-05,
+      "loss": 0.8775,
+      "step": 1169
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815850955994152e-05,
+      "loss": 0.8049,
+      "step": 1170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8154977794185588e-05,
+      "loss": 0.8152,
+      "step": 1171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8151442989064003e-05,
+      "loss": 0.8853,
+      "step": 1172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814790514589419e-05,
+      "loss": 0.782,
+      "step": 1173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8144364265994712e-05,
+      "loss": 0.8775,
+      "step": 1174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8140820350685253e-05,
+      "loss": 0.8536,
+      "step": 1175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813727340128663e-05,
+      "loss": 0.8376,
+      "step": 1176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813372341912079e-05,
+      "loss": 0.8612,
+      "step": 1177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8130170405510826e-05,
+      "loss": 0.8744,
+      "step": 1178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812661436178094e-05,
+      "loss": 0.8518,
+      "step": 1179
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8123055289256476e-05,
+      "loss": 0.8871,
+      "step": 1180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8119493189263897e-05,
+      "loss": 0.8143,
+      "step": 1181
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8115928063130806e-05,
+      "loss": 0.8492,
+      "step": 1182
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8112359912185923e-05,
+      "loss": 0.9123,
+      "step": 1183
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8108788737759106e-05,
+      "loss": 0.8546,
+      "step": 1184
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810521454118133e-05,
+      "loss": 0.8524,
+      "step": 1185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8101637323784698e-05,
+      "loss": 0.7984,
+      "step": 1186
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8098057086902447e-05,
+      "loss": 0.8961,
+      "step": 1187
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8094473831868927e-05,
+      "loss": 0.8708,
+      "step": 1188
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809088756001963e-05,
+      "loss": 0.8582,
+      "step": 1189
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808729827269115e-05,
+      "loss": 0.8838,
+      "step": 1190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8083705971221226e-05,
+      "loss": 0.8979,
+      "step": 1191
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8080110656948705e-05,
+      "loss": 0.8939,
+      "step": 1192
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8076512331213564e-05,
+      "loss": 0.8426,
+      "step": 1193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.807291099535691e-05,
+      "loss": 0.8337,
+      "step": 1194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806930665072095e-05,
+      "loss": 0.8305,
+      "step": 1195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806569929864903e-05,
+      "loss": 0.8824,
+      "step": 1196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806208894048562e-05,
+      "loss": 0.881,
+      "step": 1197
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8058475577576295e-05,
+      "loss": 0.8719,
+      "step": 1198
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8054859211267757e-05,
+      "loss": 0.8836,
+      "step": 1199
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8051239842907833e-05,
+      "loss": 0.8268,
+      "step": 1200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8047617473845463e-05,
+      "loss": 0.8882,
+      "step": 1201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8043992105430703e-05,
+      "loss": 0.8341,
+      "step": 1202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8040363739014735e-05,
+      "loss": 0.8673,
+      "step": 1203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803673237594985e-05,
+      "loss": 0.8612,
+      "step": 1204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8033098017589462e-05,
+      "loss": 0.9172,
+      "step": 1205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8029460665288098e-05,
+      "loss": 0.7638,
+      "step": 1206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80258203204014e-05,
+      "loss": 0.8445,
+      "step": 1207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8022176984286125e-05,
+      "loss": 0.858,
+      "step": 1208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801853065830015e-05,
+      "loss": 0.2552,
+      "step": 1209
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801488134380247e-05,
+      "loss": 0.8254,
+      "step": 1210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8011229042153175e-05,
+      "loss": 0.7962,
+      "step": 1211
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8007573754713484e-05,
+      "loss": 0.8467,
+      "step": 1212
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8003915482845727e-05,
+      "loss": 0.8204,
+      "step": 1213
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8000254227913346e-05,
+      "loss": 0.8599,
+      "step": 1214
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799658999128089e-05,
+      "loss": 0.8999,
+      "step": 1215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7992922774314022e-05,
+      "loss": 0.8262,
+      "step": 1216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7989252578379517e-05,
+      "loss": 0.8701,
+      "step": 1217
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798557940484526e-05,
+      "loss": 0.8923,
+      "step": 1218
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7981903255080245e-05,
+      "loss": 0.8761,
+      "step": 1219
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7978224130454576e-05,
+      "loss": 0.8199,
+      "step": 1220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7974542032339464e-05,
+      "loss": 0.8222,
+      "step": 1221
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797085696210723e-05,
+      "loss": 0.8259,
+      "step": 1222
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79671689211313e-05,
+      "loss": 0.849,
+      "step": 1223
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7963477910786212e-05,
+      "loss": 0.8394,
+      "step": 1224
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795978393244761e-05,
+      "loss": 0.8007,
+      "step": 1225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795608698749224e-05,
+      "loss": 0.8349,
+      "step": 1226
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7952387077297956e-05,
+      "loss": 0.866,
+      "step": 1227
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7948684203243715e-05,
+      "loss": 0.8032,
+      "step": 1228
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7944978366709584e-05,
+      "loss": 0.2625,
+      "step": 1229
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7941269569076726e-05,
+      "loss": 0.7694,
+      "step": 1230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7937557811727416e-05,
+      "loss": 0.7863,
+      "step": 1231
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793384309604503e-05,
+      "loss": 0.8398,
+      "step": 1232
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7930125423414038e-05,
+      "loss": 0.8479,
+      "step": 1233
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7926404795220033e-05,
+      "loss": 0.8711,
+      "step": 1234
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7922681212849678e-05,
+      "loss": 0.8062,
+      "step": 1235
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7918954677690766e-05,
+      "loss": 0.8801,
+      "step": 1236
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7915225191132174e-05,
+      "loss": 0.8032,
+      "step": 1237
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791149275456389e-05,
+      "loss": 0.8605,
+      "step": 1238
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7907757369376984e-05,
+      "loss": 0.8953,
+      "step": 1239
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7904019036963652e-05,
+      "loss": 0.8712,
+      "step": 1240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7900277758717165e-05,
+      "loss": 0.8378,
+      "step": 1241
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.78965335360319e-05,
+      "loss": 0.8736,
+      "step": 1242
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789278637030333e-05,
+      "loss": 0.2565,
+      "step": 1243
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7889036262928025e-05,
+      "loss": 0.8739,
+      "step": 1244
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788528321530366e-05,
+      "loss": 0.8023,
+      "step": 1245
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7881527228828992e-05,
+      "loss": 0.8952,
+      "step": 1246
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787776830490388e-05,
+      "loss": 0.8156,
+      "step": 1247
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787400644492928e-05,
+      "loss": 0.277,
+      "step": 1248
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787024165030724e-05,
+      "loss": 0.7959,
+      "step": 1249
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7866473922440898e-05,
+      "loss": 0.8145,
+      "step": 1250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7862703262734488e-05,
+      "loss": 0.8711,
+      "step": 1251
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7858929672593344e-05,
+      "loss": 0.7795,
+      "step": 1252
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785515315342388e-05,
+      "loss": 0.8676,
+      "step": 1253
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7851373706633607e-05,
+      "loss": 0.8348,
+      "step": 1254
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7847591333631133e-05,
+      "loss": 0.8855,
+      "step": 1255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7843806035826143e-05,
+      "loss": 0.8351,
+      "step": 1256
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7840017814629426e-05,
+      "loss": 0.795,
+      "step": 1257
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783622667145285e-05,
+      "loss": 0.8115,
+      "step": 1258
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7832432607709385e-05,
+      "loss": 0.865,
+      "step": 1259
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7828635624813067e-05,
+      "loss": 0.8583,
+      "step": 1260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782483572417905e-05,
+      "loss": 0.8512,
+      "step": 1261
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7821032907223547e-05,
+      "loss": 0.8355,
+      "step": 1262
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7817227175363877e-05,
+      "loss": 0.8745,
+      "step": 1263
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781341853001844e-05,
+      "loss": 0.8478,
+      "step": 1264
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7809606972606713e-05,
+      "loss": 0.8724,
+      "step": 1265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780579250454927e-05,
+      "loss": 0.8278,
+      "step": 1266
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780197512726777e-05,
+      "loss": 0.89,
+      "step": 1267
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7798154842184943e-05,
+      "loss": 0.8287,
+      "step": 1268
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7794331650724623e-05,
+      "loss": 0.2673,
+      "step": 1269
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779050555431171e-05,
+      "loss": 0.8311,
+      "step": 1270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778667655437219e-05,
+      "loss": 0.2827,
+      "step": 1271
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7782844652333136e-05,
+      "loss": 0.9231,
+      "step": 1272
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7779009849622705e-05,
+      "loss": 0.837,
+      "step": 1273
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7775172147670124e-05,
+      "loss": 0.8921,
+      "step": 1274
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7771331547905707e-05,
+      "loss": 0.7942,
+      "step": 1275
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7767488051760858e-05,
+      "loss": 0.8214,
+      "step": 1276
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776364166066804e-05,
+      "loss": 0.8623,
+      "step": 1277
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7759792376060806e-05,
+      "loss": 0.8806,
+      "step": 1278
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7755940199373792e-05,
+      "loss": 0.8247,
+      "step": 1279
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7752085132042707e-05,
+      "loss": 0.8906,
+      "step": 1280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7748227175504333e-05,
+      "loss": 0.8253,
+      "step": 1281
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7744366331196533e-05,
+      "loss": 0.8846,
+      "step": 1282
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774050260055825e-05,
+      "loss": 0.8468,
+      "step": 1283
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7736635985029496e-05,
+      "loss": 0.8102,
+      "step": 1284
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7732766486051365e-05,
+      "loss": 0.8592,
+      "step": 1285
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772889410506602e-05,
+      "loss": 0.8676,
+      "step": 1286
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7725018843516696e-05,
+      "loss": 0.8315,
+      "step": 1287
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7721140702847707e-05,
+      "loss": 0.797,
+      "step": 1288
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771725968450444e-05,
+      "loss": 0.8503,
+      "step": 1289
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7713375789933356e-05,
+      "loss": 0.8095,
+      "step": 1290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7709489020581983e-05,
+      "loss": 0.8051,
+      "step": 1291
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770559937789892e-05,
+      "loss": 0.8239,
+      "step": 1292
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7701706863333842e-05,
+      "loss": 0.8781,
+      "step": 1293
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769781147833749e-05,
+      "loss": 0.834,
+      "step": 1294
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769391322436168e-05,
+      "loss": 0.8045,
+      "step": 1295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769001210285929e-05,
+      "loss": 0.7915,
+      "step": 1296
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7686108115284273e-05,
+      "loss": 0.8683,
+      "step": 1297
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7682201263091646e-05,
+      "loss": 0.2718,
+      "step": 1298
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.76782915477375e-05,
+      "loss": 0.8578,
+      "step": 1299
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.767437897067898e-05,
+      "loss": 0.7822,
+      "step": 1300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.767046353337431e-05,
+      "loss": 0.2732,
+      "step": 1301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7666545237282783e-05,
+      "loss": 0.8398,
+      "step": 1302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766262408386474e-05,
+      "loss": 0.8634,
+      "step": 1303
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76587000745816e-05,
+      "loss": 0.8717,
+      "step": 1304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765477321089585e-05,
+      "loss": 0.7889,
+      "step": 1305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7650843494271028e-05,
+      "loss": 0.842,
+      "step": 1306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7646910926171747e-05,
+      "loss": 0.8431,
+      "step": 1307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7642975508063676e-05,
+      "loss": 0.8381,
+      "step": 1308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7639037241413544e-05,
+      "loss": 0.2598,
+      "step": 1309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7635096127689152e-05,
+      "loss": 0.8359,
+      "step": 1310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7631152168359352e-05,
+      "loss": 0.847,
+      "step": 1311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762720536489406e-05,
+      "loss": 0.8932,
+      "step": 1312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7623255718764252e-05,
+      "loss": 0.8654,
+      "step": 1313
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7619303231441965e-05,
+      "loss": 0.8094,
+      "step": 1314
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7615347904400296e-05,
+      "loss": 0.8275,
+      "step": 1315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7611389739113396e-05,
+      "loss": 0.7994,
+      "step": 1316
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7607428737056473e-05,
+      "loss": 0.2793,
+      "step": 1317
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76034648997058e-05,
+      "loss": 0.8676,
+      "step": 1318
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7599498228538696e-05,
+      "loss": 0.8695,
+      "step": 1319
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759552872503355e-05,
+      "loss": 0.8388,
+      "step": 1320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759155639066979e-05,
+      "loss": 0.8078,
+      "step": 1321
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758758122692791e-05,
+      "loss": 0.826,
+      "step": 1322
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7583603235289463e-05,
+      "loss": 0.2907,
+      "step": 1323
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757962241723704e-05,
+      "loss": 0.279,
+      "step": 1324
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75756387742543e-05,
+      "loss": 0.8134,
+      "step": 1325
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7571652307825947e-05,
+      "loss": 0.8349,
+      "step": 1326
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756766301943774e-05,
+      "loss": 0.8624,
+      "step": 1327
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7563670910576492e-05,
+      "loss": 0.877,
+      "step": 1328
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755967598273006e-05,
+      "loss": 0.8264,
+      "step": 1329
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7555678237387362e-05,
+      "loss": 0.8214,
+      "step": 1330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7551677676038354e-05,
+      "loss": 0.8396,
+      "step": 1331
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754767430017405e-05,
+      "loss": 0.8958,
+      "step": 1332
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7543668111286514e-05,
+      "loss": 0.8009,
+      "step": 1333
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7539659110868856e-05,
+      "loss": 0.7932,
+      "step": 1334
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7535647300415228e-05,
+      "loss": 0.2697,
+      "step": 1335
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7531632681420836e-05,
+      "loss": 0.942,
+      "step": 1336
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7527615255381934e-05,
+      "loss": 0.8957,
+      "step": 1337
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7523595023795814e-05,
+      "loss": 0.8424,
+      "step": 1338
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7519571988160825e-05,
+      "loss": 0.8227,
+      "step": 1339
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751554614997635e-05,
+      "loss": 0.7775,
+      "step": 1340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7511517510742825e-05,
+      "loss": 0.8406,
+      "step": 1341
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7507486071961727e-05,
+      "loss": 0.8314,
+      "step": 1342
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7503451835135572e-05,
+      "loss": 0.7901,
+      "step": 1343
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7499414801767925e-05,
+      "loss": 0.7755,
+      "step": 1344
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7495374973363392e-05,
+      "loss": 0.7935,
+      "step": 1345
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7491332351427618e-05,
+      "loss": 0.8734,
+      "step": 1346
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7487286937467293e-05,
+      "loss": 0.7902,
+      "step": 1347
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7483238732990145e-05,
+      "loss": 0.916,
+      "step": 1348
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747918773950494e-05,
+      "loss": 0.2593,
+      "step": 1349
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7475133958521485e-05,
+      "loss": 0.8766,
+      "step": 1350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7471077391550636e-05,
+      "loss": 0.8011,
+      "step": 1351
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7467018040104266e-05,
+      "loss": 0.8306,
+      "step": 1352
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7462955905695307e-05,
+      "loss": 0.8465,
+      "step": 1353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7458890989837714e-05,
+      "loss": 0.8616,
+      "step": 1354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7454823294046485e-05,
+      "loss": 0.8294,
+      "step": 1355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7450752819837658e-05,
+      "loss": 0.8199,
+      "step": 1356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744667956872829e-05,
+      "loss": 0.8375,
+      "step": 1357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74426035422365e-05,
+      "loss": 0.7927,
+      "step": 1358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743852474188141e-05,
+      "loss": 0.8556,
+      "step": 1359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7434443169183204e-05,
+      "loss": 0.8071,
+      "step": 1360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7430358825663082e-05,
+      "loss": 0.8136,
+      "step": 1361
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7426271712843277e-05,
+      "loss": 0.8809,
+      "step": 1362
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7422181832247064e-05,
+      "loss": 0.8467,
+      "step": 1363
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7418089185398746e-05,
+      "loss": 0.8755,
+      "step": 1364
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741399377382365e-05,
+      "loss": 0.8306,
+      "step": 1365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7409895599048142e-05,
+      "loss": 0.7851,
+      "step": 1366
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7405794662599613e-05,
+      "loss": 0.8831,
+      "step": 1367
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7401690966006483e-05,
+      "loss": 0.8595,
+      "step": 1368
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7397584510798208e-05,
+      "loss": 0.8587,
+      "step": 1369
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7393475298505263e-05,
+      "loss": 0.9361,
+      "step": 1370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7389363330659154e-05,
+      "loss": 0.8462,
+      "step": 1371
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7385248608792414e-05,
+      "loss": 0.8431,
+      "step": 1372
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7381131134438606e-05,
+      "loss": 0.9001,
+      "step": 1373
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7377010909132308e-05,
+      "loss": 0.8779,
+      "step": 1374
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7372887934409138e-05,
+      "loss": 0.8561,
+      "step": 1375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7368762211805726e-05,
+      "loss": 0.8524,
+      "step": 1376
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736463374285974e-05,
+      "loss": 0.8448,
+      "step": 1377
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7360502529109854e-05,
+      "loss": 0.8356,
+      "step": 1378
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7356368572095778e-05,
+      "loss": 0.8735,
+      "step": 1379
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7352231873358238e-05,
+      "loss": 0.8951,
+      "step": 1380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734809243443899e-05,
+      "loss": 0.8088,
+      "step": 1381
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73439502568808e-05,
+      "loss": 0.8433,
+      "step": 1382
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733980534222746e-05,
+      "loss": 0.8109,
+      "step": 1383
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733565769202379e-05,
+      "loss": 0.8855,
+      "step": 1384
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7331507307815613e-05,
+      "loss": 0.8723,
+      "step": 1385
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7327354191149787e-05,
+      "loss": 0.8684,
+      "step": 1386
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7323198343574173e-05,
+      "loss": 0.2434,
+      "step": 1387
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731903976663767e-05,
+      "loss": 0.8153,
+      "step": 1388
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731487846189017e-05,
+      "loss": 0.8806,
+      "step": 1389
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73107144308826e-05,
+      "loss": 0.851,
+      "step": 1390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73065476751669e-05,
+      "loss": 0.7847,
+      "step": 1391
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730237819629602e-05,
+      "loss": 0.8265,
+      "step": 1392
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7298205995823924e-05,
+      "loss": 0.8862,
+      "step": 1393
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72940310753056e-05,
+      "loss": 0.8168,
+      "step": 1394
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728985343629704e-05,
+      "loss": 0.8761,
+      "step": 1395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728567308035525e-05,
+      "loss": 0.8268,
+      "step": 1396
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7281490009038258e-05,
+      "loss": 0.8367,
+      "step": 1397
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727730422390509e-05,
+      "loss": 0.8501,
+      "step": 1398
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7273115726515802e-05,
+      "loss": 0.2818,
+      "step": 1399
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7268924518431437e-05,
+      "loss": 0.8902,
+      "step": 1400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7264730601214072e-05,
+      "loss": 0.8226,
+      "step": 1401
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7260533976426773e-05,
+      "loss": 0.9022,
+      "step": 1402
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7256334645633634e-05,
+      "loss": 0.8438,
+      "step": 1403
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7252132610399743e-05,
+      "loss": 0.7923,
+      "step": 1404
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72479278722912e-05,
+      "loss": 0.8694,
+      "step": 1405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7243720432875117e-05,
+      "loss": 0.8141,
+      "step": 1406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723951029371961e-05,
+      "loss": 0.895,
+      "step": 1407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72352974563938e-05,
+      "loss": 0.847,
+      "step": 1408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723108192246781e-05,
+      "loss": 0.8343,
+      "step": 1409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7226863693512777e-05,
+      "loss": 0.8653,
+      "step": 1410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722264277110084e-05,
+      "loss": 0.8441,
+      "step": 1411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721841915680513e-05,
+      "loss": 0.8899,
+      "step": 1412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72141928521998e-05,
+      "loss": 0.8383,
+      "step": 1413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7209963858859995e-05,
+      "loss": 0.8092,
+      "step": 1414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720573217836186e-05,
+      "loss": 0.8662,
+      "step": 1415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720149781228255e-05,
+      "loss": 0.8276,
+      "step": 1416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719726076220021e-05,
+      "loss": 0.8927,
+      "step": 1417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7193021029693997e-05,
+      "loss": 0.841,
+      "step": 1418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7188778616344056e-05,
+      "loss": 0.7868,
+      "step": 1419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7184533523731545e-05,
+      "loss": 0.8142,
+      "step": 1420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718028575343861e-05,
+      "loss": 0.8558,
+      "step": 1421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717603530704839e-05,
+      "loss": 0.8605,
+      "step": 1422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717178218614504e-05,
+      "loss": 0.8735,
+      "step": 1423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71675263923137e-05,
+      "loss": 0.8305,
+      "step": 1424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71632679271405e-05,
+      "loss": 0.8726,
+      "step": 1425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715900679221258e-05,
+      "loss": 0.8912,
+      "step": 1426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715474298911806e-05,
+      "loss": 0.8236,
+      "step": 1427
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715047651944607e-05,
+      "loss": 0.2786,
+      "step": 1428
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7146207384786727e-05,
+      "loss": 0.8642,
+      "step": 1429
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7141935586731132e-05,
+      "loss": 0.8406,
+      "step": 1430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713766112687139e-05,
+      "loss": 0.842,
+      "step": 1431
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71333840068006e-05,
+      "loss": 0.8672,
+      "step": 1432
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7129104228112843e-05,
+      "loss": 0.8572,
+      "step": 1433
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712482179240319e-05,
+      "loss": 0.8667,
+      "step": 1434
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7120536701267723e-05,
+      "loss": 0.8319,
+      "step": 1435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7116248956303486e-05,
+      "loss": 0.8542,
+      "step": 1436
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711195855910853e-05,
+      "loss": 0.7798,
+      "step": 1437
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710766551128188e-05,
+      "loss": 0.8519,
+      "step": 1438
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7103369814423573e-05,
+      "loss": 0.8559,
+      "step": 1439
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70990714701346e-05,
+      "loss": 0.8743,
+      "step": 1440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7094770480016972e-05,
+      "loss": 0.8733,
+      "step": 1441
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7090466845673665e-05,
+      "loss": 0.8585,
+      "step": 1442
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7086160568708644e-05,
+      "loss": 0.8596,
+      "step": 1443
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7081851650726867e-05,
+      "loss": 0.8901,
+      "step": 1444
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7077540093334265e-05,
+      "loss": 0.8255,
+      "step": 1445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707322589813776e-05,
+      "loss": 0.8467,
+      "step": 1446
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7068909066745258e-05,
+      "loss": 0.864,
+      "step": 1447
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7064589600765642e-05,
+      "loss": 0.8171,
+      "step": 1448
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706026750180878e-05,
+      "loss": 0.8255,
+      "step": 1449
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7055942771485522e-05,
+      "loss": 0.8831,
+      "step": 1450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70516154114077e-05,
+      "loss": 0.8847,
+      "step": 1451
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704728542318812e-05,
+      "loss": 0.7907,
+      "step": 1452
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7042952808440576e-05,
+      "loss": 0.7931,
+      "step": 1453
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703861756877983e-05,
+      "loss": 0.8658,
+      "step": 1454
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7034279705821638e-05,
+      "loss": 0.8537,
+      "step": 1455
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7029939221182713e-05,
+      "loss": 0.82,
+      "step": 1456
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7025596116480763e-05,
+      "loss": 0.7831,
+      "step": 1457
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7021250393334466e-05,
+      "loss": 0.8367,
+      "step": 1458
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7016902053363477e-05,
+      "loss": 0.847,
+      "step": 1459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7012551098188425e-05,
+      "loss": 0.8435,
+      "step": 1460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700819752943091e-05,
+      "loss": 0.8046,
+      "step": 1461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700384134871351e-05,
+      "loss": 0.8367,
+      "step": 1462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699948255765978e-05,
+      "loss": 0.7923,
+      "step": 1463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6995121157894246e-05,
+      "loss": 0.7975,
+      "step": 1464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69907571510424e-05,
+      "loss": 0.781,
+      "step": 1465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698639053873071e-05,
+      "loss": 0.8137,
+      "step": 1466
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6982021322586618e-05,
+      "loss": 0.8125,
+      "step": 1467
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6977649504238533e-05,
+      "loss": 0.7966,
+      "step": 1468
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697327508531584e-05,
+      "loss": 0.8838,
+      "step": 1469
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6968898067448875e-05,
+      "loss": 0.8477,
+      "step": 1470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696451845226897e-05,
+      "loss": 0.2682,
+      "step": 1471
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6960136241408397e-05,
+      "loss": 0.8037,
+      "step": 1472
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695575143650042e-05,
+      "loss": 0.8416,
+      "step": 1473
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695136403917925e-05,
+      "loss": 0.7792,
+      "step": 1474
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694697405108008e-05,
+      "loss": 0.8634,
+      "step": 1475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6942581473839058e-05,
+      "loss": 0.7975,
+      "step": 1476
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69381863090933e-05,
+      "loss": 0.8658,
+      "step": 1477
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6933788558480892e-05,
+      "loss": 0.8528,
+      "step": 1478
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6929388223640867e-05,
+      "loss": 0.9349,
+      "step": 1479
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6924985306213243e-05,
+      "loss": 0.782,
+      "step": 1480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692057980783899e-05,
+      "loss": 0.7898,
+      "step": 1481
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6916171730160032e-05,
+      "loss": 0.7764,
+      "step": 1482
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6911761074819274e-05,
+      "loss": 0.8782,
+      "step": 1483
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690734784346056e-05,
+      "loss": 0.7894,
+      "step": 1484
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6902932037728712e-05,
+      "loss": 0.807,
+      "step": 1485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6898513659269497e-05,
+      "loss": 0.8472,
+      "step": 1486
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689409270972965e-05,
+      "loss": 0.8762,
+      "step": 1487
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688966919075687e-05,
+      "loss": 0.7919,
+      "step": 1488
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688524310399979e-05,
+      "loss": 0.8737,
+      "step": 1489
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6880814451108024e-05,
+      "loss": 0.8377,
+      "step": 1490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6876383233732135e-05,
+      "loss": 0.8402,
+      "step": 1491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6871949453523643e-05,
+      "loss": 0.8513,
+      "step": 1492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6867513112135012e-05,
+      "loss": 0.8101,
+      "step": 1493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6863074211219673e-05,
+      "loss": 0.8494,
+      "step": 1494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6858632752432017e-05,
+      "loss": 0.8415,
+      "step": 1495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685418873742736e-05,
+      "loss": 0.8596,
+      "step": 1496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6849742167862005e-05,
+      "loss": 0.8666,
+      "step": 1497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6845293045393187e-05,
+      "loss": 0.8308,
+      "step": 1498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6840841371679096e-05,
+      "loss": 0.7944,
+      "step": 1499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683638714837887e-05,
+      "loss": 0.7847,
+      "step": 1500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6831930377152607e-05,
+      "loss": 0.8534,
+      "step": 1501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6827471059661343e-05,
+      "loss": 0.8069,
+      "step": 1502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6823009197567077e-05,
+      "loss": 0.8451,
+      "step": 1503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6818544792532742e-05,
+      "loss": 0.292,
+      "step": 1504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681407784622222e-05,
+      "loss": 0.8741,
+      "step": 1505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680960836030036e-05,
+      "loss": 0.8343,
+      "step": 1506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6805136336432926e-05,
+      "loss": 0.8286,
+      "step": 1507
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680066177628665e-05,
+      "loss": 0.7962,
+      "step": 1508
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6796184681529208e-05,
+      "loss": 0.8616,
+      "step": 1509
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679170505382921e-05,
+      "loss": 0.8321,
+      "step": 1510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6787222894856222e-05,
+      "loss": 0.8754,
+      "step": 1511
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6782738206280744e-05,
+      "loss": 0.8377,
+      "step": 1512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6778250989774223e-05,
+      "loss": 0.7944,
+      "step": 1513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677376124700905e-05,
+      "loss": 0.8835,
+      "step": 1514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676926897965855e-05,
+      "loss": 0.8316,
+      "step": 1515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6764774189397e-05,
+      "loss": 0.9189,
+      "step": 1516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67602768778996e-05,
+      "loss": 0.816,
+      "step": 1517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6755777046842518e-05,
+      "loss": 0.7451,
+      "step": 1518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675127469790283e-05,
+      "loss": 0.8202,
+      "step": 1519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674676983275857e-05,
+      "loss": 0.8256,
+      "step": 1520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6742262453088708e-05,
+      "loss": 0.8692,
+      "step": 1521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6737752560573136e-05,
+      "loss": 0.7767,
+      "step": 1522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6733240156892708e-05,
+      "loss": 0.8273,
+      "step": 1523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672872524372919e-05,
+      "loss": 0.8433,
+      "step": 1524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6724207822765298e-05,
+      "loss": 0.8352,
+      "step": 1525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6719687895684673e-05,
+      "loss": 0.8009,
+      "step": 1526
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6715165464171903e-05,
+      "loss": 0.8195,
+      "step": 1527
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671064052991249e-05,
+      "loss": 0.8177,
+      "step": 1528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670611309459289e-05,
+      "loss": 0.9223,
+      "step": 1529
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6701583159900477e-05,
+      "loss": 0.8819,
+      "step": 1530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6697050727523557e-05,
+      "loss": 0.3176,
+      "step": 1531
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669251579915138e-05,
+      "loss": 0.2747,
+      "step": 1532
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6687978376474107e-05,
+      "loss": 0.8781,
+      "step": 1533
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6683438461182847e-05,
+      "loss": 0.8747,
+      "step": 1534
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6678896054969622e-05,
+      "loss": 0.8055,
+      "step": 1535
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6674351159527394e-05,
+      "loss": 0.8448,
+      "step": 1536
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6669803776550044e-05,
+      "loss": 0.899,
+      "step": 1537
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666525390773239e-05,
+      "loss": 0.8568,
+      "step": 1538
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6660701554770176e-05,
+      "loss": 0.8646,
+      "step": 1539
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6656146719360054e-05,
+      "loss": 0.8022,
+      "step": 1540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665158940319962e-05,
+      "loss": 0.8034,
+      "step": 1541
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664702960798739e-05,
+      "loss": 0.8233,
+      "step": 1542
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6642467335422803e-05,
+      "loss": 0.8346,
+      "step": 1543
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6637902587206222e-05,
+      "loss": 0.8401,
+      "step": 1544
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6633335365038926e-05,
+      "loss": 0.7873,
+      "step": 1545
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6628765670623127e-05,
+      "loss": 0.8188,
+      "step": 1546
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6624193505661952e-05,
+      "loss": 0.8328,
+      "step": 1547
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661961887185945e-05,
+      "loss": 0.8662,
+      "step": 1548
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6615041770920587e-05,
+      "loss": 0.8378,
+      "step": 1549
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6610462204551256e-05,
+      "loss": 0.8173,
+      "step": 1550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6605880174458265e-05,
+      "loss": 0.8645,
+      "step": 1551
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6601295682349335e-05,
+      "loss": 0.774,
+      "step": 1552
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6596708729933107e-05,
+      "loss": 0.7837,
+      "step": 1553
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6592119318919152e-05,
+      "loss": 0.8691,
+      "step": 1554
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658752745101794e-05,
+      "loss": 0.8128,
+      "step": 1555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6582933127940858e-05,
+      "loss": 0.8313,
+      "step": 1556
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6578336351400226e-05,
+      "loss": 0.8525,
+      "step": 1557
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6573737123109254e-05,
+      "loss": 0.8437,
+      "step": 1558
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6569135444782082e-05,
+      "loss": 0.8289,
+      "step": 1559
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6564531318133762e-05,
+      "loss": 0.8116,
+      "step": 1560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6559924744880248e-05,
+      "loss": 0.8067,
+      "step": 1561
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6555315726738417e-05,
+      "loss": 0.8159,
+      "step": 1562
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6550704265426053e-05,
+      "loss": 0.867,
+      "step": 1563
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6546090362661852e-05,
+      "loss": 0.8444,
+      "step": 1564
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654147402016541e-05,
+      "loss": 0.7926,
+      "step": 1565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6536855239657253e-05,
+      "loss": 0.3031,
+      "step": 1566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.65322340228588e-05,
+      "loss": 0.8583,
+      "step": 1567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652761037149238e-05,
+      "loss": 0.2949,
+      "step": 1568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652298428728123e-05,
+      "loss": 0.8666,
+      "step": 1569
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651835577194949e-05,
+      "loss": 0.8395,
+      "step": 1570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6513724827222225e-05,
+      "loss": 0.2654,
+      "step": 1571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6509091454825376e-05,
+      "loss": 0.8379,
+      "step": 1572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6504455656485813e-05,
+      "loss": 0.858,
+      "step": 1573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64998174339313e-05,
+      "loss": 0.8597,
+      "step": 1574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64951767888905e-05,
+      "loss": 0.7985,
+      "step": 1575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6490533723092994e-05,
+      "loss": 0.8496,
+      "step": 1576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6485888238269248e-05,
+      "loss": 0.2897,
+      "step": 1577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648124033615064e-05,
+      "loss": 0.8347,
+      "step": 1578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6476590018469445e-05,
+      "loss": 0.8602,
+      "step": 1579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647193728695884e-05,
+      "loss": 0.8108,
+      "step": 1580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6467282143352904e-05,
+      "loss": 0.8463,
+      "step": 1581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6462624589386605e-05,
+      "loss": 0.2578,
+      "step": 1582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6457964626795825e-05,
+      "loss": 0.8666,
+      "step": 1583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6453302257317333e-05,
+      "loss": 0.8098,
+      "step": 1584
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6448637482688794e-05,
+      "loss": 0.8109,
+      "step": 1585
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644397030464877e-05,
+      "loss": 0.8587,
+      "step": 1586
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643930072493673e-05,
+      "loss": 0.7658,
+      "step": 1587
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6434628745293025e-05,
+      "loss": 0.8106,
+      "step": 1588
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6429954367458903e-05,
+      "loss": 0.7377,
+      "step": 1589
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642527759317651e-05,
+      "loss": 0.8728,
+      "step": 1590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6420598424188883e-05,
+      "loss": 0.8612,
+      "step": 1591
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641591686223995e-05,
+      "loss": 0.8402,
+      "step": 1592
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6411232909074533e-05,
+      "loss": 0.7499,
+      "step": 1593
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6406546566438347e-05,
+      "loss": 0.8228,
+      "step": 1594
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640185783607799e-05,
+      "loss": 0.2911,
+      "step": 1595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6397166719740958e-05,
+      "loss": 0.8322,
+      "step": 1596
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6392473219175632e-05,
+      "loss": 0.7997,
+      "step": 1597
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6387777336131286e-05,
+      "loss": 0.8261,
+      "step": 1598
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6383079072358076e-05,
+      "loss": 0.8002,
+      "step": 1599
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6378378429607046e-05,
+      "loss": 0.7966,
+      "step": 1600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637367540963013e-05,
+      "loss": 0.7861,
+      "step": 1601
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636897001418015e-05,
+      "loss": 0.8563,
+      "step": 1602
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6364262245010807e-05,
+      "loss": 0.2664,
+      "step": 1603
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6359552103876692e-05,
+      "loss": 0.845,
+      "step": 1604
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6354839592533273e-05,
+      "loss": 0.7635,
+      "step": 1605
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6350124712736914e-05,
+      "loss": 0.805,
+      "step": 1606
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6345407466244844e-05,
+      "loss": 0.8352,
+      "step": 1607
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634068785481519e-05,
+      "loss": 0.2466,
+      "step": 1608
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6335965880206957e-05,
+      "loss": 0.8076,
+      "step": 1609
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633124154418002e-05,
+      "loss": 0.7955,
+      "step": 1610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632651484849515e-05,
+      "loss": 0.8536,
+      "step": 1611
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6321785794913983e-05,
+      "loss": 0.8415,
+      "step": 1612
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6317054385199042e-05,
+      "loss": 0.8033,
+      "step": 1613
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6312320621113733e-05,
+      "loss": 0.8145,
+      "step": 1614
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6307584504422326e-05,
+      "loss": 0.8139,
+      "step": 1615
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6302846036889974e-05,
+      "loss": 0.8436,
+      "step": 1616
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6298105220282715e-05,
+      "loss": 0.8252,
+      "step": 1617
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6293362056367446e-05,
+      "loss": 0.7926,
+      "step": 1618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6288616546911955e-05,
+      "loss": 0.8419,
+      "step": 1619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6283868693684887e-05,
+      "loss": 0.8311,
+      "step": 1620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6279118498455776e-05,
+      "loss": 0.796,
+      "step": 1621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6274365962995026e-05,
+      "loss": 0.2913,
+      "step": 1622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6269611089073904e-05,
+      "loss": 0.8873,
+      "step": 1623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6264853878464557e-05,
+      "loss": 0.8086,
+      "step": 1624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6260094332940002e-05,
+      "loss": 0.8768,
+      "step": 1625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625533245427412e-05,
+      "loss": 0.8245,
+      "step": 1626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6250568244241675e-05,
+      "loss": 0.855,
+      "step": 1627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6245801704618285e-05,
+      "loss": 0.8307,
+      "step": 1628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6241032837180443e-05,
+      "loss": 0.8296,
+      "step": 1629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623626164370551e-05,
+      "loss": 0.8171,
+      "step": 1630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623148812597171e-05,
+      "loss": 0.8015,
+      "step": 1631
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622671228575814e-05,
+      "loss": 0.8212,
+      "step": 1632
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622193412484476e-05,
+      "loss": 0.8102,
+      "step": 1633
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621715364501239e-05,
+      "loss": 0.751,
+      "step": 1634
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6212370848042714e-05,
+      "loss": 0.7723,
+      "step": 1635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6207585735718296e-05,
+      "loss": 0.8446,
+      "step": 1636
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6202798309822542e-05,
+      "loss": 0.8101,
+      "step": 1637
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619800857213973e-05,
+      "loss": 0.8197,
+      "step": 1638
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6193216524455e-05,
+      "loss": 0.7745,
+      "step": 1639
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6188422168554348e-05,
+      "loss": 0.7933,
+      "step": 1640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618362550622464e-05,
+      "loss": 0.8314,
+      "step": 1641
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6178826539253588e-05,
+      "loss": 0.8431,
+      "step": 1642
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6174025269429772e-05,
+      "loss": 0.794,
+      "step": 1643
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6169221698542632e-05,
+      "loss": 0.8481,
+      "step": 1644
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616441582838246e-05,
+      "loss": 0.8829,
+      "step": 1645
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6159607660740406e-05,
+      "loss": 0.7717,
+      "step": 1646
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6154797197408477e-05,
+      "loss": 0.8403,
+      "step": 1647
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614998444017954e-05,
+      "loss": 0.8302,
+      "step": 1648
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6145169390847305e-05,
+      "loss": 0.2461,
+      "step": 1649
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614035205120635e-05,
+      "loss": 0.89,
+      "step": 1650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6135532423052102e-05,
+      "loss": 0.8896,
+      "step": 1651
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6130710508180837e-05,
+      "loss": 0.8148,
+      "step": 1652
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6125886308389686e-05,
+      "loss": 0.858,
+      "step": 1653
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612105982547663e-05,
+      "loss": 0.8175,
+      "step": 1654
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6116231061240504e-05,
+      "loss": 0.7848,
+      "step": 1655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6111400017480994e-05,
+      "loss": 0.852,
+      "step": 1656
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6106566695998627e-05,
+      "loss": 0.8497,
+      "step": 1657
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6101731098594794e-05,
+      "loss": 0.8539,
+      "step": 1658
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6096893227071718e-05,
+      "loss": 0.8379,
+      "step": 1659
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6092053083232476e-05,
+      "loss": 0.818,
+      "step": 1660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6087210668880998e-05,
+      "loss": 0.2639,
+      "step": 1661
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6082365985822055e-05,
+      "loss": 0.7986,
+      "step": 1662
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607751903586126e-05,
+      "loss": 0.829,
+      "step": 1663
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6072669820805078e-05,
+      "loss": 0.7963,
+      "step": 1664
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606781834246081e-05,
+      "loss": 0.8107,
+      "step": 1665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606296460263661e-05,
+      "loss": 0.871,
+      "step": 1666
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605810860314147e-05,
+      "loss": 0.7714,
+      "step": 1667
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6053250345785216e-05,
+      "loss": 0.823,
+      "step": 1668
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604838983237854e-05,
+      "loss": 0.8123,
+      "step": 1669
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604352706473294e-05,
+      "loss": 0.8162,
+      "step": 1670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6038662044660783e-05,
+      "loss": 0.8247,
+      "step": 1671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6033794773975267e-05,
+      "loss": 0.8135,
+      "step": 1672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6028925254490426e-05,
+      "loss": 0.9073,
+      "step": 1673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6024053488021122e-05,
+      "loss": 0.2791,
+      "step": 1674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6019179476383076e-05,
+      "loss": 0.8271,
+      "step": 1675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601430322139284e-05,
+      "loss": 0.8925,
+      "step": 1676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600942472486779e-05,
+      "loss": 0.7834,
+      "step": 1677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6004543988626146e-05,
+      "loss": 0.27,
+      "step": 1678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5999661014486956e-05,
+      "loss": 0.8259,
+      "step": 1679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599477580427012e-05,
+      "loss": 0.8697,
+      "step": 1680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598988835979635e-05,
+      "loss": 0.782,
+      "step": 1681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5984998682887206e-05,
+      "loss": 0.8663,
+      "step": 1682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5980106775365065e-05,
+      "loss": 0.8151,
+      "step": 1683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597521263905315e-05,
+      "loss": 0.2676,
+      "step": 1684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5970316275775512e-05,
+      "loss": 0.8695,
+      "step": 1685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5965417687357023e-05,
+      "loss": 0.8733,
+      "step": 1686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5960516875623394e-05,
+      "loss": 0.8146,
+      "step": 1687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595561384240116e-05,
+      "loss": 0.8506,
+      "step": 1688
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5950708589517677e-05,
+      "loss": 0.852,
+      "step": 1689
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5945801118801153e-05,
+      "loss": 0.7976,
+      "step": 1690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5940891432080587e-05,
+      "loss": 0.8333,
+      "step": 1691
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5935979531185837e-05,
+      "loss": 0.7938,
+      "step": 1692
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5931065417947562e-05,
+      "loss": 0.7808,
+      "step": 1693
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592614909419726e-05,
+      "loss": 0.8197,
+      "step": 1694
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5921230561767248e-05,
+      "loss": 0.8161,
+      "step": 1695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5916309822490666e-05,
+      "loss": 0.7953,
+      "step": 1696
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5911386878201476e-05,
+      "loss": 0.8172,
+      "step": 1697
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590646173073446e-05,
+      "loss": 0.293,
+      "step": 1698
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590153438192523e-05,
+      "loss": 0.8175,
+      "step": 1699
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5896604833610206e-05,
+      "loss": 0.8218,
+      "step": 1700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589167308762664e-05,
+      "loss": 0.8124,
+      "step": 1701
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5886739145812593e-05,
+      "loss": 0.8475,
+      "step": 1702
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5881803010006948e-05,
+      "loss": 0.8422,
+      "step": 1703
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5876864682049407e-05,
+      "loss": 0.8685,
+      "step": 1704
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5871924163780487e-05,
+      "loss": 0.8419,
+      "step": 1705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5866981457041526e-05,
+      "loss": 0.8064,
+      "step": 1706
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5862036563674673e-05,
+      "loss": 0.8573,
+      "step": 1707
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585708948552289e-05,
+      "loss": 0.8394,
+      "step": 1708
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5852140224429958e-05,
+      "loss": 0.2748,
+      "step": 1709
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5847188782240473e-05,
+      "loss": 0.7779,
+      "step": 1710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5842235160799835e-05,
+      "loss": 0.8179,
+      "step": 1711
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583727936195427e-05,
+      "loss": 0.7945,
+      "step": 1712
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58323213875508e-05,
+      "loss": 0.7819,
+      "step": 1713
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582736123943726e-05,
+      "loss": 0.8405,
+      "step": 1714
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5822398919462315e-05,
+      "loss": 0.8335,
+      "step": 1715
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581743442947542e-05,
+      "loss": 0.8535,
+      "step": 1716
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5812467771326838e-05,
+      "loss": 0.8588,
+      "step": 1717
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5807498946867657e-05,
+      "loss": 0.8625,
+      "step": 1718
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5802527957949747e-05,
+      "loss": 0.8555,
+      "step": 1719
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579755480642581e-05,
+      "loss": 0.79,
+      "step": 1720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5792579494149336e-05,
+      "loss": 0.7653,
+      "step": 1721
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578760202297463e-05,
+      "loss": 0.8188,
+      "step": 1722
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.57826223947568e-05,
+      "loss": 0.7734,
+      "step": 1723
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5777640611351755e-05,
+      "loss": 0.8175,
+      "step": 1724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5772656674616208e-05,
+      "loss": 0.2707,
+      "step": 1725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576767058640768e-05,
+      "loss": 0.7907,
+      "step": 1726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5762682348584485e-05,
+      "loss": 0.7828,
+      "step": 1727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5757691963005746e-05,
+      "loss": 0.8642,
+      "step": 1728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575269943153138e-05,
+      "loss": 0.8392,
+      "step": 1729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5747704756022106e-05,
+      "loss": 0.8548,
+      "step": 1730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5742707938339447e-05,
+      "loss": 0.9063,
+      "step": 1731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573770898034572e-05,
+      "loss": 0.3049,
+      "step": 1732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5732707883904032e-05,
+      "loss": 0.8502,
+      "step": 1733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5727704650878307e-05,
+      "loss": 0.832,
+      "step": 1734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5722699283133243e-05,
+      "loss": 0.8604,
+      "step": 1735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571769178253435e-05,
+      "loss": 0.8203,
+      "step": 1736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5712682150947926e-05,
+      "loss": 0.8731,
+      "step": 1737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5707670390241062e-05,
+      "loss": 0.2896,
+      "step": 1738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5702656502281645e-05,
+      "loss": 0.295,
+      "step": 1739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5697640488938354e-05,
+      "loss": 0.8133,
+      "step": 1740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5692622352080662e-05,
+      "loss": 0.8296,
+      "step": 1741
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5687602093578832e-05,
+      "loss": 0.878,
+      "step": 1742
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568257971530392e-05,
+      "loss": 0.8891,
+      "step": 1743
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567755521912776e-05,
+      "loss": 0.8728,
+      "step": 1744
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5672528606922996e-05,
+      "loss": 0.8255,
+      "step": 1745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5667499880563044e-05,
+      "loss": 0.8868,
+      "step": 1746
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5662469041922113e-05,
+      "loss": 0.8124,
+      "step": 1747
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565743609287521e-05,
+      "loss": 0.76,
+      "step": 1748
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5652401035298105e-05,
+      "loss": 0.7757,
+      "step": 1749
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5647363871067372e-05,
+      "loss": 0.787,
+      "step": 1750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564232460206037e-05,
+      "loss": 0.7668,
+      "step": 1751
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563728323015523e-05,
+      "loss": 0.8573,
+      "step": 1752
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5632239757230883e-05,
+      "loss": 0.8342,
+      "step": 1753
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562719418516703e-05,
+      "loss": 0.7965,
+      "step": 1754
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562214651584416e-05,
+      "loss": 0.8552,
+      "step": 1755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5617096751143538e-05,
+      "loss": 0.8751,
+      "step": 1756
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561204489294722e-05,
+      "loss": 0.798,
+      "step": 1757
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560699094313804e-05,
+      "loss": 0.7951,
+      "step": 1758
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5601934903599603e-05,
+      "loss": 0.812,
+      "step": 1759
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5596876776216298e-05,
+      "loss": 0.7979,
+      "step": 1760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5591816562873298e-05,
+      "loss": 0.7874,
+      "step": 1761
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558675426545654e-05,
+      "loss": 0.8718,
+      "step": 1762
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5581689885852752e-05,
+      "loss": 0.828,
+      "step": 1763
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557662342594943e-05,
+      "loss": 0.8625,
+      "step": 1764
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5571554887634844e-05,
+      "loss": 0.8154,
+      "step": 1765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5566484272798047e-05,
+      "loss": 0.865,
+      "step": 1766
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5561411583328853e-05,
+      "loss": 0.7692,
+      "step": 1767
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5556336821117863e-05,
+      "loss": 0.832,
+      "step": 1768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5551259988056444e-05,
+      "loss": 0.8996,
+      "step": 1769
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5546181086036733e-05,
+      "loss": 0.8393,
+      "step": 1770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554110011695164e-05,
+      "loss": 0.802,
+      "step": 1771
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5536017082694846e-05,
+      "loss": 0.8367,
+      "step": 1772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55309319851608e-05,
+      "loss": 0.794,
+      "step": 1773
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5525844826244726e-05,
+      "loss": 0.8209,
+      "step": 1774
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552075560784261e-05,
+      "loss": 0.8646,
+      "step": 1775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55156643318512e-05,
+      "loss": 0.7963,
+      "step": 1776
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5510571000168025e-05,
+      "loss": 0.7955,
+      "step": 1777
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5505475614691374e-05,
+      "loss": 0.8277,
+      "step": 1778
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5500378177320294e-05,
+      "loss": 0.8074,
+      "step": 1779
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5495278689954608e-05,
+      "loss": 0.829,
+      "step": 1780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54901771544949e-05,
+      "loss": 0.8184,
+      "step": 1781
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5485073572842512e-05,
+      "loss": 0.8275,
+      "step": 1782
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5479967946899552e-05,
+      "loss": 0.8235,
+      "step": 1783
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547486027856889e-05,
+      "loss": 0.8058,
+      "step": 1784
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5469750569754166e-05,
+      "loss": 0.8286,
+      "step": 1785
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546463882235976e-05,
+      "loss": 0.7929,
+      "step": 1786
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5459525038290827e-05,
+      "loss": 0.8872,
+      "step": 1787
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5454409219453284e-05,
+      "loss": 0.7907,
+      "step": 1788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544929136775379e-05,
+      "loss": 0.821,
+      "step": 1789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5444171485099785e-05,
+      "loss": 0.8866,
+      "step": 1790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5439049573399438e-05,
+      "loss": 0.8822,
+      "step": 1791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5433925634561698e-05,
+      "loss": 0.3005,
+      "step": 1792
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5428799670496256e-05,
+      "loss": 0.8836,
+      "step": 1793
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542367168311357e-05,
+      "loss": 0.8386,
+      "step": 1794
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541854167432483e-05,
+      "loss": 0.7969,
+      "step": 1795
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541340964604201e-05,
+      "loss": 0.8421,
+      "step": 1796
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540827560017781e-05,
+      "loss": 0.8227,
+      "step": 1797
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5403139538645695e-05,
+      "loss": 0.7726,
+      "step": 1798
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539800146335988e-05,
+      "loss": 0.7997,
+      "step": 1799
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5392861376235325e-05,
+      "loss": 0.8811,
+      "step": 1800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538771927918775e-05,
+      "loss": 0.8179,
+      "step": 1801
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5382575174133608e-05,
+      "loss": 0.8266,
+      "step": 1802
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5377429062990122e-05,
+      "loss": 0.7745,
+      "step": 1803
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5372280947675244e-05,
+      "loss": 0.8076,
+      "step": 1804
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5367130830107683e-05,
+      "loss": 0.8658,
+      "step": 1805
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536197871220689e-05,
+      "loss": 0.8717,
+      "step": 1806
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5356824595893063e-05,
+      "loss": 0.8329,
+      "step": 1807
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5351668483087145e-05,
+      "loss": 0.856,
+      "step": 1808
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5346510375710822e-05,
+      "loss": 0.8085,
+      "step": 1809
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5341350275686526e-05,
+      "loss": 0.8137,
+      "step": 1810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5336188184937424e-05,
+      "loss": 0.8694,
+      "step": 1811
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5331024105387434e-05,
+      "loss": 0.8994,
+      "step": 1812
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532585803896122e-05,
+      "loss": 0.3195,
+      "step": 1813
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5320689987584164e-05,
+      "loss": 0.8819,
+      "step": 1814
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5315519953182413e-05,
+      "loss": 0.8961,
+      "step": 1815
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5310347937682837e-05,
+      "loss": 0.8416,
+      "step": 1816
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530517394301305e-05,
+      "loss": 0.8521,
+      "step": 1817
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5299997971101405e-05,
+      "loss": 0.7642,
+      "step": 1818
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5294820023876987e-05,
+      "loss": 0.8114,
+      "step": 1819
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5289640103269626e-05,
+      "loss": 0.8907,
+      "step": 1820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528445821120988e-05,
+      "loss": 0.8553,
+      "step": 1821
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5279274349629038e-05,
+      "loss": 0.2625,
+      "step": 1822
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5274088520459134e-05,
+      "loss": 0.8357,
+      "step": 1823
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526890072563293e-05,
+      "loss": 0.8171,
+      "step": 1824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5263710967083918e-05,
+      "loss": 0.2724,
+      "step": 1825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5258519246746329e-05,
+      "loss": 0.8625,
+      "step": 1826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5253325566555117e-05,
+      "loss": 0.7787,
+      "step": 1827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5248129928445968e-05,
+      "loss": 0.8199,
+      "step": 1828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5242932334355304e-05,
+      "loss": 0.7572,
+      "step": 1829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.523773278622027e-05,
+      "loss": 0.8635,
+      "step": 1830
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5232531285978742e-05,
+      "loss": 0.8688,
+      "step": 1831
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5227327835569321e-05,
+      "loss": 0.7572,
+      "step": 1832
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5222122436931337e-05,
+      "loss": 0.8292,
+      "step": 1833
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5216915092004847e-05,
+      "loss": 0.8166,
+      "step": 1834
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5211705802730628e-05,
+      "loss": 0.8208,
+      "step": 1835
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5206494571050192e-05,
+      "loss": 0.8696,
+      "step": 1836
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5201281398905763e-05,
+      "loss": 0.8125,
+      "step": 1837
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5196066288240298e-05,
+      "loss": 0.8143,
+      "step": 1838
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5190849240997465e-05,
+      "loss": 0.8116,
+      "step": 1839
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5185630259121666e-05,
+      "loss": 0.8066,
+      "step": 1840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5180409344558023e-05,
+      "loss": 0.2834,
+      "step": 1841
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5175186499252368e-05,
+      "loss": 0.7686,
+      "step": 1842
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5169961725151263e-05,
+      "loss": 0.8146,
+      "step": 1843
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5164735024201979e-05,
+      "loss": 0.7735,
+      "step": 1844
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515950639835252e-05,
+      "loss": 0.8583,
+      "step": 1845
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5154275849551588e-05,
+      "loss": 0.8534,
+      "step": 1846
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.514904337974862e-05,
+      "loss": 0.8263,
+      "step": 1847
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5143808990893756e-05,
+      "loss": 0.8156,
+      "step": 1848
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5138572684937858e-05,
+      "loss": 0.8069,
+      "step": 1849
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5133334463832504e-05,
+      "loss": 0.7995,
+      "step": 1850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5128094329529976e-05,
+      "loss": 0.8286,
+      "step": 1851
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5122852283983282e-05,
+      "loss": 0.8504,
+      "step": 1852
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5117608329146132e-05,
+      "loss": 0.7989,
+      "step": 1853
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5112362466972949e-05,
+      "loss": 0.835,
+      "step": 1854
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5107114699418879e-05,
+      "loss": 0.8533,
+      "step": 1855
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5101865028439756e-05,
+      "loss": 0.2737,
+      "step": 1856
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5096613455992146e-05,
+      "loss": 0.8253,
+      "step": 1857
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5091359984033308e-05,
+      "loss": 0.8636,
+      "step": 1858
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5086104614521218e-05,
+      "loss": 0.8226,
+      "step": 1859
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5080847349414552e-05,
+      "loss": 0.8595,
+      "step": 1860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5075588190672696e-05,
+      "loss": 0.8581,
+      "step": 1861
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5070327140255745e-05,
+      "loss": 0.2915,
+      "step": 1862
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5065064200124496e-05,
+      "loss": 0.8146,
+      "step": 1863
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5059799372240452e-05,
+      "loss": 0.7627,
+      "step": 1864
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.505453265856581e-05,
+      "loss": 0.7938,
+      "step": 1865
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5049264061063484e-05,
+      "loss": 0.8376,
+      "step": 1866
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5043993581697082e-05,
+      "loss": 0.8426,
+      "step": 1867
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503872122243092e-05,
+      "loss": 0.8322,
+      "step": 1868
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503344698523e-05,
+      "loss": 0.774,
+      "step": 1869
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.502817087206004e-05,
+      "loss": 0.7944,
+      "step": 1870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5022892884887457e-05,
+      "loss": 0.8164,
+      "step": 1871
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5017613025679344e-05,
+      "loss": 0.8289,
+      "step": 1872
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5012331296403523e-05,
+      "loss": 0.8324,
+      "step": 1873
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5007047699028489e-05,
+      "loss": 0.8728,
+      "step": 1874
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5001762235523452e-05,
+      "loss": 0.7785,
+      "step": 1875
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4996474907858297e-05,
+      "loss": 0.8016,
+      "step": 1876
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4991185718003617e-05,
+      "loss": 0.8794,
+      "step": 1877
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4985894667930702e-05,
+      "loss": 0.7994,
+      "step": 1878
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4980601759611527e-05,
+      "loss": 0.8001,
+      "step": 1879
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497530699501876e-05,
+      "loss": 0.8245,
+      "step": 1880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4970010376125766e-05,
+      "loss": 0.845,
+      "step": 1881
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.49647119049066e-05,
+      "loss": 0.7554,
+      "step": 1882
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4959411583335998e-05,
+      "loss": 0.8095,
+      "step": 1883
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.49541094133894e-05,
+      "loss": 0.8327,
+      "step": 1884
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4948805397042928e-05,
+      "loss": 0.8217,
+      "step": 1885
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4943499536273395e-05,
+      "loss": 0.8156,
+      "step": 1886
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4938191833058294e-05,
+      "loss": 0.8505,
+      "step": 1887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4932882289375808e-05,
+      "loss": 0.8792,
+      "step": 1888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4927570907204811e-05,
+      "loss": 0.8283,
+      "step": 1889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4922257688524856e-05,
+      "loss": 0.8274,
+      "step": 1890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4916942635316189e-05,
+      "loss": 0.814,
+      "step": 1891
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4911625749559731e-05,
+      "loss": 0.8412,
+      "step": 1892
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4906307033237084e-05,
+      "loss": 0.8396,
+      "step": 1893
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4900986488330542e-05,
+      "loss": 0.85,
+      "step": 1894
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4895664116823074e-05,
+      "loss": 0.8621,
+      "step": 1895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4890339920698334e-05,
+      "loss": 0.8004,
+      "step": 1896
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.488501390194065e-05,
+      "loss": 0.8017,
+      "step": 1897
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4879686062535034e-05,
+      "loss": 0.7733,
+      "step": 1898
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4874356404467176e-05,
+      "loss": 0.8208,
+      "step": 1899
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4869024929723441e-05,
+      "loss": 0.8343,
+      "step": 1900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4863691640290879e-05,
+      "loss": 0.8573,
+      "step": 1901
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4858356538157202e-05,
+      "loss": 0.8112,
+      "step": 1902
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4853019625310813e-05,
+      "loss": 0.8136,
+      "step": 1903
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4847680903740777e-05,
+      "loss": 0.869,
+      "step": 1904
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4842340375436843e-05,
+      "loss": 0.7971,
+      "step": 1905
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4836998042389432e-05,
+      "loss": 0.6835,
+      "step": 1906
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4831653906589632e-05,
+      "loss": 0.8591,
+      "step": 1907
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4826307970029204e-05,
+      "loss": 0.8376,
+      "step": 1908
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4820960234700583e-05,
+      "loss": 0.8826,
+      "step": 1909
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4815610702596877e-05,
+      "loss": 0.8364,
+      "step": 1910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4810259375711856e-05,
+      "loss": 0.8033,
+      "step": 1911
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4804906256039961e-05,
+      "loss": 0.8441,
+      "step": 1912
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4799551345576316e-05,
+      "loss": 0.8376,
+      "step": 1913
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4794194646316684e-05,
+      "loss": 0.8026,
+      "step": 1914
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.478883616025752e-05,
+      "loss": 0.7991,
+      "step": 1915
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4783475889395931e-05,
+      "loss": 0.8239,
+      "step": 1916
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4778113835729695e-05,
+      "loss": 0.8116,
+      "step": 1917
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4772750001257255e-05,
+      "loss": 0.7566,
+      "step": 1918
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.476738438797771e-05,
+      "loss": 0.8037,
+      "step": 1919
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4762016997890834e-05,
+      "loss": 0.7852,
+      "step": 1920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4756647832997053e-05,
+      "loss": 0.8245,
+      "step": 1921
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4751276895297459e-05,
+      "loss": 0.7144,
+      "step": 1922
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4745904186793804e-05,
+      "loss": 0.7876,
+      "step": 1923
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4740529709488507e-05,
+      "loss": 0.8215,
+      "step": 1924
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4735153465384623e-05,
+      "loss": 0.8729,
+      "step": 1925
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4729775456485897e-05,
+      "loss": 0.8057,
+      "step": 1926
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.472439568479671e-05,
+      "loss": 0.8306,
+      "step": 1927
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4719014152322109e-05,
+      "loss": 0.8165,
+      "step": 1928
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4713630861067792e-05,
+      "loss": 0.8227,
+      "step": 1929
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4708245813040116e-05,
+      "loss": 0.797,
+      "step": 1930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4702859010246096e-05,
+      "loss": 0.8498,
+      "step": 1931
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.469747045469339e-05,
+      "loss": 0.8135,
+      "step": 1932
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4692080148390318e-05,
+      "loss": 0.8529,
+      "step": 1933
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4686688093345855e-05,
+      "loss": 0.8406,
+      "step": 1934
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4681294291569623e-05,
+      "loss": 0.8114,
+      "step": 1935
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.467589874507189e-05,
+      "loss": 0.8617,
+      "step": 1936
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4670501455863583e-05,
+      "loss": 0.7531,
+      "step": 1937
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.466510242595628e-05,
+      "loss": 0.8069,
+      "step": 1938
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4659701657362193e-05,
+      "loss": 0.7935,
+      "step": 1939
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4654299152094202e-05,
+      "loss": 0.812,
+      "step": 1940
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4648894912165818e-05,
+      "loss": 0.8476,
+      "step": 1941
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4643488939591206e-05,
+      "loss": 0.8222,
+      "step": 1942
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4638081236385177e-05,
+      "loss": 0.8472,
+      "step": 1943
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4632671804563185e-05,
+      "loss": 0.8316,
+      "step": 1944
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4627260646141329e-05,
+      "loss": 0.7524,
+      "step": 1945
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4621847763136351e-05,
+      "loss": 0.2725,
+      "step": 1946
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4616433157565638e-05,
+      "loss": 0.8117,
+      "step": 1947
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.461101683144721e-05,
+      "loss": 0.8103,
+      "step": 1948
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4605598786799744e-05,
+      "loss": 0.7936,
+      "step": 1949
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4600179025642547e-05,
+      "loss": 0.8382,
+      "step": 1950
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4594757549995566e-05,
+      "loss": 0.8444,
+      "step": 1951
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4589334361879388e-05,
+      "loss": 0.7465,
+      "step": 1952
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4583909463315236e-05,
+      "loss": 0.8115,
+      "step": 1953
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.457848285632498e-05,
+      "loss": 0.8439,
+      "step": 1954
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4573054542931116e-05,
+      "loss": 0.807,
+      "step": 1955
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4567624525156783e-05,
+      "loss": 0.7986,
+      "step": 1956
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4562192805025748e-05,
+      "loss": 0.869,
+      "step": 1957
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4556759384562418e-05,
+      "loss": 0.8526,
+      "step": 1958
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4551324265791834e-05,
+      "loss": 0.8389,
+      "step": 1959
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4545887450739666e-05,
+      "loss": 0.7322,
+      "step": 1960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4540448941432223e-05,
+      "loss": 0.8574,
+      "step": 1961
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4535008739896436e-05,
+      "loss": 0.8055,
+      "step": 1962
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4529566848159875e-05,
+      "loss": 0.8507,
+      "step": 1963
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4524123268250737e-05,
+      "loss": 0.8281,
+      "step": 1964
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4518678002197844e-05,
+      "loss": 0.8547,
+      "step": 1965
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4513231052030654e-05,
+      "loss": 0.8115,
+      "step": 1966
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4507782419779248e-05,
+      "loss": 0.7817,
+      "step": 1967
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4502332107474335e-05,
+      "loss": 0.8357,
+      "step": 1968
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4496880117147247e-05,
+      "loss": 0.8466,
+      "step": 1969
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4491426450829946e-05,
+      "loss": 0.8057,
+      "step": 1970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4485971110555022e-05,
+      "loss": 0.7665,
+      "step": 1971
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4480514098355678e-05,
+      "loss": 0.8321,
+      "step": 1972
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4475055416265751e-05,
+      "loss": 0.7433,
+      "step": 1973
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.446959506631969e-05,
+      "loss": 0.8987,
+      "step": 1974
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4464133050552574e-05,
+      "loss": 0.8135,
+      "step": 1975
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4458669371000098e-05,
+      "loss": 0.8048,
+      "step": 1976
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.445320402969858e-05,
+      "loss": 0.7908,
+      "step": 1977
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.444773702868496e-05,
+      "loss": 0.8663,
+      "step": 1978
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.444226836999679e-05,
+      "loss": 0.7915,
+      "step": 1979
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4436798055672242e-05,
+      "loss": 0.7628,
+      "step": 1980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4431326087750104e-05,
+      "loss": 0.7999,
+      "step": 1981
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4425852468269785e-05,
+      "loss": 0.8234,
+      "step": 1982
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4420377199271305e-05,
+      "loss": 0.7944,
+      "step": 1983
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4414900282795304e-05,
+      "loss": 0.8031,
+      "step": 1984
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.440942172088303e-05,
+      "loss": 0.8361,
+      "step": 1985
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4403941515576344e-05,
+      "loss": 0.2869,
+      "step": 1986
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4398459668917726e-05,
+      "loss": 0.76,
+      "step": 1987
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4392976182950262e-05,
+      "loss": 0.8365,
+      "step": 1988
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4387491059717653e-05,
+      "loss": 0.8215,
+      "step": 1989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4382004301264205e-05,
+      "loss": 0.8139,
+      "step": 1990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4376515909634838e-05,
+      "loss": 0.808,
+      "step": 1991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.437102588687508e-05,
+      "loss": 0.7729,
+      "step": 1992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4365534235031065e-05,
+      "loss": 0.8484,
+      "step": 1993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4360040956149535e-05,
+      "loss": 0.2728,
+      "step": 1994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4354546052277842e-05,
+      "loss": 0.8476,
+      "step": 1995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4349049525463936e-05,
+      "loss": 0.838,
+      "step": 1996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4343551377756378e-05,
+      "loss": 0.7815,
+      "step": 1997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433805161120433e-05,
+      "loss": 0.8297,
+      "step": 1998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433255022785756e-05,
+      "loss": 0.8307,
+      "step": 1999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4327047229766437e-05,
+      "loss": 0.7756,
+      "step": 2000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4321542618981937e-05,
+      "loss": 0.8027,
+      "step": 2001
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.431603639755562e-05,
+      "loss": 0.8526,
+      "step": 2002
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.431052856753967e-05,
+      "loss": 0.8635,
+      "step": 2003
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4305019130986852e-05,
+      "loss": 0.8281,
+      "step": 2004
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.429950808995054e-05,
+      "loss": 0.854,
+      "step": 2005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4293995446484706e-05,
+      "loss": 0.8014,
+      "step": 2006
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4288481202643905e-05,
+      "loss": 0.8708,
+      "step": 2007
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.428296536048331e-05,
+      "loss": 0.8469,
+      "step": 2008
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4277447922058674e-05,
+      "loss": 0.7517,
+      "step": 2009
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4271928889426351e-05,
+      "loss": 0.827,
+      "step": 2010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4266408264643289e-05,
+      "loss": 0.8592,
+      "step": 2011
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4260886049767029e-05,
+      "loss": 0.8014,
+      "step": 2012
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42553622468557e-05,
+      "loss": 0.8185,
+      "step": 2013
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4249836857968031e-05,
+      "loss": 0.8494,
+      "step": 2014
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424430988516334e-05,
+      "loss": 0.8293,
+      "step": 2015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4238781330501527e-05,
+      "loss": 0.7829,
+      "step": 2016
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4233251196043095e-05,
+      "loss": 0.8297,
+      "step": 2017
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.422771948384912e-05,
+      "loss": 0.7968,
+      "step": 2018
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4222186195981285e-05,
+      "loss": 0.7785,
+      "step": 2019
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.421665133450184e-05,
+      "loss": 0.8376,
+      "step": 2020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4211114901473639e-05,
+      "loss": 0.8008,
+      "step": 2021
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4205576898960117e-05,
+      "loss": 0.8784,
+      "step": 2022
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.420003732902528e-05,
+      "loss": 0.7747,
+      "step": 2023
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.419449619373374e-05,
+      "loss": 0.787,
+      "step": 2024
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4188953495150674e-05,
+      "loss": 0.752,
+      "step": 2025
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4183409235341853e-05,
+      "loss": 0.8231,
+      "step": 2026
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.417786341637363e-05,
+      "loss": 0.8113,
+      "step": 2027
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4172316040312927e-05,
+      "loss": 0.8062,
+      "step": 2028
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4166767109227264e-05,
+      "loss": 0.8424,
+      "step": 2029
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4161216625184723e-05,
+      "loss": 0.8246,
+      "step": 2030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.415566459025398e-05,
+      "loss": 0.8447,
+      "step": 2031
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4150111006504275e-05,
+      "loss": 0.8146,
+      "step": 2032
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4144555876005438e-05,
+      "loss": 0.8068,
+      "step": 2033
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4138999200827871e-05,
+      "loss": 0.8219,
+      "step": 2034
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4133440983042544e-05,
+      "loss": 0.8038,
+      "step": 2035
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4127881224721014e-05,
+      "loss": 0.814,
+      "step": 2036
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4122319927935402e-05,
+      "loss": 0.8066,
+      "step": 2037
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4116757094758413e-05,
+      "loss": 0.8429,
+      "step": 2038
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4111192727263316e-05,
+      "loss": 0.8085,
+      "step": 2039
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4105626827523953e-05,
+      "loss": 0.8282,
+      "step": 2040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4100059397614737e-05,
+      "loss": 0.8418,
+      "step": 2041
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4094490439610655e-05,
+      "loss": 0.8342,
+      "step": 2042
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4088919955587264e-05,
+      "loss": 0.8606,
+      "step": 2043
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4083347947620684e-05,
+      "loss": 0.848,
+      "step": 2044
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4077774417787608e-05,
+      "loss": 0.2764,
+      "step": 2045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.407219936816529e-05,
+      "loss": 0.8196,
+      "step": 2046
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4066622800831563e-05,
+      "loss": 0.7989,
+      "step": 2047
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4061044717864805e-05,
+      "loss": 0.8317,
+      "step": 2048
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4055465121343984e-05,
+      "loss": 0.8188,
+      "step": 2049
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4049884013348611e-05,
+      "loss": 0.7536,
+      "step": 2050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.404430139595877e-05,
+      "loss": 0.7726,
+      "step": 2051
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4038717271255112e-05,
+      "loss": 0.8566,
+      "step": 2052
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4033131641318835e-05,
+      "loss": 0.8329,
+      "step": 2053
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4027544508231714e-05,
+      "loss": 0.7636,
+      "step": 2054
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4021955874076074e-05,
+      "loss": 0.7943,
+      "step": 2055
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4016365740934802e-05,
+      "loss": 0.8329,
+      "step": 2056
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.401077411089135e-05,
+      "loss": 0.8571,
+      "step": 2057
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4005180986029714e-05,
+      "loss": 0.8217,
+      "step": 2058
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3999586368434462e-05,
+      "loss": 0.7832,
+      "step": 2059
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3993990260190708e-05,
+      "loss": 0.8648,
+      "step": 2060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3988392663384127e-05,
+      "loss": 0.8096,
+      "step": 2061
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3982793580100945e-05,
+      "loss": 0.8245,
+      "step": 2062
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3977193012427943e-05,
+      "loss": 0.823,
+      "step": 2063
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3971590962452463e-05,
+      "loss": 0.8367,
+      "step": 2064
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.396598743226239e-05,
+      "loss": 0.8386,
+      "step": 2065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.396038242394616e-05,
+      "loss": 0.821,
+      "step": 2066
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3954775939592761e-05,
+      "loss": 0.8068,
+      "step": 2067
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3949167981291742e-05,
+      "loss": 0.7923,
+      "step": 2068
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3943558551133186e-05,
+      "loss": 0.8201,
+      "step": 2069
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3937947651207735e-05,
+      "loss": 0.8154,
+      "step": 2070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3932335283606578e-05,
+      "loss": 0.7935,
+      "step": 2071
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392672145042144e-05,
+      "loss": 0.8715,
+      "step": 2072
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3921106153744606e-05,
+      "loss": 0.864,
+      "step": 2073
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3915489395668896e-05,
+      "loss": 0.7843,
+      "step": 2074
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3909871178287685e-05,
+      "loss": 0.7512,
+      "step": 2075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3904251503694884e-05,
+      "loss": 0.7943,
+      "step": 2076
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3898630373984951e-05,
+      "loss": 0.8041,
+      "step": 2077
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3893007791252881e-05,
+      "loss": 0.8082,
+      "step": 2078
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3887383757594217e-05,
+      "loss": 0.8336,
+      "step": 2079
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3881758275105042e-05,
+      "loss": 0.7897,
+      "step": 2080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3876131345881971e-05,
+      "loss": 0.8613,
+      "step": 2081
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3870502972022175e-05,
+      "loss": 0.8226,
+      "step": 2082
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.386487315562334e-05,
+      "loss": 0.8142,
+      "step": 2083
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.385924189878371e-05,
+      "loss": 0.7732,
+      "step": 2084
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3853609203602053e-05,
+      "loss": 0.8259,
+      "step": 2085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3847975072177683e-05,
+      "loss": 0.8058,
+      "step": 2086
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3842339506610448e-05,
+      "loss": 0.8216,
+      "step": 2087
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3836702509000719e-05,
+      "loss": 0.8665,
+      "step": 2088
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.383106408144941e-05,
+      "loss": 0.8084,
+      "step": 2089
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3825424226057968e-05,
+      "loss": 0.8459,
+      "step": 2090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3819782944928375e-05,
+      "loss": 0.8344,
+      "step": 2091
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3814140240163134e-05,
+      "loss": 0.3025,
+      "step": 2092
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3808496113865288e-05,
+      "loss": 0.824,
+      "step": 2093
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3802850568138408e-05,
+      "loss": 0.8217,
+      "step": 2094
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3797203605086589e-05,
+      "loss": 0.8225,
+      "step": 2095
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.379155522681446e-05,
+      "loss": 0.7866,
+      "step": 2096
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3785905435427173e-05,
+      "loss": 0.7682,
+      "step": 2097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3780254233030414e-05,
+      "loss": 0.814,
+      "step": 2098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3774601621730386e-05,
+      "loss": 0.8804,
+      "step": 2099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3768947603633822e-05,
+      "loss": 0.8658,
+      "step": 2100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3763292180847977e-05,
+      "loss": 0.8266,
+      "step": 2101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3757635355480632e-05,
+      "loss": 0.3042,
+      "step": 2102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375197712964009e-05,
+      "loss": 0.8162,
+      "step": 2103
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3746317505435175e-05,
+      "loss": 0.8174,
+      "step": 2104
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3740656484975236e-05,
+      "loss": 0.7705,
+      "step": 2105
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3734994070370132e-05,
+      "loss": 0.7901,
+      "step": 2106
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3729330263730252e-05,
+      "loss": 0.8573,
+      "step": 2107
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3723665067166506e-05,
+      "loss": 0.8346,
+      "step": 2108
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3717998482790307e-05,
+      "loss": 0.8008,
+      "step": 2109
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3712330512713603e-05,
+      "loss": 0.8466,
+      "step": 2110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3706661159048842e-05,
+      "loss": 0.8674,
+      "step": 2111
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3700990423909005e-05,
+      "loss": 0.7805,
+      "step": 2112
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.369531830940757e-05,
+      "loss": 0.2763,
+      "step": 2113
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3689644817658545e-05,
+      "loss": 0.8295,
+      "step": 2114
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3683969950776441e-05,
+      "loss": 0.8179,
+      "step": 2115
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3678293710876285e-05,
+      "loss": 0.808,
+      "step": 2116
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3672616100073615e-05,
+      "loss": 0.8245,
+      "step": 2117
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.366693712048448e-05,
+      "loss": 0.8253,
+      "step": 2118
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3661256774225441e-05,
+      "loss": 0.8495,
+      "step": 2119
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3655575063413567e-05,
+      "loss": 0.8324,
+      "step": 2120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3649891990166431e-05,
+      "loss": 0.7972,
+      "step": 2121
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3644207556602124e-05,
+      "loss": 0.8202,
+      "step": 2122
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3638521764839234e-05,
+      "loss": 0.8883,
+      "step": 2123
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3632834616996863e-05,
+      "loss": 0.8484,
+      "step": 2124
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3627146115194613e-05,
+      "loss": 0.7979,
+      "step": 2125
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3621456261552593e-05,
+      "loss": 0.8325,
+      "step": 2126
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3615765058191413e-05,
+      "loss": 0.808,
+      "step": 2127
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3610072507232186e-05,
+      "loss": 0.8381,
+      "step": 2128
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.360437861079654e-05,
+      "loss": 0.8202,
+      "step": 2129
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3598683371006583e-05,
+      "loss": 0.7741,
+      "step": 2130
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3592986789984943e-05,
+      "loss": 0.8128,
+      "step": 2131
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358728886985473e-05,
+      "loss": 0.8113,
+      "step": 2132
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3581589612739574e-05,
+      "loss": 0.8441,
+      "step": 2133
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.357588902076358e-05,
+      "loss": 0.893,
+      "step": 2134
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.357018709605137e-05,
+      "loss": 0.7684,
+      "step": 2135
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3564483840728058e-05,
+      "loss": 0.8715,
+      "step": 2136
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3558779256919244e-05,
+      "loss": 0.7663,
+      "step": 2137
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3553073346751033e-05,
+      "loss": 0.8828,
+      "step": 2138
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.354736611235002e-05,
+      "loss": 0.8395,
+      "step": 2139
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.35416575558433e-05,
+      "loss": 0.8257,
+      "step": 2140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3535947679358446e-05,
+      "loss": 0.7337,
+      "step": 2141
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3530236485023544e-05,
+      "loss": 0.7953,
+      "step": 2142
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3524523974967154e-05,
+      "loss": 0.7132,
+      "step": 2143
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.351881015131833e-05,
+      "loss": 0.8078,
+      "step": 2144
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3513095016206626e-05,
+      "loss": 0.8104,
+      "step": 2145
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3507378571762068e-05,
+      "loss": 0.7665,
+      "step": 2146
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3501660820115186e-05,
+      "loss": 0.7748,
+      "step": 2147
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3495941763396985e-05,
+      "loss": 0.7889,
+      "step": 2148
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3490221403738963e-05,
+      "loss": 0.8657,
+      "step": 2149
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3484499743273099e-05,
+      "loss": 0.8069,
+      "step": 2150
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3478776784131862e-05,
+      "loss": 0.8052,
+      "step": 2151
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3473052528448203e-05,
+      "loss": 0.7851,
+      "step": 2152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.346732697835556e-05,
+      "loss": 0.7818,
+      "step": 2153
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.346160013598784e-05,
+      "loss": 0.7937,
+      "step": 2154
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3455872003479446e-05,
+      "loss": 0.7874,
+      "step": 2155
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3450142582965259e-05,
+      "loss": 0.8212,
+      "step": 2156
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3444411876580632e-05,
+      "loss": 0.8014,
+      "step": 2157
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.343867988646141e-05,
+      "loss": 0.7565,
+      "step": 2158
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3432946614743906e-05,
+      "loss": 0.8655,
+      "step": 2159
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3427212063564912e-05,
+      "loss": 0.7876,
+      "step": 2160
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3421476235061705e-05,
+      "loss": 0.7782,
+      "step": 2161
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3415739131372023e-05,
+      "loss": 0.7838,
+      "step": 2162
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3410000754634101e-05,
+      "loss": 0.7972,
+      "step": 2163
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3404261106986624e-05,
+      "loss": 0.8277,
+      "step": 2164
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3398520190568766e-05,
+      "loss": 0.7758,
+      "step": 2165
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3392778007520175e-05,
+      "loss": 0.7791,
+      "step": 2166
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3387034559980961e-05,
+      "loss": 0.7763,
+      "step": 2167
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3381289850091712e-05,
+      "loss": 0.7573,
+      "step": 2168
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3375543879993486e-05,
+      "loss": 0.8307,
+      "step": 2169
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3369796651827814e-05,
+      "loss": 0.8078,
+      "step": 2170
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3364048167736683e-05,
+      "loss": 0.8465,
+      "step": 2171
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.335829842986256e-05,
+      "loss": 0.7943,
+      "step": 2172
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3352547440348382e-05,
+      "loss": 0.8164,
+      "step": 2173
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.334679520133754e-05,
+      "loss": 0.8296,
+      "step": 2174
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3341041714973901e-05,
+      "loss": 0.809,
+      "step": 2175
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.333528698340179e-05,
+      "loss": 0.8034,
+      "step": 2176
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3329531008766e-05,
+      "loss": 0.8517,
+      "step": 2177
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3323773793211787e-05,
+      "loss": 0.2772,
+      "step": 2178
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3318015338884868e-05,
+      "loss": 0.7853,
+      "step": 2179
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3312255647931434e-05,
+      "loss": 0.8126,
+      "step": 2180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3306494722498107e-05,
+      "loss": 0.8279,
+      "step": 2181
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3300732564732e-05,
+      "loss": 0.8377,
+      "step": 2182
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3294969176780668e-05,
+      "loss": 0.8671,
+      "step": 2183
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3289204560792133e-05,
+      "loss": 0.8261,
+      "step": 2184
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3283438718914865e-05,
+      "loss": 0.8071,
+      "step": 2185
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3277671653297805e-05,
+      "loss": 0.7722,
+      "step": 2186
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3271903366090342e-05,
+      "loss": 0.8229,
+      "step": 2187
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3266133859442312e-05,
+      "loss": 0.7761,
+      "step": 2188
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3260363135504023e-05,
+      "loss": 0.8759,
+      "step": 2189
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3254591196426221e-05,
+      "loss": 0.8433,
+      "step": 2190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3248818044360115e-05,
+      "loss": 0.7977,
+      "step": 2191
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3243043681457364e-05,
+      "loss": 0.8048,
+      "step": 2192
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3237268109870072e-05,
+      "loss": 0.8159,
+      "step": 2193
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3231491331750804e-05,
+      "loss": 0.8341,
+      "step": 2194
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3225713349252562e-05,
+      "loss": 0.8356,
+      "step": 2195
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3219934164528813e-05,
+      "loss": 0.8533,
+      "step": 2196
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3214153779733457e-05,
+      "loss": 0.2829,
+      "step": 2197
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3208372197020848e-05,
+      "loss": 0.8483,
+      "step": 2198
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3202589418545786e-05,
+      "loss": 0.8343,
+      "step": 2199
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3196805446463517e-05,
+      "loss": 0.8586,
+      "step": 2200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3191020282929729e-05,
+      "loss": 0.7608,
+      "step": 2201
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3185233930100563e-05,
+      "loss": 0.2869,
+      "step": 2202
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3179446390132588e-05,
+      "loss": 0.7672,
+      "step": 2203
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3173657665182828e-05,
+      "loss": 0.8164,
+      "step": 2204
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3167867757408745e-05,
+      "loss": 0.7588,
+      "step": 2205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.316207666896824e-05,
+      "loss": 0.8174,
+      "step": 2206
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.315628440201966e-05,
+      "loss": 0.8783,
+      "step": 2207
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.315049095872178e-05,
+      "loss": 0.8005,
+      "step": 2208
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.314469634123382e-05,
+      "loss": 0.8307,
+      "step": 2209
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3138900551715446e-05,
+      "loss": 0.7891,
+      "step": 2210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3133103592326743e-05,
+      "loss": 0.8877,
+      "step": 2211
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3127305465228251e-05,
+      "loss": 0.7951,
+      "step": 2212
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3121506172580931e-05,
+      "loss": 0.7915,
+      "step": 2213
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3115705716546183e-05,
+      "loss": 0.7841,
+      "step": 2214
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.310990409928584e-05,
+      "loss": 0.8213,
+      "step": 2215
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3104101322962168e-05,
+      "loss": 0.2748,
+      "step": 2216
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3098297389737868e-05,
+      "loss": 0.8203,
+      "step": 2217
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3092492301776072e-05,
+      "loss": 0.8035,
+      "step": 2218
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3086686061240336e-05,
+      "loss": 0.8192,
+      "step": 2219
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.308087867029465e-05,
+      "loss": 0.8041,
+      "step": 2220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3075070131103433e-05,
+      "loss": 0.8532,
+      "step": 2221
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3069260445831534e-05,
+      "loss": 0.8374,
+      "step": 2222
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3063449616644224e-05,
+      "loss": 0.7904,
+      "step": 2223
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3057637645707203e-05,
+      "loss": 0.7646,
+      "step": 2224
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3051824535186595e-05,
+      "loss": 0.8269,
+      "step": 2225
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3046010287248956e-05,
+      "loss": 0.8945,
+      "step": 2226
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3040194904061253e-05,
+      "loss": 0.8342,
+      "step": 2227
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3034378387790887e-05,
+      "loss": 0.7535,
+      "step": 2228
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3028560740605678e-05,
+      "loss": 0.2449,
+      "step": 2229
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3022741964673862e-05,
+      "loss": 0.827,
+      "step": 2230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3016922062164108e-05,
+      "loss": 0.7842,
+      "step": 2231
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.301110103524549e-05,
+      "loss": 0.3054,
+      "step": 2232
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3005278886087518e-05,
+      "loss": 0.816,
+      "step": 2233
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2999455616860099e-05,
+      "loss": 0.8456,
+      "step": 2234
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2993631229733584e-05,
+      "loss": 0.2575,
+      "step": 2235
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2987805726878711e-05,
+      "loss": 0.8419,
+      "step": 2236
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2981979110466655e-05,
+      "loss": 0.8304,
+      "step": 2237
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2976151382669001e-05,
+      "loss": 0.2737,
+      "step": 2238
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2970322545657749e-05,
+      "loss": 0.7985,
+      "step": 2239
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2964492601605307e-05,
+      "loss": 0.7872,
+      "step": 2240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2958661552684499e-05,
+      "loss": 0.8619,
+      "step": 2241
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2952829401068562e-05,
+      "loss": 0.8276,
+      "step": 2242
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2946996148931144e-05,
+      "loss": 0.7808,
+      "step": 2243
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2941161798446296e-05,
+      "loss": 0.8329,
+      "step": 2244
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2935326351788491e-05,
+      "loss": 0.2994,
+      "step": 2245
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2929489811132603e-05,
+      "loss": 0.7755,
+      "step": 2246
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.292365217865391e-05,
+      "loss": 0.8433,
+      "step": 2247
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2917813456528106e-05,
+      "loss": 0.8212,
+      "step": 2248
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2911973646931287e-05,
+      "loss": 0.7664,
+      "step": 2249
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2906132752039948e-05,
+      "loss": 0.765,
+      "step": 2250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2900290774031001e-05,
+      "loss": 0.8807,
+      "step": 2251
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2894447715081753e-05,
+      "loss": 0.8769,
+      "step": 2252
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2888603577369914e-05,
+      "loss": 0.7339,
+      "step": 2253
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.28827583630736e-05,
+      "loss": 0.8235,
+      "step": 2254
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2876912074371326e-05,
+      "loss": 0.8099,
+      "step": 2255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2871064713442007e-05,
+      "loss": 0.7936,
+      "step": 2256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2865216282464963e-05,
+      "loss": 0.2915,
+      "step": 2257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2859366783619898e-05,
+      "loss": 0.841,
+      "step": 2258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2853516219086934e-05,
+      "loss": 0.8138,
+      "step": 2259
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2847664591046574e-05,
+      "loss": 0.8335,
+      "step": 2260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2841811901679726e-05,
+      "loss": 0.807,
+      "step": 2261
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2835958153167693e-05,
+      "loss": 0.7906,
+      "step": 2262
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2830103347692169e-05,
+      "loss": 0.7943,
+      "step": 2263
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2824247487435242e-05,
+      "loss": 0.8213,
+      "step": 2264
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2818390574579397e-05,
+      "loss": 0.781,
+      "step": 2265
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2812532611307514e-05,
+      "loss": 0.8461,
+      "step": 2266
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.280667359980285e-05,
+      "loss": 0.8149,
+      "step": 2267
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2800813542249073e-05,
+      "loss": 0.7943,
+      "step": 2268
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2794952440830219e-05,
+      "loss": 0.7299,
+      "step": 2269
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2789090297730734e-05,
+      "loss": 0.8293,
+      "step": 2270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2783227115135439e-05,
+      "loss": 0.8375,
+      "step": 2271
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2777362895229546e-05,
+      "loss": 0.8581,
+      "step": 2272
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2771497640198653e-05,
+      "loss": 0.7558,
+      "step": 2273
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2765631352228744e-05,
+      "loss": 0.8117,
+      "step": 2274
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2759764033506192e-05,
+      "loss": 0.8206,
+      "step": 2275
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2753895686217748e-05,
+      "loss": 0.8311,
+      "step": 2276
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.274802631255055e-05,
+      "loss": 0.7998,
+      "step": 2277
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2742155914692116e-05,
+      "loss": 0.7705,
+      "step": 2278
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2736284494830346e-05,
+      "loss": 0.7999,
+      "step": 2279
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2730412055153528e-05,
+      "loss": 0.765,
+      "step": 2280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2724538597850315e-05,
+      "loss": 0.7788,
+      "step": 2281
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2718664125109757e-05,
+      "loss": 0.831,
+      "step": 2282
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.271278863912127e-05,
+      "loss": 0.8054,
+      "step": 2283
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2706912142074654e-05,
+      "loss": 0.808,
+      "step": 2284
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2701034636160077e-05,
+      "loss": 0.776,
+      "step": 2285
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2695156123568097e-05,
+      "loss": 0.8175,
+      "step": 2286
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2689276606489634e-05,
+      "loss": 0.805,
+      "step": 2287
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.268339608711599e-05,
+      "loss": 0.8282,
+      "step": 2288
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2677514567638845e-05,
+      "loss": 0.8087,
+      "step": 2289
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2671632050250231e-05,
+      "loss": 0.825,
+      "step": 2290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2665748537142578e-05,
+      "loss": 0.8512,
+      "step": 2291
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.265986403050867e-05,
+      "loss": 0.8196,
+      "step": 2292
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2653978532541672e-05,
+      "loss": 0.878,
+      "step": 2293
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2648092045435103e-05,
+      "loss": 0.7691,
+      "step": 2294
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2642204571382868e-05,
+      "loss": 0.7823,
+      "step": 2295
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2636316112579234e-05,
+      "loss": 0.7782,
+      "step": 2296
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2630426671218828e-05,
+      "loss": 0.8156,
+      "step": 2297
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2624536249496652e-05,
+      "loss": 0.7717,
+      "step": 2298
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2618644849608068e-05,
+      "loss": 0.817,
+      "step": 2299
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2612752473748808e-05,
+      "loss": 0.7905,
+      "step": 2300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2606859124114963e-05,
+      "loss": 0.87,
+      "step": 2301
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2600964802902986e-05,
+      "loss": 0.8135,
+      "step": 2302
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2595069512309697e-05,
+      "loss": 0.8316,
+      "step": 2303
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2589173254532272e-05,
+      "loss": 0.278,
+      "step": 2304
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2583276031768255e-05,
+      "loss": 0.7922,
+      "step": 2305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2577377846215541e-05,
+      "loss": 0.8256,
+      "step": 2306
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.257147870007239e-05,
+      "loss": 0.8003,
+      "step": 2307
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2565578595537414e-05,
+      "loss": 0.8351,
+      "step": 2308
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2559677534809588e-05,
+      "loss": 0.7554,
+      "step": 2309
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2553775520088243e-05,
+      "loss": 0.7829,
+      "step": 2310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2547872553573062e-05,
+      "loss": 0.8896,
+      "step": 2311
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2541968637464086e-05,
+      "loss": 0.8522,
+      "step": 2312
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2536063773961704e-05,
+      "loss": 0.8075,
+      "step": 2313
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.253015796526667e-05,
+      "loss": 0.8248,
+      "step": 2314
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2524251213580075e-05,
+      "loss": 0.8321,
+      "step": 2315
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2518343521103376e-05,
+      "loss": 0.7513,
+      "step": 2316
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2512434890038369e-05,
+      "loss": 0.7606,
+      "step": 2317
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2506525322587207e-05,
+      "loss": 0.3084,
+      "step": 2318
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.250061482095239e-05,
+      "loss": 0.2722,
+      "step": 2319
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2494703387336765e-05,
+      "loss": 0.7829,
+      "step": 2320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2488791023943528e-05,
+      "loss": 0.7724,
+      "step": 2321
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.248287773297622e-05,
+      "loss": 0.7908,
+      "step": 2322
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2476963516638727e-05,
+      "loss": 0.8565,
+      "step": 2323
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2471048377135285e-05,
+      "loss": 0.7869,
+      "step": 2324
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2465132316670467e-05,
+      "loss": 0.847,
+      "step": 2325
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2459215337449195e-05,
+      "loss": 0.7874,
+      "step": 2326
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2453297441676733e-05,
+      "loss": 0.7914,
+      "step": 2327
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2447378631558679e-05,
+      "loss": 0.8587,
+      "step": 2328
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2441458909300981e-05,
+      "loss": 0.7847,
+      "step": 2329
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2435538277109919e-05,
+      "loss": 0.7744,
+      "step": 2330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2429616737192128e-05,
+      "loss": 0.8317,
+      "step": 2331
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2423694291754557e-05,
+      "loss": 0.8153,
+      "step": 2332
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2417770943004514e-05,
+      "loss": 0.7871,
+      "step": 2333
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.241184669314963e-05,
+      "loss": 0.8137,
+      "step": 2334
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2405921544397879e-05,
+      "loss": 0.8045,
+      "step": 2335
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2399995498957564e-05,
+      "loss": 0.8456,
+      "step": 2336
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2394068559037333e-05,
+      "loss": 0.7585,
+      "step": 2337
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2388140726846159e-05,
+      "loss": 0.7695,
+      "step": 2338
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2382212004593344e-05,
+      "loss": 0.7962,
+      "step": 2339
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.237628239448853e-05,
+      "loss": 0.8145,
+      "step": 2340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2370351898741687e-05,
+      "loss": 0.7765,
+      "step": 2341
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2364420519563111e-05,
+      "loss": 0.7297,
+      "step": 2342
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2358488259163438e-05,
+      "loss": 0.8648,
+      "step": 2343
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.235255511975362e-05,
+      "loss": 0.858,
+      "step": 2344
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2346621103544945e-05,
+      "loss": 0.8059,
+      "step": 2345
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.234068621274902e-05,
+      "loss": 0.8434,
+      "step": 2346
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2334750449577788e-05,
+      "loss": 0.8614,
+      "step": 2347
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2328813816243508e-05,
+      "loss": 0.835,
+      "step": 2348
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2322876314958772e-05,
+      "loss": 0.8077,
+      "step": 2349
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2316937947936488e-05,
+      "loss": 0.8459,
+      "step": 2350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2310998717389887e-05,
+      "loss": 0.8233,
+      "step": 2351
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2305058625532524e-05,
+      "loss": 0.7364,
+      "step": 2352
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.229911767457828e-05,
+      "loss": 0.7941,
+      "step": 2353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2293175866741348e-05,
+      "loss": 0.7873,
+      "step": 2354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2287233204236246e-05,
+      "loss": 0.8597,
+      "step": 2355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2281289689277806e-05,
+      "loss": 0.851,
+      "step": 2356
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2275345324081181e-05,
+      "loss": 0.7883,
+      "step": 2357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.226940011086184e-05,
+      "loss": 0.8238,
+      "step": 2358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2263454051835567e-05,
+      "loss": 0.8147,
+      "step": 2359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2257507149218462e-05,
+      "loss": 0.7553,
+      "step": 2360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2251559405226943e-05,
+      "loss": 0.7749,
+      "step": 2361
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2245610822077731e-05,
+      "loss": 0.8329,
+      "step": 2362
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2239661401987875e-05,
+      "loss": 0.8207,
+      "step": 2363
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2233711147174724e-05,
+      "loss": 0.863,
+      "step": 2364
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.222776005985594e-05,
+      "loss": 0.8149,
+      "step": 2365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2221808142249501e-05,
+      "loss": 0.8036,
+      "step": 2366
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2215855396573687e-05,
+      "loss": 0.7705,
+      "step": 2367
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2209901825047093e-05,
+      "loss": 0.8122,
+      "step": 2368
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2203947429888618e-05,
+      "loss": 0.7775,
+      "step": 2369
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2197992213317471e-05,
+      "loss": 0.7873,
+      "step": 2370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2192036177553164e-05,
+      "loss": 0.8431,
+      "step": 2371
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2186079324815513e-05,
+      "loss": 0.8323,
+      "step": 2372
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2180121657324641e-05,
+      "loss": 0.8206,
+      "step": 2373
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2174163177300979e-05,
+      "loss": 0.771,
+      "step": 2374
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2168203886965255e-05,
+      "loss": 0.8264,
+      "step": 2375
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.21622437885385e-05,
+      "loss": 0.838,
+      "step": 2376
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2156282884242046e-05,
+      "loss": 0.7811,
+      "step": 2377
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2150321176297523e-05,
+      "loss": 0.8525,
+      "step": 2378
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.214435866692687e-05,
+      "loss": 0.8521,
+      "step": 2379
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2138395358352313e-05,
+      "loss": 0.817,
+      "step": 2380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2132431252796384e-05,
+      "loss": 0.7812,
+      "step": 2381
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2126466352481909e-05,
+      "loss": 0.7859,
+      "step": 2382
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2120500659632007e-05,
+      "loss": 0.8331,
+      "step": 2383
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.21145341764701e-05,
+      "loss": 0.8039,
+      "step": 2384
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2108566905219896e-05,
+      "loss": 0.854,
+      "step": 2385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2102598848105404e-05,
+      "loss": 0.8648,
+      "step": 2386
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2096630007350925e-05,
+      "loss": 0.8243,
+      "step": 2387
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2090660385181046e-05,
+      "loss": 0.7973,
+      "step": 2388
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.208468998382065e-05,
+      "loss": 0.7986,
+      "step": 2389
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2078718805494907e-05,
+      "loss": 0.7767,
+      "step": 2390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2072746852429285e-05,
+      "loss": 0.7838,
+      "step": 2391
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.206677412684953e-05,
+      "loss": 0.809,
+      "step": 2392
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2060800630981686e-05,
+      "loss": 0.8299,
+      "step": 2393
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2054826367052073e-05,
+      "loss": 0.8077,
+      "step": 2394
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2048851337287303e-05,
+      "loss": 0.3073,
+      "step": 2395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2042875543914283e-05,
+      "loss": 0.7972,
+      "step": 2396
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2036898989160181e-05,
+      "loss": 0.2675,
+      "step": 2397
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2030921675252476e-05,
+      "loss": 0.7861,
+      "step": 2398
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2024943604418911e-05,
+      "loss": 0.7712,
+      "step": 2399
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2018964778887516e-05,
+      "loss": 0.8381,
+      "step": 2400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2012985200886602e-05,
+      "loss": 0.8378,
+      "step": 2401
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2007004872644765e-05,
+      "loss": 0.7962,
+      "step": 2402
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2001023796390879e-05,
+      "loss": 0.8335,
+      "step": 2403
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1995041974354091e-05,
+      "loss": 0.792,
+      "step": 2404
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1989059408763833e-05,
+      "loss": 0.8578,
+      "step": 2405
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1983076101849806e-05,
+      "loss": 0.8365,
+      "step": 2406
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1977092055842e-05,
+      "loss": 0.8366,
+      "step": 2407
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1971107272970666e-05,
+      "loss": 0.8191,
+      "step": 2408
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1965121755466342e-05,
+      "loss": 0.7898,
+      "step": 2409
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1959135505559834e-05,
+      "loss": 0.8174,
+      "step": 2410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1953148525482214e-05,
+      "loss": 0.7788,
+      "step": 2411
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1947160817464845e-05,
+      "loss": 0.8418,
+      "step": 2412
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1941172383739337e-05,
+      "loss": 0.8657,
+      "step": 2413
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1935183226537594e-05,
+      "loss": 0.7918,
+      "step": 2414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1929193348091776e-05,
+      "loss": 0.8582,
+      "step": 2415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1923202750634311e-05,
+      "loss": 0.7686,
+      "step": 2416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1917211436397902e-05,
+      "loss": 0.821,
+      "step": 2417
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1911219407615513e-05,
+      "loss": 0.7601,
+      "step": 2418
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1905226666520382e-05,
+      "loss": 0.8006,
+      "step": 2419
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1899233215346004e-05,
+      "loss": 0.7735,
+      "step": 2420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1893239056326143e-05,
+      "loss": 0.2829,
+      "step": 2421
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1887244191694822e-05,
+      "loss": 0.8492,
+      "step": 2422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.188124862368634e-05,
+      "loss": 0.8256,
+      "step": 2423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187525235453524e-05,
+      "loss": 0.8056,
+      "step": 2424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1869255386476343e-05,
+      "loss": 0.7652,
+      "step": 2425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1863257721744719e-05,
+      "loss": 0.7882,
+      "step": 2426
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.18572593625757e-05,
+      "loss": 0.2665,
+      "step": 2427
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1851260311204884e-05,
+      "loss": 0.7985,
+      "step": 2428
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1845260569868113e-05,
+      "loss": 0.7648,
+      "step": 2429
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1839260140801505e-05,
+      "loss": 0.7756,
+      "step": 2430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1833259026241416e-05,
+      "loss": 0.7207,
+      "step": 2431
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1827257228424469e-05,
+      "loss": 0.7829,
+      "step": 2432
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1821254749587538e-05,
+      "loss": 0.826,
+      "step": 2433
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1815251591967747e-05,
+      "loss": 0.7825,
+      "step": 2434
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1809247757802482e-05,
+      "loss": 0.8006,
+      "step": 2435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1803243249329374e-05,
+      "loss": 0.7942,
+      "step": 2436
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1797238068786307e-05,
+      "loss": 0.8488,
+      "step": 2437
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1791232218411415e-05,
+      "loss": 0.8174,
+      "step": 2438
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1785225700443081e-05,
+      "loss": 0.7305,
+      "step": 2439
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1779218517119943e-05,
+      "loss": 0.829,
+      "step": 2440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1773210670680879e-05,
+      "loss": 0.8598,
+      "step": 2441
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.176720216336502e-05,
+      "loss": 0.8578,
+      "step": 2442
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1761192997411736e-05,
+      "loss": 0.7619,
+      "step": 2443
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1755183175060654e-05,
+      "loss": 0.8308,
+      "step": 2444
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.174917269855163e-05,
+      "loss": 0.8137,
+      "step": 2445
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.174316157012478e-05,
+      "loss": 0.8182,
+      "step": 2446
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1737149792020455e-05,
+      "loss": 0.8203,
+      "step": 2447
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1731137366479242e-05,
+      "loss": 0.8647,
+      "step": 2448
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1725124295741983e-05,
+      "loss": 0.8358,
+      "step": 2449
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1719110582049749e-05,
+      "loss": 0.7879,
+      "step": 2450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1713096227643859e-05,
+      "loss": 0.8089,
+      "step": 2451
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1707081234765862e-05,
+      "loss": 0.793,
+      "step": 2452
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1701065605657558e-05,
+      "loss": 0.7517,
+      "step": 2453
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1695049342560969e-05,
+      "loss": 0.7838,
+      "step": 2454
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.168903244771836e-05,
+      "loss": 0.8077,
+      "step": 2455
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1683014923372235e-05,
+      "loss": 0.8233,
+      "step": 2456
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1676996771765328e-05,
+      "loss": 0.8365,
+      "step": 2457
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.167097799514061e-05,
+      "loss": 0.7907,
+      "step": 2458
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.166495859574128e-05,
+      "loss": 0.8132,
+      "step": 2459
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1658938575810772e-05,
+      "loss": 0.7923,
+      "step": 2460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1652917937592756e-05,
+      "loss": 0.8081,
+      "step": 2461
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.164689668333112e-05,
+      "loss": 0.8036,
+      "step": 2462
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1640874815269997e-05,
+      "loss": 0.8027,
+      "step": 2463
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.163485233565374e-05,
+      "loss": 0.2675,
+      "step": 2464
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1628829246726929e-05,
+      "loss": 0.8148,
+      "step": 2465
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1622805550734373e-05,
+      "loss": 0.7809,
+      "step": 2466
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1616781249921108e-05,
+      "loss": 0.7802,
+      "step": 2467
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1610756346532397e-05,
+      "loss": 0.7561,
+      "step": 2468
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1604730842813723e-05,
+      "loss": 0.742,
+      "step": 2469
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1598704741010796e-05,
+      "loss": 0.784,
+      "step": 2470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1592678043369547e-05,
+      "loss": 0.819,
+      "step": 2471
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1586650752136133e-05,
+      "loss": 0.2684,
+      "step": 2472
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1580622869556926e-05,
+      "loss": 0.7922,
+      "step": 2473
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1574594397878527e-05,
+      "loss": 0.8259,
+      "step": 2474
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1568565339347745e-05,
+      "loss": 0.2682,
+      "step": 2475
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1562535696211618e-05,
+      "loss": 0.8312,
+      "step": 2476
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1556505470717399e-05,
+      "loss": 0.7117,
+      "step": 2477
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1550474665112552e-05,
+      "loss": 0.2967,
+      "step": 2478
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1544443281644769e-05,
+      "loss": 0.8147,
+      "step": 2479
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1538411322561947e-05,
+      "loss": 0.8575,
+      "step": 2480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1532378790112197e-05,
+      "loss": 0.8478,
+      "step": 2481
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1526345686543856e-05,
+      "loss": 0.8106,
+      "step": 2482
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1520312014105462e-05,
+      "loss": 0.8118,
+      "step": 2483
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1514277775045768e-05,
+      "loss": 0.8052,
+      "step": 2484
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1508242971613741e-05,
+      "loss": 0.7977,
+      "step": 2485
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1502207606058552e-05,
+      "loss": 0.8144,
+      "step": 2486
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.149617168062959e-05,
+      "loss": 0.8435,
+      "step": 2487
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1490135197576447e-05,
+      "loss": 0.7946,
+      "step": 2488
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1484098159148924e-05,
+      "loss": 0.8841,
+      "step": 2489
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1478060567597032e-05,
+      "loss": 0.8075,
+      "step": 2490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1472022425170981e-05,
+      "loss": 0.8395,
+      "step": 2491
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.146598373412119e-05,
+      "loss": 0.8533,
+      "step": 2492
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.145994449669829e-05,
+      "loss": 0.7849,
+      "step": 2493
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1453904715153102e-05,
+      "loss": 0.7897,
+      "step": 2494
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.144786439173666e-05,
+      "loss": 0.7907,
+      "step": 2495
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1441823528700197e-05,
+      "loss": 0.7938,
+      "step": 2496
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1435782128295143e-05,
+      "loss": 0.8042,
+      "step": 2497
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1429740192773138e-05,
+      "loss": 0.7914,
+      "step": 2498
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1423697724386008e-05,
+      "loss": 0.8108,
+      "step": 2499
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1417654725385789e-05,
+      "loss": 0.7716,
+      "step": 2500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1411611198024712e-05,
+      "loss": 0.8015,
+      "step": 2501
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1405567144555205e-05,
+      "loss": 0.771,
+      "step": 2502
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1399522567229884e-05,
+      "loss": 0.7389,
+      "step": 2503
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1393477468301568e-05,
+      "loss": 0.8073,
+      "step": 2504
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1387431850023279e-05,
+      "loss": 0.782,
+      "step": 2505
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.138138571464821e-05,
+      "loss": 0.7918,
+      "step": 2506
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.137533906442977e-05,
+      "loss": 0.8015,
+      "step": 2507
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1369291901621546e-05,
+      "loss": 0.8224,
+      "step": 2508
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1363244228477319e-05,
+      "loss": 0.825,
+      "step": 2509
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1357196047251058e-05,
+      "loss": 0.7754,
+      "step": 2510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1351147360196928e-05,
+      "loss": 0.7355,
+      "step": 2511
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.134509816956928e-05,
+      "loss": 0.8448,
+      "step": 2512
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1339048477622647e-05,
+      "loss": 0.7797,
+      "step": 2513
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1332998286611759e-05,
+      "loss": 0.7759,
+      "step": 2514
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1326947598791518e-05,
+      "loss": 0.299,
+      "step": 2515
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1320896416417026e-05,
+      "loss": 0.7972,
+      "step": 2516
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1314844741743558e-05,
+      "loss": 0.8145,
+      "step": 2517
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1308792577026585e-05,
+      "loss": 0.8543,
+      "step": 2518
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1302739924521743e-05,
+      "loss": 0.7865,
+      "step": 2519
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1296686786484865e-05,
+      "loss": 0.8096,
+      "step": 2520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.129063316517196e-05,
+      "loss": 0.8102,
+      "step": 2521
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1284579062839214e-05,
+      "loss": 0.8318,
+      "step": 2522
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1278524481742999e-05,
+      "loss": 0.8186,
+      "step": 2523
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1272469424139859e-05,
+      "loss": 0.8018,
+      "step": 2524
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1266413892286514e-05,
+      "loss": 0.7898,
+      "step": 2525
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1260357888439872e-05,
+      "loss": 0.7375,
+      "step": 2526
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1254301414857004e-05,
+      "loss": 0.7958,
+      "step": 2527
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1248244473795167e-05,
+      "loss": 0.8494,
+      "step": 2528
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1242187067511784e-05,
+      "loss": 0.8068,
+      "step": 2529
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1236129198264453e-05,
+      "loss": 0.7711,
+      "step": 2530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1230070868310948e-05,
+      "loss": 0.806,
+      "step": 2531
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1224012079909208e-05,
+      "loss": 0.797,
+      "step": 2532
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1217952835317354e-05,
+      "loss": 0.7311,
+      "step": 2533
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.121189313679367e-05,
+      "loss": 0.7928,
+      "step": 2534
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1205832986596605e-05,
+      "loss": 0.731,
+      "step": 2535
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1199772386984784e-05,
+      "loss": 0.7979,
+      "step": 2536
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1193711340216996e-05,
+      "loss": 0.7613,
+      "step": 2537
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1187649848552198e-05,
+      "loss": 0.2581,
+      "step": 2538
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.118158791424951e-05,
+      "loss": 0.811,
+      "step": 2539
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1175525539568222e-05,
+      "loss": 0.8719,
+      "step": 2540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1169462726767779e-05,
+      "loss": 0.8519,
+      "step": 2541
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1163399478107804e-05,
+      "loss": 0.7917,
+      "step": 2542
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1157335795848066e-05,
+      "loss": 0.8496,
+      "step": 2543
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.115127168224851e-05,
+      "loss": 0.7599,
+      "step": 2544
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1145207139569233e-05,
+      "loss": 0.7835,
+      "step": 2545
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1139142170070487e-05,
+      "loss": 0.7622,
+      "step": 2546
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1133076776012702e-05,
+      "loss": 0.7504,
+      "step": 2547
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1127010959656443e-05,
+      "loss": 0.8075,
+      "step": 2548
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1120944723262454e-05,
+      "loss": 0.7845,
+      "step": 2549
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1114878069091618e-05,
+      "loss": 0.7662,
+      "step": 2550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.110881099940499e-05,
+      "loss": 0.7996,
+      "step": 2551
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1102743516463756e-05,
+      "loss": 0.8049,
+      "step": 2552
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1096675622529282e-05,
+      "loss": 0.7996,
+      "step": 2553
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1090607319863076e-05,
+      "loss": 0.7956,
+      "step": 2554
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1084538610726794e-05,
+      "loss": 0.7851,
+      "step": 2555
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1078469497382252e-05,
+      "loss": 0.7644,
+      "step": 2556
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.107239998209141e-05,
+      "loss": 0.769,
+      "step": 2557
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1066330067116385e-05,
+      "loss": 0.2757,
+      "step": 2558
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1060259754719427e-05,
+      "loss": 0.807,
+      "step": 2559
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1054189047162957e-05,
+      "loss": 0.8272,
+      "step": 2560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1048117946709526e-05,
+      "loss": 0.8073,
+      "step": 2561
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1042046455621842e-05,
+      "loss": 0.8383,
+      "step": 2562
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1035974576162749e-05,
+      "loss": 0.7518,
+      "step": 2563
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.102990231059524e-05,
+      "loss": 0.8176,
+      "step": 2564
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1023829661182457e-05,
+      "loss": 0.7503,
+      "step": 2565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1017756630187676e-05,
+      "loss": 0.8253,
+      "step": 2566
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1011683219874324e-05,
+      "loss": 0.8174,
+      "step": 2567
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1005609432505961e-05,
+      "loss": 0.8683,
+      "step": 2568
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0999535270346294e-05,
+      "loss": 0.7879,
+      "step": 2569
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0993460735659166e-05,
+      "loss": 0.2881,
+      "step": 2570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.098738583070856e-05,
+      "loss": 0.7371,
+      "step": 2571
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.09813105577586e-05,
+      "loss": 0.8921,
+      "step": 2572
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0975234919073542e-05,
+      "loss": 0.7352,
+      "step": 2573
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0969158916917782e-05,
+      "loss": 0.8265,
+      "step": 2574
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0963082553555844e-05,
+      "loss": 0.8324,
+      "step": 2575
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0957005831252399e-05,
+      "loss": 0.8084,
+      "step": 2576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0950928752272248e-05,
+      "loss": 0.809,
+      "step": 2577
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0944851318880314e-05,
+      "loss": 0.234,
+      "step": 2578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0938773533341667e-05,
+      "loss": 0.7725,
+      "step": 2579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0932695397921497e-05,
+      "loss": 0.823,
+      "step": 2580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0926616914885133e-05,
+      "loss": 0.83,
+      "step": 2581
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0920538086498028e-05,
+      "loss": 0.2727,
+      "step": 2582
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0914458915025768e-05,
+      "loss": 0.7473,
+      "step": 2583
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0908379402734059e-05,
+      "loss": 0.8387,
+      "step": 2584
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0902299551888741e-05,
+      "loss": 0.8572,
+      "step": 2585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0896219364755782e-05,
+      "loss": 0.8046,
+      "step": 2586
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0890138843601268e-05,
+      "loss": 0.8313,
+      "step": 2587
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0884057990691416e-05,
+      "loss": 0.7963,
+      "step": 2588
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0877976808292563e-05,
+      "loss": 0.821,
+      "step": 2589
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0871895298671169e-05,
+      "loss": 0.7883,
+      "step": 2590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0865813464093818e-05,
+      "loss": 0.7712,
+      "step": 2591
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0859731306827208e-05,
+      "loss": 0.8069,
+      "step": 2592
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0853648829138172e-05,
+      "loss": 0.7652,
+      "step": 2593
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0847566033293651e-05,
+      "loss": 0.8161,
+      "step": 2594
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0841482921560702e-05,
+      "loss": 0.8047,
+      "step": 2595
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0835399496206507e-05,
+      "loss": 0.7428,
+      "step": 2596
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0829315759498364e-05,
+      "loss": 0.7435,
+      "step": 2597
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.082323171370369e-05,
+      "loss": 0.7901,
+      "step": 2598
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0817147361090004e-05,
+      "loss": 0.7895,
+      "step": 2599
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0811062703924957e-05,
+      "loss": 0.7959,
+      "step": 2600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0804977744476296e-05,
+      "loss": 0.7748,
+      "step": 2601
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0798892485011898e-05,
+      "loss": 0.8147,
+      "step": 2602
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0792806927799737e-05,
+      "loss": 0.7961,
+      "step": 2603
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0786721075107906e-05,
+      "loss": 0.8042,
+      "step": 2604
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0780634929204607e-05,
+      "loss": 0.8579,
+      "step": 2605
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0774548492358154e-05,
+      "loss": 0.8004,
+      "step": 2606
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0768461766836959e-05,
+      "loss": 0.741,
+      "step": 2607
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0762374754909552e-05,
+      "loss": 0.7875,
+      "step": 2608
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.075628745884457e-05,
+      "loss": 0.8345,
+      "step": 2609
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0750199880910743e-05,
+      "loss": 0.7534,
+      "step": 2610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0744112023376924e-05,
+      "loss": 0.8137,
+      "step": 2611
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0738023888512062e-05,
+      "loss": 0.8346,
+      "step": 2612
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.07319354785852e-05,
+      "loss": 0.833,
+      "step": 2613
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0725846795865502e-05,
+      "loss": 0.8336,
+      "step": 2614
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0719757842622219e-05,
+      "loss": 0.8124,
+      "step": 2615
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.071366862112471e-05,
+      "loss": 0.7746,
+      "step": 2616
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0707579133642429e-05,
+      "loss": 0.7949,
+      "step": 2617
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0701489382444936e-05,
+      "loss": 0.8169,
+      "step": 2618
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0695399369801883e-05,
+      "loss": 0.7883,
+      "step": 2619
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.068930909798302e-05,
+      "loss": 0.8384,
+      "step": 2620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.06832185692582e-05,
+      "loss": 0.8309,
+      "step": 2621
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0677127785897365e-05,
+      "loss": 0.8401,
+      "step": 2622
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0671036750170554e-05,
+      "loss": 0.8121,
+      "step": 2623
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0664945464347895e-05,
+      "loss": 0.8584,
+      "step": 2624
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0658853930699621e-05,
+      "loss": 0.7972,
+      "step": 2625
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0652762151496052e-05,
+      "loss": 0.8046,
+      "step": 2626
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0646670129007592e-05,
+      "loss": 0.7737,
+      "step": 2627
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0640577865504747e-05,
+      "loss": 0.769,
+      "step": 2628
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0634485363258103e-05,
+      "loss": 0.8268,
+      "step": 2629
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0628392624538342e-05,
+      "loss": 0.8075,
+      "step": 2630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.062229965161623e-05,
+      "loss": 0.7789,
+      "step": 2631
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0616206446762627e-05,
+      "loss": 0.8666,
+      "step": 2632
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.061011301224847e-05,
+      "loss": 0.755,
+      "step": 2633
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0604019350344788e-05,
+      "loss": 0.2623,
+      "step": 2634
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0597925463322691e-05,
+      "loss": 0.7114,
+      "step": 2635
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0591831353453375e-05,
+      "loss": 0.8529,
+      "step": 2636
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0585737023008122e-05,
+      "loss": 0.2642,
+      "step": 2637
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0579642474258292e-05,
+      "loss": 0.8313,
+      "step": 2638
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0573547709475325e-05,
+      "loss": 0.8388,
+      "step": 2639
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0567452730930743e-05,
+      "loss": 0.8425,
+      "step": 2640
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0561357540896149e-05,
+      "loss": 0.7468,
+      "step": 2641
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055526214164323e-05,
+      "loss": 0.7935,
+      "step": 2642
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.054916653544374e-05,
+      "loss": 0.2605,
+      "step": 2643
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0543070724569517e-05,
+      "loss": 0.8199,
+      "step": 2644
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0536974711292473e-05,
+      "loss": 0.8503,
+      "step": 2645
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0530878497884599e-05,
+      "loss": 0.8433,
+      "step": 2646
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.052478208661795e-05,
+      "loss": 0.7948,
+      "step": 2647
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0518685479764673e-05,
+      "loss": 0.8064,
+      "step": 2648
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0512588679596974e-05,
+      "loss": 0.8762,
+      "step": 2649
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0506491688387128e-05,
+      "loss": 0.7943,
+      "step": 2650
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0500394508407497e-05,
+      "loss": 0.7556,
+      "step": 2651
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0494297141930496e-05,
+      "loss": 0.8572,
+      "step": 2652
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0488199591228626e-05,
+      "loss": 0.7834,
+      "step": 2653
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.048210185857444e-05,
+      "loss": 0.8452,
+      "step": 2654
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0476003946240569e-05,
+      "loss": 0.8338,
+      "step": 2655
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0469905856499714e-05,
+      "loss": 0.7727,
+      "step": 2656
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046380759162463e-05,
+      "loss": 0.8268,
+      "step": 2657
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.045770915388815e-05,
+      "loss": 0.8057,
+      "step": 2658
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0451610545563165e-05,
+      "loss": 0.8437,
+      "step": 2659
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0445511768922632e-05,
+      "loss": 0.8151,
+      "step": 2660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0439412826239559e-05,
+      "loss": 0.8289,
+      "step": 2661
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0433313719787034e-05,
+      "loss": 0.2969,
+      "step": 2662
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0427214451838203e-05,
+      "loss": 0.8391,
+      "step": 2663
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0421115024666257e-05,
+      "loss": 0.773,
+      "step": 2664
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0415015440544462e-05,
+      "loss": 0.8441,
+      "step": 2665
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0408915701746138e-05,
+      "loss": 0.7845,
+      "step": 2666
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.040281581054466e-05,
+      "loss": 0.81,
+      "step": 2667
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.039671576921346e-05,
+      "loss": 0.8562,
+      "step": 2668
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0390615580026027e-05,
+      "loss": 0.8092,
+      "step": 2669
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0384515245255908e-05,
+      "loss": 0.7639,
+      "step": 2670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0378414767176706e-05,
+      "loss": 0.8024,
+      "step": 2671
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0372314148062064e-05,
+      "loss": 0.7264,
+      "step": 2672
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0366213390185691e-05,
+      "loss": 0.8287,
+      "step": 2673
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0360112495821346e-05,
+      "loss": 0.8244,
+      "step": 2674
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0354011467242831e-05,
+      "loss": 0.2427,
+      "step": 2675
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0347910306724009e-05,
+      "loss": 0.8131,
+      "step": 2676
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0341809016538787e-05,
+      "loss": 0.7718,
+      "step": 2677
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0335707598961111e-05,
+      "loss": 0.7945,
+      "step": 2678
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0329606056264994e-05,
+      "loss": 0.8216,
+      "step": 2679
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0323504390724477e-05,
+      "loss": 0.7872,
+      "step": 2680
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0317402604613662e-05,
+      "loss": 0.7973,
+      "step": 2681
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0311300700206683e-05,
+      "loss": 0.7813,
+      "step": 2682
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0305198679777725e-05,
+      "loss": 0.8033,
+      "step": 2683
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0299096545601018e-05,
+      "loss": 0.8148,
+      "step": 2684
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0292994299950828e-05,
+      "loss": 0.7355,
+      "step": 2685
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.028689194510147e-05,
+      "loss": 0.7712,
+      "step": 2686
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0280789483327291e-05,
+      "loss": 0.7721,
+      "step": 2687
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0274686916902686e-05,
+      "loss": 0.7671,
+      "step": 2688
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0268584248102081e-05,
+      "loss": 0.8251,
+      "step": 2689
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0262481479199949e-05,
+      "loss": 0.8062,
+      "step": 2690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0256378612470797e-05,
+      "loss": 0.7094,
+      "step": 2691
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0250275650189166e-05,
+      "loss": 0.778,
+      "step": 2692
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0244172594629634e-05,
+      "loss": 0.8195,
+      "step": 2693
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.023806944806681e-05,
+      "loss": 0.798,
+      "step": 2694
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0231966212775345e-05,
+      "loss": 0.8298,
+      "step": 2695
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0225862891029918e-05,
+      "loss": 0.7968,
+      "step": 2696
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0219759485105245e-05,
+      "loss": 0.7447,
+      "step": 2697
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0213655997276064e-05,
+      "loss": 0.7952,
+      "step": 2698
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.020755242981715e-05,
+      "loss": 0.8615,
+      "step": 2699
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0201448785003305e-05,
+      "loss": 0.8193,
+      "step": 2700
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0195345065109366e-05,
+      "loss": 0.7881,
+      "step": 2701
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0189241272410191e-05,
+      "loss": 0.8206,
+      "step": 2702
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.018313740918067e-05,
+      "loss": 0.7771,
+      "step": 2703
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0177033477695713e-05,
+      "loss": 0.8107,
+      "step": 2704
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.017092948023026e-05,
+      "loss": 0.8109,
+      "step": 2705
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0164825419059274e-05,
+      "loss": 0.7306,
+      "step": 2706
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.015872129645775e-05,
+      "loss": 0.8046,
+      "step": 2707
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0152617114700689e-05,
+      "loss": 0.7609,
+      "step": 2708
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0146512876063133e-05,
+      "loss": 0.829,
+      "step": 2709
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0140408582820126e-05,
+      "loss": 0.8452,
+      "step": 2710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0134304237246749e-05,
+      "loss": 0.7826,
+      "step": 2711
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0128199841618089e-05,
+      "loss": 0.2774,
+      "step": 2712
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0122095398209264e-05,
+      "loss": 0.7871,
+      "step": 2713
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0115990909295405e-05,
+      "loss": 0.8039,
+      "step": 2714
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0109886377151658e-05,
+      "loss": 0.7534,
+      "step": 2715
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0103781804053185e-05,
+      "loss": 0.8049,
+      "step": 2716
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0097677192275161e-05,
+      "loss": 0.8125,
+      "step": 2717
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0091572544092788e-05,
+      "loss": 0.2496,
+      "step": 2718
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0085467861781261e-05,
+      "loss": 0.2735,
+      "step": 2719
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0079363147615809e-05,
+      "loss": 0.7708,
+      "step": 2720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0073258403871658e-05,
+      "loss": 0.8688,
+      "step": 2721
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.006715363282405e-05,
+      "loss": 0.8089,
+      "step": 2722
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.006104883674824e-05,
+      "loss": 0.8971,
+      "step": 2723
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0054944017919486e-05,
+      "loss": 0.8238,
+      "step": 2724
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0048839178613061e-05,
+      "loss": 0.7817,
+      "step": 2725
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0042734321104243e-05,
+      "loss": 0.818,
+      "step": 2726
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0036629447668313e-05,
+      "loss": 0.7588,
+      "step": 2727
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0030524560580566e-05,
+      "loss": 0.7569,
+      "step": 2728
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0024419662116292e-05,
+      "loss": 0.2648,
+      "step": 2729
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0018314754550795e-05,
+      "loss": 0.7787,
+      "step": 2730
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0012209840159377e-05,
+      "loss": 0.7928,
+      "step": 2731
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0006104921217346e-05,
+      "loss": 0.8086,
+      "step": 2732
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 0.7589,
+      "step": 2733
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.993895078782659e-06,
+      "loss": 0.8684,
+      "step": 2734
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.987790159840625e-06,
+      "loss": 0.8072,
+      "step": 2735
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.981685245449208e-06,
+      "loss": 0.8174,
+      "step": 2736
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.975580337883711e-06,
+      "loss": 0.7971,
+      "step": 2737
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.969475439419438e-06,
+      "loss": 0.8357,
+      "step": 2738
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.963370552331687e-06,
+      "loss": 0.8228,
+      "step": 2739
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.957265678895762e-06,
+      "loss": 0.8263,
+      "step": 2740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.951160821386942e-06,
+      "loss": 0.7531,
+      "step": 2741
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.945055982080517e-06,
+      "loss": 0.8081,
+      "step": 2742
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.938951163251764e-06,
+      "loss": 0.8293,
+      "step": 2743
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.932846367175952e-06,
+      "loss": 0.7903,
+      "step": 2744
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.926741596128346e-06,
+      "loss": 0.8256,
+      "step": 2745
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.920636852384194e-06,
+      "loss": 0.8346,
+      "step": 2746
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.914532138218742e-06,
+      "loss": 0.7721,
+      "step": 2747
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.908427455907216e-06,
+      "loss": 0.8422,
+      "step": 2748
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.902322807724839e-06,
+      "loss": 0.7641,
+      "step": 2749
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89621819594682e-06,
+      "loss": 0.7966,
+      "step": 2750
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.890113622848343e-06,
+      "loss": 0.8464,
+      "step": 2751
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.884009090704598e-06,
+      "loss": 0.7475,
+      "step": 2752
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.877904601790736e-06,
+      "loss": 0.8254,
+      "step": 2753
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.871800158381916e-06,
+      "loss": 0.7883,
+      "step": 2754
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.865695762753258e-06,
+      "loss": 0.7887,
+      "step": 2755
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.859591417179878e-06,
+      "loss": 0.8024,
+      "step": 2756
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.85348712393687e-06,
+      "loss": 0.8257,
+      "step": 2757
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.847382885299313e-06,
+      "loss": 0.7647,
+      "step": 2758
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.841278703542256e-06,
+      "loss": 0.853,
+      "step": 2759
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.83517458094073e-06,
+      "loss": 0.7876,
+      "step": 2760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.829070519769744e-06,
+      "loss": 0.7793,
+      "step": 2761
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.82296652230429e-06,
+      "loss": 0.8243,
+      "step": 2762
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.816862590819333e-06,
+      "loss": 0.828,
+      "step": 2763
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.810758727589814e-06,
+      "loss": 0.7522,
+      "step": 2764
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.804654934890637e-06,
+      "loss": 0.8138,
+      "step": 2765
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.798551214996697e-06,
+      "loss": 0.7945,
+      "step": 2766
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.792447570182852e-06,
+      "loss": 0.79,
+      "step": 2767
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.786344002723941e-06,
+      "loss": 0.753,
+      "step": 2768
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.780240514894758e-06,
+      "loss": 0.2864,
+      "step": 2769
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.774137108970084e-06,
+      "loss": 0.7628,
+      "step": 2770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.768033787224656e-06,
+      "loss": 0.7929,
+      "step": 2771
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.761930551933191e-06,
+      "loss": 0.8315,
+      "step": 2772
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.755827405370373e-06,
+      "loss": 0.766,
+      "step": 2773
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.749724349810838e-06,
+      "loss": 0.8214,
+      "step": 2774
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.743621387529207e-06,
+      "loss": 0.7625,
+      "step": 2775
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.737518520800053e-06,
+      "loss": 0.8445,
+      "step": 2776
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.73141575189792e-06,
+      "loss": 0.7976,
+      "step": 2777
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.725313083097319e-06,
+      "loss": 0.8227,
+      "step": 2778
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.719210516672712e-06,
+      "loss": 0.8227,
+      "step": 2779
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.713108054898535e-06,
+      "loss": 0.7893,
+      "step": 2780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.707005700049174e-06,
+      "loss": 0.8018,
+      "step": 2781
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.700903454398987e-06,
+      "loss": 0.8127,
+      "step": 2782
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.694801320222278e-06,
+      "loss": 0.8016,
+      "step": 2783
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.688699299793319e-06,
+      "loss": 0.7671,
+      "step": 2784
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.682597395386341e-06,
+      "loss": 0.7876,
+      "step": 2785
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.676495609275523e-06,
+      "loss": 0.7824,
+      "step": 2786
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.670393943735011e-06,
+      "loss": 0.7856,
+      "step": 2787
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.664292401038892e-06,
+      "loss": 0.8185,
+      "step": 2788
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.658190983461218e-06,
+      "loss": 0.8535,
+      "step": 2789
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.652089693275993e-06,
+      "loss": 0.8079,
+      "step": 2790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.645988532757169e-06,
+      "loss": 0.6858,
+      "step": 2791
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.63988750417866e-06,
+      "loss": 0.7732,
+      "step": 2792
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.633786609814312e-06,
+      "loss": 0.8456,
+      "step": 2793
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.627685851937939e-06,
+      "loss": 0.8215,
+      "step": 2794
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.6215852328233e-06,
+      "loss": 0.2696,
+      "step": 2795
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.615484754744094e-06,
+      "loss": 0.762,
+      "step": 2796
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.609384419973978e-06,
+      "loss": 0.7797,
+      "step": 2797
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.603284230786544e-06,
+      "loss": 0.8026,
+      "step": 2798
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.597184189455342e-06,
+      "loss": 0.8106,
+      "step": 2799
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.591084298253862e-06,
+      "loss": 0.8065,
+      "step": 2800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58498455945554e-06,
+      "loss": 0.8503,
+      "step": 2801
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.578884975333746e-06,
+      "loss": 0.2742,
+      "step": 2802
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.5727855481618e-06,
+      "loss": 0.7951,
+      "step": 2803
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.566686280212964e-06,
+      "loss": 0.8473,
+      "step": 2804
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.560587173760445e-06,
+      "loss": 0.8037,
+      "step": 2805
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.554488231077375e-06,
+      "loss": 0.846,
+      "step": 2806
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.548389454436839e-06,
+      "loss": 0.811,
+      "step": 2807
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54229084611185e-06,
+      "loss": 0.8192,
+      "step": 2808
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53619240837537e-06,
+      "loss": 0.8122,
+      "step": 2809
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.53009414350029e-06,
+      "loss": 0.7397,
+      "step": 2810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.523996053759433e-06,
+      "loss": 0.8108,
+      "step": 2811
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.517898141425563e-06,
+      "loss": 0.2589,
+      "step": 2812
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.51180040877138e-06,
+      "loss": 0.7993,
+      "step": 2813
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.505702858069504e-06,
+      "loss": 0.7548,
+      "step": 2814
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.499605491592508e-06,
+      "loss": 0.7936,
+      "step": 2815
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.493508311612874e-06,
+      "loss": 0.8212,
+      "step": 2816
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.48741132040303e-06,
+      "loss": 0.7228,
+      "step": 2817
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.481314520235329e-06,
+      "loss": 0.7749,
+      "step": 2818
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.475217913382053e-06,
+      "loss": 0.7833,
+      "step": 2819
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.469121502115407e-06,
+      "loss": 0.787,
+      "step": 2820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.46302528870753e-06,
+      "loss": 0.8056,
+      "step": 2821
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.456929275430485e-06,
+      "loss": 0.7705,
+      "step": 2822
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.450833464556263e-06,
+      "loss": 0.7916,
+      "step": 2823
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.444737858356776e-06,
+      "loss": 0.2725,
+      "step": 2824
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.438642459103854e-06,
+      "loss": 0.7703,
+      "step": 2825
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.43254726906926e-06,
+      "loss": 0.7935,
+      "step": 2826
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.42645229052468e-06,
+      "loss": 0.7946,
+      "step": 2827
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.420357525741712e-06,
+      "loss": 0.8281,
+      "step": 2828
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41426297699188e-06,
+      "loss": 0.7303,
+      "step": 2829
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.408168646546627e-06,
+      "loss": 0.8108,
+      "step": 2830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.402074536677312e-06,
+      "loss": 0.756,
+      "step": 2831
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.395980649655214e-06,
+      "loss": 0.7638,
+      "step": 2832
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.389886987751533e-06,
+      "loss": 0.785,
+      "step": 2833
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.383793553237376e-06,
+      "loss": 0.2511,
+      "step": 2834
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.377700348383772e-06,
+      "loss": 0.8349,
+      "step": 2835
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.371607375461661e-06,
+      "loss": 0.812,
+      "step": 2836
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.365514636741899e-06,
+      "loss": 0.7572,
+      "step": 2837
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.359422134495258e-06,
+      "loss": 0.8518,
+      "step": 2838
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.353329870992411e-06,
+      "loss": 0.7819,
+      "step": 2839
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.347237848503951e-06,
+      "loss": 0.7614,
+      "step": 2840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.341146069300379e-06,
+      "loss": 0.7619,
+      "step": 2841
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.335054535652105e-06,
+      "loss": 0.2613,
+      "step": 2842
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.328963249829451e-06,
+      "loss": 0.797,
+      "step": 2843
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.322872214102638e-06,
+      "loss": 0.7832,
+      "step": 2844
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.316781430741802e-06,
+      "loss": 0.8004,
+      "step": 2845
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.31069090201698e-06,
+      "loss": 0.8075,
+      "step": 2846
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.304600630198122e-06,
+      "loss": 0.8301,
+      "step": 2847
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.298510617555067e-06,
+      "loss": 0.8096,
+      "step": 2848
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.292420866357573e-06,
+      "loss": 0.3012,
+      "step": 2849
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.286331378875292e-06,
+      "loss": 0.759,
+      "step": 2850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.280242157377783e-06,
+      "loss": 0.7651,
+      "step": 2851
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.274153204134501e-06,
+      "loss": 0.8459,
+      "step": 2852
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.268064521414802e-06,
+      "loss": 0.836,
+      "step": 2853
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.261976111487942e-06,
+      "loss": 0.7272,
+      "step": 2854
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.255887976623074e-06,
+      "loss": 0.8146,
+      "step": 2855
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.249800119089257e-06,
+      "loss": 0.7822,
+      "step": 2856
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.243712541155437e-06,
+      "loss": 0.802,
+      "step": 2857
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.23762524509045e-06,
+      "loss": 0.7796,
+      "step": 2858
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.231538233163043e-06,
+      "loss": 0.8501,
+      "step": 2859
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.225451507641847e-06,
+      "loss": 0.798,
+      "step": 2860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.219365070795394e-06,
+      "loss": 0.797,
+      "step": 2861
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.213278924892097e-06,
+      "loss": 0.805,
+      "step": 2862
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.207193072200265e-06,
+      "loss": 0.7753,
+      "step": 2863
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.201107514988104e-06,
+      "loss": 0.8249,
+      "step": 2864
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.195022255523702e-06,
+      "loss": 0.7641,
+      "step": 2865
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.188937296075046e-06,
+      "loss": 0.7983,
+      "step": 2866
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.182852638909998e-06,
+      "loss": 0.2872,
+      "step": 2867
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.176768286296314e-06,
+      "loss": 0.7751,
+      "step": 2868
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.170684240501635e-06,
+      "loss": 0.8086,
+      "step": 2869
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.164600503793493e-06,
+      "loss": 0.7885,
+      "step": 2870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.158517078439303e-06,
+      "loss": 0.8081,
+      "step": 2871
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.152433966706354e-06,
+      "loss": 0.8474,
+      "step": 2872
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.14635117086183e-06,
+      "loss": 0.8002,
+      "step": 2873
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.140268693172792e-06,
+      "loss": 0.7386,
+      "step": 2874
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.134186535906189e-06,
+      "loss": 0.7616,
+      "step": 2875
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.128104701328834e-06,
+      "loss": 0.7551,
+      "step": 2876
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.122023191707439e-06,
+      "loss": 0.2733,
+      "step": 2877
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.115942009308586e-06,
+      "loss": 0.8241,
+      "step": 2878
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.109861156398732e-06,
+      "loss": 0.8198,
+      "step": 2879
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.103780635244222e-06,
+      "loss": 0.7457,
+      "step": 2880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.09770044811126e-06,
+      "loss": 0.8285,
+      "step": 2881
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.091620597265942e-06,
+      "loss": 0.8036,
+      "step": 2882
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.085541084974237e-06,
+      "loss": 0.7826,
+      "step": 2883
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.079461913501974e-06,
+      "loss": 0.7856,
+      "step": 2884
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.07338308511487e-06,
+      "loss": 0.8402,
+      "step": 2885
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.067304602078506e-06,
+      "loss": 0.7548,
+      "step": 2886
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.061226466658335e-06,
+      "loss": 0.759,
+      "step": 2887
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.055148681119688e-06,
+      "loss": 0.7507,
+      "step": 2888
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.049071247727757e-06,
+      "loss": 0.772,
+      "step": 2889
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.042994168747603e-06,
+      "loss": 0.8022,
+      "step": 2890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.036917446444158e-06,
+      "loss": 0.7646,
+      "step": 2891
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.030841083082222e-06,
+      "loss": 0.7871,
+      "step": 2892
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.024765080926461e-06,
+      "loss": 0.8081,
+      "step": 2893
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.018689442241403e-06,
+      "loss": 0.2652,
+      "step": 2894
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.012614169291443e-06,
+      "loss": 0.8267,
+      "step": 2895
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.006539264340837e-06,
+      "loss": 0.7692,
+      "step": 2896
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.000464729653708e-06,
+      "loss": 0.8022,
+      "step": 2897
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.994390567494044e-06,
+      "loss": 0.7874,
+      "step": 2898
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.98831678012568e-06,
+      "loss": 0.7922,
+      "step": 2899
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.982243369812326e-06,
+      "loss": 0.8159,
+      "step": 2900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.976170338817546e-06,
+      "loss": 0.7906,
+      "step": 2901
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.97009768940476e-06,
+      "loss": 0.7894,
+      "step": 2902
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.964025423837258e-06,
+      "loss": 0.7646,
+      "step": 2903
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.957953544378163e-06,
+      "loss": 0.7675,
+      "step": 2904
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.951882053290475e-06,
+      "loss": 0.8258,
+      "step": 2905
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.945810952837045e-06,
+      "loss": 0.7772,
+      "step": 2906
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.939740245280573e-06,
+      "loss": 0.7964,
+      "step": 2907
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.933669932883622e-06,
+      "loss": 0.7886,
+      "step": 2908
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.927600017908591e-06,
+      "loss": 0.7928,
+      "step": 2909
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.92153050261775e-06,
+      "loss": 0.8594,
+      "step": 2910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.915461389273206e-06,
+      "loss": 0.8128,
+      "step": 2911
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.909392680136929e-06,
+      "loss": 0.7979,
+      "step": 2912
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.903324377470721e-06,
+      "loss": 0.8271,
+      "step": 2913
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.897256483536246e-06,
+      "loss": 0.777,
+      "step": 2914
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.891189000595016e-06,
+      "loss": 0.8221,
+      "step": 2915
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.885121930908382e-06,
+      "loss": 0.7851,
+      "step": 2916
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.87905527673755e-06,
+      "loss": 0.7755,
+      "step": 2917
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.872989040343559e-06,
+      "loss": 0.2763,
+      "step": 2918
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.866923223987303e-06,
+      "loss": 0.8398,
+      "step": 2919
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.860857829929513e-06,
+      "loss": 0.773,
+      "step": 2920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.854792860430772e-06,
+      "loss": 0.8467,
+      "step": 2921
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.848728317751495e-06,
+      "loss": 0.7449,
+      "step": 2922
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.842664204151936e-06,
+      "loss": 0.7833,
+      "step": 2923
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8366005218922e-06,
+      "loss": 0.7789,
+      "step": 2924
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.83053727323222e-06,
+      "loss": 0.7463,
+      "step": 2925
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.824474460431783e-06,
+      "loss": 0.7926,
+      "step": 2926
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.818412085750493e-06,
+      "loss": 0.8067,
+      "step": 2927
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.812350151447806e-06,
+      "loss": 0.774,
+      "step": 2928
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.806288659783005e-06,
+      "loss": 0.7776,
+      "step": 2929
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.800227613015216e-06,
+      "loss": 0.8201,
+      "step": 2930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.794167013403398e-06,
+      "loss": 0.7722,
+      "step": 2931
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.788106863206333e-06,
+      "loss": 0.742,
+      "step": 2932
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.782047164682648e-06,
+      "loss": 0.7744,
+      "step": 2933
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.775987920090792e-06,
+      "loss": 0.75,
+      "step": 2934
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.769929131689054e-06,
+      "loss": 0.7801,
+      "step": 2935
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.763870801735552e-06,
+      "loss": 0.733,
+      "step": 2936
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.757812932488218e-06,
+      "loss": 0.7946,
+      "step": 2937
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.751755526204836e-06,
+      "loss": 0.7926,
+      "step": 2938
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.745698585142995e-06,
+      "loss": 0.8348,
+      "step": 2939
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.739642111560131e-06,
+      "loss": 0.8168,
+      "step": 2940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.733586107713487e-06,
+      "loss": 0.7742,
+      "step": 2941
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.727530575860143e-06,
+      "loss": 0.8491,
+      "step": 2942
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.721475518257003e-06,
+      "loss": 0.8408,
+      "step": 2943
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.715420937160786e-06,
+      "loss": 0.8009,
+      "step": 2944
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.709366834828045e-06,
+      "loss": 0.7962,
+      "step": 2945
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.703313213515138e-06,
+      "loss": 0.8066,
+      "step": 2946
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69726007547826e-06,
+      "loss": 0.8171,
+      "step": 2947
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69120742297342e-06,
+      "loss": 0.7978,
+      "step": 2948
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.685155258256444e-06,
+      "loss": 0.8016,
+      "step": 2949
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.67910358358298e-06,
+      "loss": 0.7904,
+      "step": 2950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.673052401208486e-06,
+      "loss": 0.7598,
+      "step": 2951
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.667001713388246e-06,
+      "loss": 0.8339,
+      "step": 2952
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.660951522377355e-06,
+      "loss": 0.8815,
+      "step": 2953
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.654901830430724e-06,
+      "loss": 0.802,
+      "step": 2954
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.648852639803075e-06,
+      "loss": 0.7693,
+      "step": 2955
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.642803952748944e-06,
+      "loss": 0.8141,
+      "step": 2956
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.636755771522684e-06,
+      "loss": 0.7929,
+      "step": 2957
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.630708098378454e-06,
+      "loss": 0.8425,
+      "step": 2958
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.624660935570233e-06,
+      "loss": 0.8082,
+      "step": 2959
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.618614285351792e-06,
+      "loss": 0.817,
+      "step": 2960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.612568149976726e-06,
+      "loss": 0.836,
+      "step": 2961
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.60652253169843e-06,
+      "loss": 0.7953,
+      "step": 2962
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.600477432770123e-06,
+      "loss": 0.8257,
+      "step": 2963
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.5944328554448e-06,
+      "loss": 0.7884,
+      "step": 2964
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.588388801975292e-06,
+      "loss": 0.7714,
+      "step": 2965
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.582345274614213e-06,
+      "loss": 0.7927,
+      "step": 2966
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.576302275613994e-06,
+      "loss": 0.7525,
+      "step": 2967
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.570259807226869e-06,
+      "loss": 0.8232,
+      "step": 2968
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.56421787170486e-06,
+      "loss": 0.7065,
+      "step": 2969
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.558176471299804e-06,
+      "loss": 0.8372,
+      "step": 2970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.552135608263343e-06,
+      "loss": 0.8129,
+      "step": 2971
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.546095284846898e-06,
+      "loss": 0.2718,
+      "step": 2972
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.540055503301714e-06,
+      "loss": 0.771,
+      "step": 2973
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.534016265878811e-06,
+      "loss": 0.7754,
+      "step": 2974
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.527977574829022e-06,
+      "loss": 0.8,
+      "step": 2975
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.521939432402971e-06,
+      "loss": 0.8529,
+      "step": 2976
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.515901840851081e-06,
+      "loss": 0.7512,
+      "step": 2977
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.509864802423557e-06,
+      "loss": 0.8092,
+      "step": 2978
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.503828319370413e-06,
+      "loss": 0.7638,
+      "step": 2979
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.49779239394145e-06,
+      "loss": 0.8197,
+      "step": 2980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.491757028386262e-06,
+      "loss": 0.8046,
+      "step": 2981
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.485722224954237e-06,
+      "loss": 0.7498,
+      "step": 2982
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.479687985894543e-06,
+      "loss": 0.7615,
+      "step": 2983
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.473654313456145e-06,
+      "loss": 0.769,
+      "step": 2984
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.467621209887803e-06,
+      "loss": 0.7335,
+      "step": 2985
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.461588677438057e-06,
+      "loss": 0.7788,
+      "step": 2986
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.455556718355234e-06,
+      "loss": 0.8127,
+      "step": 2987
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.44952533488745e-06,
+      "loss": 0.8158,
+      "step": 2988
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.443494529282605e-06,
+      "loss": 0.8307,
+      "step": 2989
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.437464303788383e-06,
+      "loss": 0.8207,
+      "step": 2990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.43143466065226e-06,
+      "loss": 0.814,
+      "step": 2991
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.425405602121478e-06,
+      "loss": 0.8221,
+      "step": 2992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.419377130443077e-06,
+      "loss": 0.7592,
+      "step": 2993
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.41334924786387e-06,
+      "loss": 0.8119,
+      "step": 2994
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.407321956630455e-06,
+      "loss": 0.7649,
+      "step": 2995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.401295258989209e-06,
+      "loss": 0.7996,
+      "step": 2996
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.39526915718628e-06,
+      "loss": 0.7671,
+      "step": 2997
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.389243653467606e-06,
+      "loss": 0.8059,
+      "step": 2998
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.383218750078894e-06,
+      "loss": 0.8036,
+      "step": 2999
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.377194449265628e-06,
+      "loss": 0.8257,
+      "step": 3000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.371170753273075e-06,
+      "loss": 0.7624,
+      "step": 3001
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.365147664346261e-06,
+      "loss": 0.7151,
+      "step": 3002
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.359125184730004e-06,
+      "loss": 0.8536,
+      "step": 3003
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.353103316668881e-06,
+      "loss": 0.7905,
+      "step": 3004
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.34708206240725e-06,
+      "loss": 0.8167,
+      "step": 3005
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.34106142418923e-06,
+      "loss": 0.8083,
+      "step": 3006
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.335041404258723e-06,
+      "loss": 0.7431,
+      "step": 3007
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.329022004859394e-06,
+      "loss": 0.7887,
+      "step": 3008
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.323003228234672e-06,
+      "loss": 0.8209,
+      "step": 3009
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.31698507662777e-06,
+      "loss": 0.7567,
+      "step": 3010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.310967552281643e-06,
+      "loss": 0.261,
+      "step": 3011
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.304950657439034e-06,
+      "loss": 0.7806,
+      "step": 3012
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.298934394342442e-06,
+      "loss": 0.7451,
+      "step": 3013
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.29291876523414e-06,
+      "loss": 0.7915,
+      "step": 3014
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.286903772356146e-06,
+      "loss": 0.2338,
+      "step": 3015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.280889417950255e-06,
+      "loss": 0.7398,
+      "step": 3016
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.274875704258022e-06,
+      "loss": 0.7314,
+      "step": 3017
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.268862633520758e-06,
+      "loss": 0.8245,
+      "step": 3018
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.262850207979552e-06,
+      "loss": 0.7538,
+      "step": 3019
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.256838429875225e-06,
+      "loss": 0.7886,
+      "step": 3020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.250827301448371e-06,
+      "loss": 0.7477,
+      "step": 3021
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.24481682493935e-06,
+      "loss": 0.7744,
+      "step": 3022
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.238807002588262e-06,
+      "loss": 0.2667,
+      "step": 3023
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.232797836634983e-06,
+      "loss": 0.786,
+      "step": 3024
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.226789329319124e-06,
+      "loss": 0.7731,
+      "step": 3025
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.220781482880058e-06,
+      "loss": 0.7755,
+      "step": 3026
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.21477429955692e-06,
+      "loss": 0.781,
+      "step": 3027
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.208767781588591e-06,
+      "loss": 0.7874,
+      "step": 3028
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.202761931213698e-06,
+      "loss": 0.8092,
+      "step": 3029
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.196756750670629e-06,
+      "loss": 0.7917,
+      "step": 3030
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.190752242197521e-06,
+      "loss": 0.7796,
+      "step": 3031
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.184748408032253e-06,
+      "loss": 0.7083,
+      "step": 3032
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.178745250412468e-06,
+      "loss": 0.7593,
+      "step": 3033
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.172742771575534e-06,
+      "loss": 0.7803,
+      "step": 3034
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.166740973758585e-06,
+      "loss": 0.8227,
+      "step": 3035
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.160739859198498e-06,
+      "loss": 0.7629,
+      "step": 3036
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.154739430131886e-06,
+      "loss": 0.7935,
+      "step": 3037
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.148739688795121e-06,
+      "loss": 0.7937,
+      "step": 3038
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.142740637424302e-06,
+      "loss": 0.7464,
+      "step": 3039
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.136742278255285e-06,
+      "loss": 0.8374,
+      "step": 3040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13074461352366e-06,
+      "loss": 0.8115,
+      "step": 3041
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.124747645464763e-06,
+      "loss": 0.7926,
+      "step": 3042
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.118751376313666e-06,
+      "loss": 0.7825,
+      "step": 3043
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.11275580830518e-06,
+      "loss": 0.7977,
+      "step": 3044
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.10676094367386e-06,
+      "loss": 0.805,
+      "step": 3045
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.100766784654e-06,
+      "loss": 0.2721,
+      "step": 3046
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.094773333479621e-06,
+      "loss": 0.8115,
+      "step": 3047
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.08878059238449e-06,
+      "loss": 0.2931,
+      "step": 3048
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.082788563602102e-06,
+      "loss": 0.791,
+      "step": 3049
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.07679724936569e-06,
+      "loss": 0.7466,
+      "step": 3050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.070806651908227e-06,
+      "loss": 0.2461,
+      "step": 3051
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.064816773462408e-06,
+      "loss": 0.7671,
+      "step": 3052
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.058827616260664e-06,
+      "loss": 0.7853,
+      "step": 3053
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.05283918253516e-06,
+      "loss": 0.8266,
+      "step": 3054
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.046851474517786e-06,
+      "loss": 0.8353,
+      "step": 3055
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.040864494440173e-06,
+      "loss": 0.8628,
+      "step": 3056
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.03487824453366e-06,
+      "loss": 0.7881,
+      "step": 3057
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.028892727029338e-06,
+      "loss": 0.7818,
+      "step": 3058
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.022907944158002e-06,
+      "loss": 0.7773,
+      "step": 3059
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.016923898150195e-06,
+      "loss": 0.7349,
+      "step": 3060
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.010940591236172e-06,
+      "loss": 0.2971,
+      "step": 3061
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.004958025645912e-06,
+      "loss": 0.8292,
+      "step": 3062
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.998976203609124e-06,
+      "loss": 0.7749,
+      "step": 3063
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.992995127355236e-06,
+      "loss": 0.802,
+      "step": 3064
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.987014799113398e-06,
+      "loss": 0.2992,
+      "step": 3065
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98103522111249e-06,
+      "loss": 0.795,
+      "step": 3066
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.975056395581092e-06,
+      "loss": 0.7811,
+      "step": 3067
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.969078324747527e-06,
+      "loss": 0.783,
+      "step": 3068
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.963101010839817e-06,
+      "loss": 0.798,
+      "step": 3069
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.957124456085723e-06,
+      "loss": 0.8366,
+      "step": 3070
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.951148662712698e-06,
+      "loss": 0.8393,
+      "step": 3071
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.945173632947929e-06,
+      "loss": 0.8541,
+      "step": 3072
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.939199369018315e-06,
+      "loss": 0.759,
+      "step": 3073
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93322587315047e-06,
+      "loss": 0.2826,
+      "step": 3074
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.92725314757072e-06,
+      "loss": 0.759,
+      "step": 3075
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.921281194505097e-06,
+      "loss": 0.7503,
+      "step": 3076
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.915310016179353e-06,
+      "loss": 0.8244,
+      "step": 3077
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.909339614818955e-06,
+      "loss": 0.7771,
+      "step": 3078
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.903369992649076e-06,
+      "loss": 0.2607,
+      "step": 3079
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.8974011518946e-06,
+      "loss": 0.8335,
+      "step": 3080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.891433094780108e-06,
+      "loss": 0.7898,
+      "step": 3081
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.885465823529902e-06,
+      "loss": 0.8253,
+      "step": 3082
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.879499340367995e-06,
+      "loss": 0.2529,
+      "step": 3083
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.873533647518096e-06,
+      "loss": 0.7992,
+      "step": 3084
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.86756874720362e-06,
+      "loss": 0.7288,
+      "step": 3085
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.861604641647692e-06,
+      "loss": 0.8149,
+      "step": 3086
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.855641333073133e-06,
+      "loss": 0.7771,
+      "step": 3087
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.849678823702479e-06,
+      "loss": 0.8052,
+      "step": 3088
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.84371711575796e-06,
+      "loss": 0.8055,
+      "step": 3089
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.837756211461504e-06,
+      "loss": 0.279,
+      "step": 3090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.831796113034747e-06,
+      "loss": 0.8123,
+      "step": 3091
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.825836822699021e-06,
+      "loss": 0.8429,
+      "step": 3092
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.819878342675362e-06,
+      "loss": 0.7782,
+      "step": 3093
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.813920675184492e-06,
+      "loss": 0.834,
+      "step": 3094
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.80796382244684e-06,
+      "loss": 0.7369,
+      "step": 3095
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.802007786682532e-06,
+      "loss": 0.8018,
+      "step": 3096
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.796052570111383e-06,
+      "loss": 0.7414,
+      "step": 3097
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.790098174952912e-06,
+      "loss": 0.7613,
+      "step": 3098
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.784144603426317e-06,
+      "loss": 0.7844,
+      "step": 3099
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.778191857750502e-06,
+      "loss": 0.7293,
+      "step": 3100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.772239940144064e-06,
+      "loss": 0.7669,
+      "step": 3101
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.766288852825278e-06,
+      "loss": 0.7596,
+      "step": 3102
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.76033859801213e-06,
+      "loss": 0.7774,
+      "step": 3103
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.75438917792227e-06,
+      "loss": 0.2789,
+      "step": 3104
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.74844059477306e-06,
+      "loss": 0.7498,
+      "step": 3105
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.74249285078154e-06,
+      "loss": 0.8132,
+      "step": 3106
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.736545948164437e-06,
+      "loss": 0.7454,
+      "step": 3107
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.730599889138165e-06,
+      "loss": 0.7897,
+      "step": 3108
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.72465467591882e-06,
+      "loss": 0.7254,
+      "step": 3109
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.718710310722195e-06,
+      "loss": 0.7914,
+      "step": 3110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.712766795763756e-06,
+      "loss": 0.8456,
+      "step": 3111
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.706824133258655e-06,
+      "loss": 0.7807,
+      "step": 3112
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.700882325421724e-06,
+      "loss": 0.8354,
+      "step": 3113
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.694941374467478e-06,
+      "loss": 0.7792,
+      "step": 3114
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.689001282610115e-06,
+      "loss": 0.7887,
+      "step": 3115
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.683062052063514e-06,
+      "loss": 0.8166,
+      "step": 3116
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.677123685041231e-06,
+      "loss": 0.8207,
+      "step": 3117
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.671186183756495e-06,
+      "loss": 0.2563,
+      "step": 3118
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.665249550422214e-06,
+      "loss": 0.8242,
+      "step": 3119
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.659313787250982e-06,
+      "loss": 0.269,
+      "step": 3120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.65337889645506e-06,
+      "loss": 0.7751,
+      "step": 3121
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.647444880246382e-06,
+      "loss": 0.7577,
+      "step": 3122
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.641511740836566e-06,
+      "loss": 0.7662,
+      "step": 3123
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.63557948043689e-06,
+      "loss": 0.865,
+      "step": 3124
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.629648101258315e-06,
+      "loss": 0.8088,
+      "step": 3125
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.623717605511475e-06,
+      "loss": 0.8529,
+      "step": 3126
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.61778799540666e-06,
+      "loss": 0.7954,
+      "step": 3127
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.611859273153843e-06,
+      "loss": 0.8141,
+      "step": 3128
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.605931440962668e-06,
+      "loss": 0.8108,
+      "step": 3129
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.600004501042435e-06,
+      "loss": 0.7607,
+      "step": 3130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.594078455602126e-06,
+      "loss": 0.865,
+      "step": 3131
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.588153306850374e-06,
+      "loss": 0.7656,
+      "step": 3132
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5822290569954886e-06,
+      "loss": 0.784,
+      "step": 3133
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.576305708245445e-06,
+      "loss": 0.8168,
+      "step": 3134
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.570383262807879e-06,
+      "loss": 0.7863,
+      "step": 3135
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.564461722890082e-06,
+      "loss": 0.8391,
+      "step": 3136
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.558541090699023e-06,
+      "loss": 0.8474,
+      "step": 3137
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.552621368441324e-06,
+      "loss": 0.7385,
+      "step": 3138
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.546702558323271e-06,
+      "loss": 0.753,
+      "step": 3139
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.54078466255081e-06,
+      "loss": 0.7549,
+      "step": 3140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.534867683329537e-06,
+      "loss": 0.8199,
+      "step": 3141
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5289516228647185e-06,
+      "loss": 0.7975,
+      "step": 3142
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.523036483361275e-06,
+      "loss": 0.8066,
+      "step": 3143
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.517122267023784e-06,
+      "loss": 0.7534,
+      "step": 3144
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5112089760564765e-06,
+      "loss": 0.8204,
+      "step": 3145
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5052966126632394e-06,
+      "loss": 0.7873,
+      "step": 3146
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.499385179047614e-06,
+      "loss": 0.6949,
+      "step": 3147
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.493474677412795e-06,
+      "loss": 0.8049,
+      "step": 3148
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.487565109961635e-06,
+      "loss": 0.7757,
+      "step": 3149
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.481656478896628e-06,
+      "loss": 0.8277,
+      "step": 3150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.475748786419928e-06,
+      "loss": 0.7968,
+      "step": 3151
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.469842034733333e-06,
+      "loss": 0.7739,
+      "step": 3152
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.463936226038297e-06,
+      "loss": 0.7335,
+      "step": 3153
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.458031362535919e-06,
+      "loss": 0.807,
+      "step": 3154
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.45212744642694e-06,
+      "loss": 0.7231,
+      "step": 3155
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.446224479911761e-06,
+      "loss": 0.7898,
+      "step": 3156
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.440322465190414e-06,
+      "loss": 0.7895,
+      "step": 3157
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4344214044625914e-06,
+      "loss": 0.8196,
+      "step": 3158
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.428521299927616e-06,
+      "loss": 0.8098,
+      "step": 3159
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.422622153784461e-06,
+      "loss": 0.7914,
+      "step": 3160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.416723968231748e-06,
+      "loss": 0.8247,
+      "step": 3161
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.410826745467728e-06,
+      "loss": 0.7666,
+      "step": 3162
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.404930487690308e-06,
+      "loss": 0.7628,
+      "step": 3163
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.399035197097018e-06,
+      "loss": 0.774,
+      "step": 3164
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.39314087588504e-06,
+      "loss": 0.7903,
+      "step": 3165
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.387247526251194e-06,
+      "loss": 0.7882,
+      "step": 3166
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3813551503919335e-06,
+      "loss": 0.8078,
+      "step": 3167
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3754637505033535e-06,
+      "loss": 0.8021,
+      "step": 3168
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.369573328781176e-06,
+      "loss": 0.82,
+      "step": 3169
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3636838874207695e-06,
+      "loss": 0.8218,
+      "step": 3170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.357795428617131e-06,
+      "loss": 0.7624,
+      "step": 3171
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.351907954564901e-06,
+      "loss": 0.7494,
+      "step": 3172
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.346021467458335e-06,
+      "loss": 0.8369,
+      "step": 3173
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.340135969491333e-06,
+      "loss": 0.789,
+      "step": 3174
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.334251462857425e-06,
+      "loss": 0.7767,
+      "step": 3175
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.328367949749769e-06,
+      "loss": 0.2846,
+      "step": 3176
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.322485432361162e-06,
+      "loss": 0.7744,
+      "step": 3177
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.316603912884013e-06,
+      "loss": 0.2751,
+      "step": 3178
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3107233935103685e-06,
+      "loss": 0.7785,
+      "step": 3179
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.304843876431907e-06,
+      "loss": 0.2545,
+      "step": 3180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.298965363839923e-06,
+      "loss": 0.7663,
+      "step": 3181
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.293087857925351e-06,
+      "loss": 0.8233,
+      "step": 3182
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.287211360878734e-06,
+      "loss": 0.8259,
+      "step": 3183
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2813358748902456e-06,
+      "loss": 0.7689,
+      "step": 3184
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.275461402149686e-06,
+      "loss": 0.7798,
+      "step": 3185
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.269587944846479e-06,
+      "loss": 0.7832,
+      "step": 3186
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.263715505169656e-06,
+      "loss": 0.7661,
+      "step": 3187
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.257844085307887e-06,
+      "loss": 0.8074,
+      "step": 3188
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.251973687449454e-06,
+      "loss": 0.7948,
+      "step": 3189
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2461043137822525e-06,
+      "loss": 0.7671,
+      "step": 3190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.240235966493812e-06,
+      "loss": 0.759,
+      "step": 3191
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.234368647771257e-06,
+      "loss": 0.798,
+      "step": 3192
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.228502359801349e-06,
+      "loss": 0.7826,
+      "step": 3193
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.222637104770458e-06,
+      "loss": 0.8448,
+      "step": 3194
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.216772884864563e-06,
+      "loss": 0.8187,
+      "step": 3195
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.210909702269271e-06,
+      "loss": 0.7179,
+      "step": 3196
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.205047559169783e-06,
+      "loss": 0.711,
+      "step": 3197
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.199186457750931e-06,
+      "loss": 0.763,
+      "step": 3198
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1933264001971514e-06,
+      "loss": 0.7598,
+      "step": 3199
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.187467388692491e-06,
+      "loss": 0.7957,
+      "step": 3200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.181609425420606e-06,
+      "loss": 0.7387,
+      "step": 3201
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.17575251256476e-06,
+      "loss": 0.7547,
+      "step": 3202
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.169896652307833e-06,
+      "loss": 0.7899,
+      "step": 3203
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1640418468323106e-06,
+      "loss": 0.8078,
+      "step": 3204
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.158188098320276e-06,
+      "loss": 0.7569,
+      "step": 3205
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.15233540895343e-06,
+      "loss": 0.7691,
+      "step": 3206
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1464837809130695e-06,
+      "loss": 0.7939,
+      "step": 3207
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.140633216380103e-06,
+      "loss": 0.7751,
+      "step": 3208
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.134783717535041e-06,
+      "loss": 0.8056,
+      "step": 3209
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.128935286557994e-06,
+      "loss": 0.7516,
+      "step": 3210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.123087925628677e-06,
+      "loss": 0.8009,
+      "step": 3211
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.117241636926402e-06,
+      "loss": 0.7521,
+      "step": 3212
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.111396422630087e-06,
+      "loss": 0.7855,
+      "step": 3213
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.105552284918251e-06,
+      "loss": 0.8189,
+      "step": 3214
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.099709225969001e-06,
+      "loss": 0.8053,
+      "step": 3215
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.093867247960055e-06,
+      "loss": 0.7277,
+      "step": 3216
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.088026353068718e-06,
+      "loss": 0.6895,
+      "step": 3217
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.082186543471894e-06,
+      "loss": 0.7928,
+      "step": 3218
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0763478213460935e-06,
+      "loss": 0.2608,
+      "step": 3219
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.070510188867401e-06,
+      "loss": 0.7889,
+      "step": 3220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.064673648211512e-06,
+      "loss": 0.8242,
+      "step": 3221
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.058838201553706e-06,
+      "loss": 0.8095,
+      "step": 3222
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.053003851068863e-06,
+      "loss": 0.8295,
+      "step": 3223
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.047170598931441e-06,
+      "loss": 0.8434,
+      "step": 3224
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.041338447315502e-06,
+      "loss": 0.8331,
+      "step": 3225
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.035507398394697e-06,
+      "loss": 0.7407,
+      "step": 3226
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.029677454342252e-06,
+      "loss": 0.7654,
+      "step": 3227
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.023848617331002e-06,
+      "loss": 0.7508,
+      "step": 3228
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.018020889533348e-06,
+      "loss": 0.7823,
+      "step": 3229
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.012194273121294e-06,
+      "loss": 0.793,
+      "step": 3230
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.006368770266421e-06,
+      "loss": 0.7701,
+      "step": 3231
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.000544383139901e-06,
+      "loss": 0.8105,
+      "step": 3232
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.994721113912488e-06,
+      "loss": 0.8073,
+      "step": 3233
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.988898964754512e-06,
+      "loss": 0.7483,
+      "step": 3234
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.983077937835896e-06,
+      "loss": 0.8009,
+      "step": 3235
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.977258035326139e-06,
+      "loss": 0.2673,
+      "step": 3236
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.971439259394328e-06,
+      "loss": 0.7901,
+      "step": 3237
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.965621612209119e-06,
+      "loss": 0.826,
+      "step": 3238
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.959805095938752e-06,
+      "loss": 0.8001,
+      "step": 3239
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.953989712751048e-06,
+      "loss": 0.7762,
+      "step": 3240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.948175464813405e-06,
+      "loss": 0.7396,
+      "step": 3241
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.942362354292802e-06,
+      "loss": 0.7967,
+      "step": 3242
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.936550383355779e-06,
+      "loss": 0.7705,
+      "step": 3243
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9307395541684695e-06,
+      "loss": 0.777,
+      "step": 3244
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.924929868896568e-06,
+      "loss": 0.7964,
+      "step": 3245
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.919121329705352e-06,
+      "loss": 0.7902,
+      "step": 3246
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.913313938759669e-06,
+      "loss": 0.2827,
+      "step": 3247
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.907507698223931e-06,
+      "loss": 0.8269,
+      "step": 3248
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.901702610262133e-06,
+      "loss": 0.7853,
+      "step": 3249
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.895898677037833e-06,
+      "loss": 0.7897,
+      "step": 3250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.890095900714167e-06,
+      "loss": 0.8008,
+      "step": 3251
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8842942834538225e-06,
+      "loss": 0.7898,
+      "step": 3252
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.878493827419071e-06,
+      "loss": 0.7969,
+      "step": 3253
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.872694534771751e-06,
+      "loss": 0.8363,
+      "step": 3254
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.866896407673256e-06,
+      "loss": 0.2589,
+      "step": 3255
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.861099448284558e-06,
+      "loss": 0.8303,
+      "step": 3256
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.855303658766182e-06,
+      "loss": 0.7424,
+      "step": 3257
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8495090412782235e-06,
+      "loss": 0.81,
+      "step": 3258
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.843715597980345e-06,
+      "loss": 0.7912,
+      "step": 3259
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.837923331031761e-06,
+      "loss": 0.825,
+      "step": 3260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.832132242591259e-06,
+      "loss": 0.7996,
+      "step": 3261
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.826342334817176e-06,
+      "loss": 0.8276,
+      "step": 3262
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.820553609867413e-06,
+      "loss": 0.837,
+      "step": 3263
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.81476606989944e-06,
+      "loss": 0.2556,
+      "step": 3264
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.808979717070272e-06,
+      "loss": 0.7682,
+      "step": 3265
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.803194553536487e-06,
+      "loss": 0.807,
+      "step": 3266
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.797410581454216e-06,
+      "loss": 0.8139,
+      "step": 3267
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7916278029791525e-06,
+      "loss": 0.7994,
+      "step": 3268
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.785846220266545e-06,
+      "loss": 0.7783,
+      "step": 3269
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.78006583547119e-06,
+      "loss": 0.7887,
+      "step": 3270
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.774286650747439e-06,
+      "loss": 0.7612,
+      "step": 3271
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7685086682492e-06,
+      "loss": 0.734,
+      "step": 3272
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.76273189012993e-06,
+      "loss": 0.8095,
+      "step": 3273
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.756956318542638e-06,
+      "loss": 0.7736,
+      "step": 3274
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.751181955639887e-06,
+      "loss": 0.8211,
+      "step": 3275
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.745408803573783e-06,
+      "loss": 0.7796,
+      "step": 3276
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.739636864495981e-06,
+      "loss": 0.8175,
+      "step": 3277
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.733866140557689e-06,
+      "loss": 0.7925,
+      "step": 3278
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.728096633909664e-06,
+      "loss": 0.6721,
+      "step": 3279
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.722328346702196e-06,
+      "loss": 0.7611,
+      "step": 3280
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.716561281085137e-06,
+      "loss": 0.807,
+      "step": 3281
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7107954392078715e-06,
+      "loss": 0.8044,
+      "step": 3282
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.705030823219334e-06,
+      "loss": 0.7929,
+      "step": 3283
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.699267435268006e-06,
+      "loss": 0.8316,
+      "step": 3284
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6935052775018974e-06,
+      "loss": 0.8162,
+      "step": 3285
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.68774435206857e-06,
+      "loss": 0.7386,
+      "step": 3286
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.68198466111513e-06,
+      "loss": 0.7608,
+      "step": 3287
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.676226206788213e-06,
+      "loss": 0.7251,
+      "step": 3288
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6704689912340045e-06,
+      "loss": 0.7365,
+      "step": 3289
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.664713016598214e-06,
+      "loss": 0.7639,
+      "step": 3290
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6589582850261025e-06,
+      "loss": 0.7443,
+      "step": 3291
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.653204798662463e-06,
+      "loss": 0.797,
+      "step": 3292
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.647452559651624e-06,
+      "loss": 0.8127,
+      "step": 3293
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.641701570137442e-06,
+      "loss": 0.8168,
+      "step": 3294
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.635951832263321e-06,
+      "loss": 0.7972,
+      "step": 3295
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6302033481721894e-06,
+      "loss": 0.8039,
+      "step": 3296
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.624456120006515e-06,
+      "loss": 0.8092,
+      "step": 3297
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6187101499082905e-06,
+      "loss": 0.8456,
+      "step": 3298
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.612965440019044e-06,
+      "loss": 0.8283,
+      "step": 3299
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.607221992479828e-06,
+      "loss": 0.7178,
+      "step": 3300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.601479809431235e-06,
+      "loss": 0.8038,
+      "step": 3301
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.595738893013381e-06,
+      "loss": 0.7703,
+      "step": 3302
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.589999245365903e-06,
+      "loss": 0.8366,
+      "step": 3303
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.584260868627979e-06,
+      "loss": 0.2762,
+      "step": 3304
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.578523764938298e-06,
+      "loss": 0.2675,
+      "step": 3305
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.572787936435088e-06,
+      "loss": 0.8217,
+      "step": 3306
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.567053385256098e-06,
+      "loss": 0.7666,
+      "step": 3307
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.561320113538592e-06,
+      "loss": 0.8166,
+      "step": 3308
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.55558812341937e-06,
+      "loss": 0.8037,
+      "step": 3309
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.549857417034744e-06,
+      "loss": 0.805,
+      "step": 3310
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.544127996520555e-06,
+      "loss": 0.8011,
+      "step": 3311
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5383998640121645e-06,
+      "loss": 0.7684,
+      "step": 3312
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.532673021644444e-06,
+      "loss": 0.7393,
+      "step": 3313
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.526947471551799e-06,
+      "loss": 0.7882,
+      "step": 3314
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.52122321586814e-06,
+      "loss": 0.7651,
+      "step": 3315
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5155002567269065e-06,
+      "loss": 0.8263,
+      "step": 3316
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.509778596261043e-06,
+      "loss": 0.7495,
+      "step": 3317
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.504058236603018e-06,
+      "loss": 0.8132,
+      "step": 3318
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.498339179884816e-06,
+      "loss": 0.2347,
+      "step": 3319
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.492621428237932e-06,
+      "loss": 0.833,
+      "step": 3320
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.486904983793378e-06,
+      "loss": 0.2925,
+      "step": 3321
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.48118984868167e-06,
+      "loss": 0.7892,
+      "step": 3322
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.475476025032849e-06,
+      "loss": 0.8018,
+      "step": 3323
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4697635149764594e-06,
+      "loss": 0.7479,
+      "step": 3324
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.464052320641554e-06,
+      "loss": 0.2708,
+      "step": 3325
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.458342444156707e-06,
+      "loss": 0.7879,
+      "step": 3326
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.452633887649983e-06,
+      "loss": 0.8185,
+      "step": 3327
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.446926653248969e-06,
+      "loss": 0.7062,
+      "step": 3328
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.441220743080756e-06,
+      "loss": 0.7871,
+      "step": 3329
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.435516159271946e-06,
+      "loss": 0.2532,
+      "step": 3330
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.429812903948631e-06,
+      "loss": 0.7647,
+      "step": 3331
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.424110979236422e-06,
+      "loss": 0.7874,
+      "step": 3332
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4184103872604295e-06,
+      "loss": 0.7535,
+      "step": 3333
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4127111301452695e-06,
+      "loss": 0.7411,
+      "step": 3334
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.407013210015063e-06,
+      "loss": 0.7913,
+      "step": 3335
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.40131662899342e-06,
+      "loss": 0.7745,
+      "step": 3336
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.395621389203464e-06,
+      "loss": 0.829,
+      "step": 3337
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.389927492767813e-06,
+      "loss": 0.7657,
+      "step": 3338
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.384234941808589e-06,
+      "loss": 0.81,
+      "step": 3339
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.378543738447411e-06,
+      "loss": 0.7378,
+      "step": 3340
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.372853884805389e-06,
+      "loss": 0.7853,
+      "step": 3341
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.367165383003139e-06,
+      "loss": 0.7969,
+      "step": 3342
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.361478235160766e-06,
+      "loss": 0.7303,
+      "step": 3343
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.35579244339788e-06,
+      "loss": 0.7832,
+      "step": 3344
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.350108009833571e-06,
+      "loss": 0.8107,
+      "step": 3345
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.344424936586436e-06,
+      "loss": 0.7166,
+      "step": 3346
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.338743225774562e-06,
+      "loss": 0.7148,
+      "step": 3347
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.333062879515521e-06,
+      "loss": 0.7306,
+      "step": 3348
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.327383899926389e-06,
+      "loss": 0.8047,
+      "step": 3349
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.321706289123719e-06,
+      "loss": 0.7314,
+      "step": 3350
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.31603004922356e-06,
+      "loss": 0.7437,
+      "step": 3351
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.310355182341458e-06,
+      "loss": 0.8255,
+      "step": 3352
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.304681690592431e-06,
+      "loss": 0.7653,
+      "step": 3353
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.299009576091001e-06,
+      "loss": 0.7801,
+      "step": 3354
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.293338840951161e-06,
+      "loss": 0.7999,
+      "step": 3355
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2876694872864e-06,
+      "loss": 0.7531,
+      "step": 3356
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2820015172096945e-06,
+      "loss": 0.7274,
+      "step": 3357
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.276334932833499e-06,
+      "loss": 0.7937,
+      "step": 3358
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.27066973626975e-06,
+      "loss": 0.832,
+      "step": 3359
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.265005929629871e-06,
+      "loss": 0.7814,
+      "step": 3360
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.259343515024767e-06,
+      "loss": 0.8006,
+      "step": 3361
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2536824945648255e-06,
+      "loss": 0.7613,
+      "step": 3362
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.248022870359912e-06,
+      "loss": 0.7478,
+      "step": 3363
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.242364644519371e-06,
+      "loss": 0.789,
+      "step": 3364
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.236707819152026e-06,
+      "loss": 0.2473,
+      "step": 3365
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.231052396366179e-06,
+      "loss": 0.2652,
+      "step": 3366
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.225398378269618e-06,
+      "loss": 0.8027,
+      "step": 3367
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2197457669695894e-06,
+      "loss": 0.7712,
+      "step": 3368
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.214094564572829e-06,
+      "loss": 0.7947,
+      "step": 3369
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.208444773185543e-06,
+      "loss": 0.7741,
+      "step": 3370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.202796394913413e-06,
+      "loss": 0.7846,
+      "step": 3371
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.197149431861597e-06,
+      "loss": 0.8033,
+      "step": 3372
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.191503886134713e-06,
+      "loss": 0.8125,
+      "step": 3373
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.185859759836868e-06,
+      "loss": 0.819,
+      "step": 3374
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.180217055071627e-06,
+      "loss": 0.7491,
+      "step": 3375
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1745757739420305e-06,
+      "loss": 0.7722,
+      "step": 3376
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.168935918550594e-06,
+      "loss": 0.8034,
+      "step": 3377
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1632974909992845e-06,
+      "loss": 0.7384,
+      "step": 3378
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1576604933895565e-06,
+      "loss": 0.7209,
+      "step": 3379
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1520249278223166e-06,
+      "loss": 0.8132,
+      "step": 3380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.146390796397951e-06,
+      "loss": 0.8005,
+      "step": 3381
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.140758101216295e-06,
+      "loss": 0.7489,
+      "step": 3382
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.135126844376663e-06,
+      "loss": 0.7794,
+      "step": 3383
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.129497027977829e-06,
+      "loss": 0.7936,
+      "step": 3384
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.123868654118028e-06,
+      "loss": 0.7699,
+      "step": 3385
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1182417248949635e-06,
+      "loss": 0.2641,
+      "step": 3386
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1126162424057855e-06,
+      "loss": 0.7753,
+      "step": 3387
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.106992208747121e-06,
+      "loss": 0.7954,
+      "step": 3388
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.10136962601505e-06,
+      "loss": 0.2532,
+      "step": 3389
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.095748496305119e-06,
+      "loss": 0.7572,
+      "step": 3390
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.09012882171232e-06,
+      "loss": 0.8018,
+      "step": 3391
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.084510604331108e-06,
+      "loss": 0.7726,
+      "step": 3392
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0788938462553985e-06,
+      "loss": 0.7882,
+      "step": 3393
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.073278549578563e-06,
+      "loss": 0.8107,
+      "step": 3394
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.067664716393428e-06,
+      "loss": 0.7788,
+      "step": 3395
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0620523487922675e-06,
+      "loss": 0.7529,
+      "step": 3396
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.056441448866817e-06,
+      "loss": 0.7973,
+      "step": 3397
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.050832018708261e-06,
+      "loss": 0.7789,
+      "step": 3398
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.04522406040724e-06,
+      "loss": 0.7713,
+      "step": 3399
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.039617576053847e-06,
+      "loss": 0.7821,
+      "step": 3400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.034012567737615e-06,
+      "loss": 0.7793,
+      "step": 3401
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.028409037547539e-06,
+      "loss": 0.7952,
+      "step": 3402
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.022806987572057e-06,
+      "loss": 0.8475,
+      "step": 3403
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.017206419899056e-06,
+      "loss": 0.7695,
+      "step": 3404
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.011607336615878e-06,
+      "loss": 0.7567,
+      "step": 3405
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0060097398092955e-06,
+      "loss": 0.7264,
+      "step": 3406
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.000413631565541e-06,
+      "loss": 0.7799,
+      "step": 3407
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.994819013970286e-06,
+      "loss": 0.7467,
+      "step": 3408
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.989225889108655e-06,
+      "loss": 0.7444,
+      "step": 3409
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.983634259065198e-06,
+      "loss": 0.8083,
+      "step": 3410
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9780441259239275e-06,
+      "loss": 0.7962,
+      "step": 3411
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.972455491768289e-06,
+      "loss": 0.7624,
+      "step": 3412
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.966868358681166e-06,
+      "loss": 0.7997,
+      "step": 3413
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9612827287448935e-06,
+      "loss": 0.8094,
+      "step": 3414
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.955698604041231e-06,
+      "loss": 0.7563,
+      "step": 3415
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.950115986651391e-06,
+      "loss": 0.7709,
+      "step": 3416
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9445348786560195e-06,
+      "loss": 0.266,
+      "step": 3417
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9389552821351945e-06,
+      "loss": 0.7863,
+      "step": 3418
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.933377199168444e-06,
+      "loss": 0.8089,
+      "step": 3419
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9278006318347125e-06,
+      "loss": 0.771,
+      "step": 3420
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.922225582212396e-06,
+      "loss": 0.7976,
+      "step": 3421
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.916652052379318e-06,
+      "loss": 0.7727,
+      "step": 3422
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.911080044412738e-06,
+      "loss": 0.7376,
+      "step": 3423
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9055095603893484e-06,
+      "loss": 0.8215,
+      "step": 3424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.899940602385266e-06,
+      "loss": 0.7654,
+      "step": 3425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8943731724760515e-06,
+      "loss": 0.7781,
+      "step": 3426
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.888807272736685e-06,
+      "loss": 0.7965,
+      "step": 3427
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.88324290524159e-06,
+      "loss": 0.8041,
+      "step": 3428
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8776800720646e-06,
+      "loss": 0.8301,
+      "step": 3429
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.872118775278989e-06,
+      "loss": 0.8038,
+      "step": 3430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8665590169574575e-06,
+      "loss": 0.8149,
+      "step": 3431
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.861000799172134e-06,
+      "loss": 0.8128,
+      "step": 3432
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.855444123994562e-06,
+      "loss": 0.6745,
+      "step": 3433
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.849888993495727e-06,
+      "loss": 0.7395,
+      "step": 3434
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8443354097460235e-06,
+      "loss": 0.7004,
+      "step": 3435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.838783374815278e-06,
+      "loss": 0.7921,
+      "step": 3436
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.833232890772739e-06,
+      "loss": 0.7481,
+      "step": 3437
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.827683959687076e-06,
+      "loss": 0.7483,
+      "step": 3438
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.822136583626374e-06,
+      "loss": 0.266,
+      "step": 3439
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8165907646581474e-06,
+      "loss": 0.7901,
+      "step": 3440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.811046504849327e-06,
+      "loss": 0.8103,
+      "step": 3441
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.805503806266265e-06,
+      "loss": 0.7863,
+      "step": 3442
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.799962670974721e-06,
+      "loss": 0.8544,
+      "step": 3443
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.794423101039886e-06,
+      "loss": 0.7953,
+      "step": 3444
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.788885098526359e-06,
+      "loss": 0.7785,
+      "step": 3445
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.78334866549816e-06,
+      "loss": 0.7562,
+      "step": 3446
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.777813804018719e-06,
+      "loss": 0.7946,
+      "step": 3447
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.772280516150884e-06,
+      "loss": 0.7944,
+      "step": 3448
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.766748803956912e-06,
+      "loss": 0.7648,
+      "step": 3449
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.761218669498476e-06,
+      "loss": 0.7454,
+      "step": 3450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.755690114836667e-06,
+      "loss": 0.7729,
+      "step": 3451
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.750163142031973e-06,
+      "loss": 0.7798,
+      "step": 3452
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.744637753144302e-06,
+      "loss": 0.8149,
+      "step": 3453
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.739113950232973e-06,
+      "loss": 0.7401,
+      "step": 3454
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.733591735356711e-06,
+      "loss": 0.718,
+      "step": 3455
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.728071110573652e-06,
+      "loss": 0.7968,
+      "step": 3456
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.722552077941328e-06,
+      "loss": 0.7696,
+      "step": 3457
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.717034639516694e-06,
+      "loss": 0.7744,
+      "step": 3458
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.711518797356098e-06,
+      "loss": 0.7466,
+      "step": 3459
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.706004553515303e-06,
+      "loss": 0.7772,
+      "step": 3460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.700491910049464e-06,
+      "loss": 0.7192,
+      "step": 3461
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6949808690131515e-06,
+      "loss": 0.788,
+      "step": 3462
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.689471432460334e-06,
+      "loss": 0.7462,
+      "step": 3463
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6839636024443815e-06,
+      "loss": 0.7618,
+      "step": 3464
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.67845738101807e-06,
+      "loss": 0.8562,
+      "step": 3465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.672952770233564e-06,
+      "loss": 0.7716,
+      "step": 3466
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.667449772142441e-06,
+      "loss": 0.8226,
+      "step": 3467
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.66194838879567e-06,
+      "loss": 0.7814,
+      "step": 3468
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.656448622243626e-06,
+      "loss": 0.7786,
+      "step": 3469
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.65095047453607e-06,
+      "loss": 0.8003,
+      "step": 3470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.645453947722164e-06,
+      "loss": 0.7634,
+      "step": 3471
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.639959043850467e-06,
+      "loss": 0.8231,
+      "step": 3472
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.634465764968936e-06,
+      "loss": 0.722,
+      "step": 3473
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.628974113124924e-06,
+      "loss": 0.7934,
+      "step": 3474
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6234840903651645e-06,
+      "loss": 0.8217,
+      "step": 3475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.617995698735796e-06,
+      "loss": 0.8148,
+      "step": 3476
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.612508940282348e-06,
+      "loss": 0.2555,
+      "step": 3477
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.607023817049737e-06,
+      "loss": 0.8318,
+      "step": 3478
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.601540331082276e-06,
+      "loss": 0.7785,
+      "step": 3479
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5960584844236565e-06,
+      "loss": 0.7871,
+      "step": 3480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.590578279116975e-06,
+      "loss": 0.7133,
+      "step": 3481
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.585099717204698e-06,
+      "loss": 0.7588,
+      "step": 3482
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5796228007286944e-06,
+      "loss": 0.7538,
+      "step": 3483
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.57414753173022e-06,
+      "loss": 0.7699,
+      "step": 3484
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5686739122499e-06,
+      "loss": 0.7662,
+      "step": 3485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.563201944327762e-06,
+      "loss": 0.7714,
+      "step": 3486
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.557731630003211e-06,
+      "loss": 0.8342,
+      "step": 3487
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5522629713150435e-06,
+      "loss": 0.7466,
+      "step": 3488
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.54679597030142e-06,
+      "loss": 0.8602,
+      "step": 3489
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.541330628999903e-06,
+      "loss": 0.7721,
+      "step": 3490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.535866949447431e-06,
+      "loss": 0.818,
+      "step": 3491
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.530404933680314e-06,
+      "loss": 0.7704,
+      "step": 3492
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5249445837342555e-06,
+      "loss": 0.7974,
+      "step": 3493
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.519485901644326e-06,
+      "loss": 0.8086,
+      "step": 3494
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5140288894449804e-06,
+      "loss": 0.8353,
+      "step": 3495
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5085735491700535e-06,
+      "loss": 0.7458,
+      "step": 3496
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5031198828527544e-06,
+      "loss": 0.7417,
+      "step": 3497
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.497667892525671e-06,
+      "loss": 0.7925,
+      "step": 3498
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.492217580220756e-06,
+      "loss": 0.7914,
+      "step": 3499
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.486768947969348e-06,
+      "loss": 0.7985,
+      "step": 3500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.48132199780216e-06,
+      "loss": 0.8017,
+      "step": 3501
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.475876731749266e-06,
+      "loss": 0.775,
+      "step": 3502
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.470433151840129e-06,
+      "loss": 0.7739,
+      "step": 3503
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.464991260103567e-06,
+      "loss": 0.7889,
+      "step": 3504
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.45955105856778e-06,
+      "loss": 0.7933,
+      "step": 3505
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.454112549260335e-06,
+      "loss": 0.8456,
+      "step": 3506
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4486757342081705e-06,
+      "loss": 0.7727,
+      "step": 3507
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.443240615437586e-06,
+      "loss": 0.7787,
+      "step": 3508
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.437807194974256e-06,
+      "loss": 0.785,
+      "step": 3509
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.432375474843219e-06,
+      "loss": 0.7618,
+      "step": 3510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.426945457068887e-06,
+      "loss": 0.8065,
+      "step": 3511
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4215171436750214e-06,
+      "loss": 0.8105,
+      "step": 3512
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.416090536684769e-06,
+      "loss": 0.2497,
+      "step": 3513
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.410665638120618e-06,
+      "loss": 0.7647,
+      "step": 3514
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.405242450004438e-06,
+      "loss": 0.7794,
+      "step": 3515
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.399820974357457e-06,
+      "loss": 0.7764,
+      "step": 3516
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.394401213200258e-06,
+      "loss": 0.7931,
+      "step": 3517
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.38898316855279e-06,
+      "loss": 0.7881,
+      "step": 3518
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.383566842434365e-06,
+      "loss": 0.7242,
+      "step": 3519
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.37815223686365e-06,
+      "loss": 0.7532,
+      "step": 3520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.372739353858674e-06,
+      "loss": 0.8189,
+      "step": 3521
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.367328195436816e-06,
+      "loss": 0.8293,
+      "step": 3522
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.361918763614823e-06,
+      "loss": 0.8328,
+      "step": 3523
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.356511060408797e-06,
+      "loss": 0.8352,
+      "step": 3524
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.351105087834188e-06,
+      "loss": 0.8031,
+      "step": 3525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.345700847905804e-06,
+      "loss": 0.8061,
+      "step": 3526
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.340298342637809e-06,
+      "loss": 0.8064,
+      "step": 3527
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.334897574043723e-06,
+      "loss": 0.8328,
+      "step": 3528
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3294985441364165e-06,
+      "loss": 0.6799,
+      "step": 3529
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3241012549281136e-06,
+      "loss": 0.7722,
+      "step": 3530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.31870570843038e-06,
+      "loss": 0.7754,
+      "step": 3531
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.313311906654146e-06,
+      "loss": 0.7108,
+      "step": 3532
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3079198516096805e-06,
+      "loss": 0.8546,
+      "step": 3533
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.302529545306614e-06,
+      "loss": 0.754,
+      "step": 3534
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.297140989753908e-06,
+      "loss": 0.7828,
+      "step": 3535
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.291754186959888e-06,
+      "loss": 0.6783,
+      "step": 3536
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.28636913893221e-06,
+      "loss": 0.7661,
+      "step": 3537
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.280985847677893e-06,
+      "loss": 0.7716,
+      "step": 3538
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2756043152032934e-06,
+      "loss": 0.8153,
+      "step": 3539
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2702245435141055e-06,
+      "loss": 0.7866,
+      "step": 3540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2648465346153775e-06,
+      "loss": 0.2771,
+      "step": 3541
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.259470290511497e-06,
+      "loss": 0.8128,
+      "step": 3542
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.254095813206193e-06,
+      "loss": 0.754,
+      "step": 3543
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.248723104702542e-06,
+      "loss": 0.7851,
+      "step": 3544
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.243352167002949e-06,
+      "loss": 0.2281,
+      "step": 3545
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.23798300210917e-06,
+      "loss": 0.8521,
+      "step": 3546
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.232615612022293e-06,
+      "loss": 0.756,
+      "step": 3547
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.227249998742748e-06,
+      "loss": 0.739,
+      "step": 3548
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.221886164270309e-06,
+      "loss": 0.7601,
+      "step": 3549
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.216524110604073e-06,
+      "loss": 0.7866,
+      "step": 3550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.211163839742484e-06,
+      "loss": 0.7444,
+      "step": 3551
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.205805353683317e-06,
+      "loss": 0.77,
+      "step": 3552
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.20044865442369e-06,
+      "loss": 0.7686,
+      "step": 3553
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.195093743960039e-06,
+      "loss": 0.7425,
+      "step": 3554
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.189740624288147e-06,
+      "loss": 0.7366,
+      "step": 3555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.184389297403129e-06,
+      "loss": 0.7479,
+      "step": 3556
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.17903976529942e-06,
+      "loss": 0.7765,
+      "step": 3557
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.173692029970803e-06,
+      "loss": 0.788,
+      "step": 3558
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1683460934103734e-06,
+      "loss": 0.7619,
+      "step": 3559
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.163001957610571e-06,
+      "loss": 0.7679,
+      "step": 3560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1576596245631565e-06,
+      "loss": 0.257,
+      "step": 3561
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.152319096259223e-06,
+      "loss": 0.8301,
+      "step": 3562
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.146980374689192e-06,
+      "loss": 0.812,
+      "step": 3563
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1416434618428e-06,
+      "loss": 0.7922,
+      "step": 3564
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.136308359709123e-06,
+      "loss": 0.8264,
+      "step": 3565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.130975070276557e-06,
+      "loss": 0.8023,
+      "step": 3566
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.125643595532826e-06,
+      "loss": 0.7894,
+      "step": 3567
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.12031393746497e-06,
+      "loss": 0.805,
+      "step": 3568
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.114986098059354e-06,
+      "loss": 0.7847,
+      "step": 3569
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.109660079301668e-06,
+      "loss": 0.8336,
+      "step": 3570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.104335883176926e-06,
+      "loss": 0.7828,
+      "step": 3571
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.099013511669462e-06,
+      "loss": 0.8203,
+      "step": 3572
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0936929667629185e-06,
+      "loss": 0.7787,
+      "step": 3573
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.088374250440272e-06,
+      "loss": 0.7901,
+      "step": 3574
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.08305736468381e-06,
+      "loss": 0.7977,
+      "step": 3575
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.077742311475144e-06,
+      "loss": 0.7596,
+      "step": 3576
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.072429092795191e-06,
+      "loss": 0.7459,
+      "step": 3577
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.067117710624194e-06,
+      "loss": 0.8208,
+      "step": 3578
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.061808166941711e-06,
+      "loss": 0.7522,
+      "step": 3579
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.056500463726608e-06,
+      "loss": 0.7503,
+      "step": 3580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.051194602957075e-06,
+      "loss": 0.8012,
+      "step": 3581
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.045890586610602e-06,
+      "loss": 0.816,
+      "step": 3582
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.040588416664005e-06,
+      "loss": 0.7597,
+      "step": 3583
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0352880950934044e-06,
+      "loss": 0.8,
+      "step": 3584
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.029989623874234e-06,
+      "loss": 0.2826,
+      "step": 3585
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.024693004981243e-06,
+      "loss": 0.7243,
+      "step": 3586
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.019398240388476e-06,
+      "loss": 0.7531,
+      "step": 3587
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0141053320692986e-06,
+      "loss": 0.8059,
+      "step": 3588
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.008814281996385e-06,
+      "loss": 0.815,
+      "step": 3589
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.00352509214171e-06,
+      "loss": 0.7957,
+      "step": 3590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.998237764476554e-06,
+      "loss": 0.7908,
+      "step": 3591
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.992952300971512e-06,
+      "loss": 0.7442,
+      "step": 3592
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.987668703596479e-06,
+      "loss": 0.7389,
+      "step": 3593
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.982386974320656e-06,
+      "loss": 0.7781,
+      "step": 3594
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.977107115112549e-06,
+      "loss": 0.7351,
+      "step": 3595
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.97182912793996e-06,
+      "loss": 0.7934,
+      "step": 3596
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9665530147700004e-06,
+      "loss": 0.8328,
+      "step": 3597
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9612787775690814e-06,
+      "loss": 0.7836,
+      "step": 3598
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.95600641830292e-06,
+      "loss": 0.7354,
+      "step": 3599
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.950735938936517e-06,
+      "loss": 0.7621,
+      "step": 3600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9454673414341945e-06,
+      "loss": 0.8218,
+      "step": 3601
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9402006277595535e-06,
+      "loss": 0.7948,
+      "step": 3602
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.934935799875505e-06,
+      "loss": 0.7526,
+      "step": 3603
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9296728597442585e-06,
+      "loss": 0.7525,
+      "step": 3604
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.924411809327308e-06,
+      "loss": 0.7283,
+      "step": 3605
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9191526505854525e-06,
+      "loss": 0.7978,
+      "step": 3606
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9138953854787855e-06,
+      "loss": 0.7522,
+      "step": 3607
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.908640015966692e-06,
+      "loss": 0.7665,
+      "step": 3608
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.903386544007857e-06,
+      "loss": 0.7696,
+      "step": 3609
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.898134971560244e-06,
+      "loss": 0.8337,
+      "step": 3610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.892885300581127e-06,
+      "loss": 0.2502,
+      "step": 3611
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.887637533027052e-06,
+      "loss": 0.7199,
+      "step": 3612
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.882391670853871e-06,
+      "loss": 0.8251,
+      "step": 3613
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.877147716016723e-06,
+      "loss": 0.8037,
+      "step": 3614
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8719056704700265e-06,
+      "loss": 0.7773,
+      "step": 3615
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.866665536167499e-06,
+      "loss": 0.7677,
+      "step": 3616
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8614273150621416e-06,
+      "loss": 0.7976,
+      "step": 3617
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.856191009106248e-06,
+      "loss": 0.7246,
+      "step": 3618
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.850956620251384e-06,
+      "loss": 0.7969,
+      "step": 3619
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.845724150448413e-06,
+      "loss": 0.8024,
+      "step": 3620
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.840493601647483e-06,
+      "loss": 0.8322,
+      "step": 3621
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.835264975798023e-06,
+      "loss": 0.7287,
+      "step": 3622
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.830038274848743e-06,
+      "loss": 0.7673,
+      "step": 3623
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.824813500747635e-06,
+      "loss": 0.7753,
+      "step": 3624
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.819590655441979e-06,
+      "loss": 0.7997,
+      "step": 3625
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.814369740878333e-06,
+      "loss": 0.7578,
+      "step": 3626
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8091507590025354e-06,
+      "loss": 0.8125,
+      "step": 3627
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.803933711759707e-06,
+      "loss": 0.7844,
+      "step": 3628
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.798718601094239e-06,
+      "loss": 0.8094,
+      "step": 3629
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.793505428949808e-06,
+      "loss": 0.8182,
+      "step": 3630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.78829419726937e-06,
+      "loss": 0.7561,
+      "step": 3631
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.783084907995156e-06,
+      "loss": 0.7539,
+      "step": 3632
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.777877563068668e-06,
+      "loss": 0.7499,
+      "step": 3633
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.772672164430683e-06,
+      "loss": 0.8108,
+      "step": 3634
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.767468714021262e-06,
+      "loss": 0.7431,
+      "step": 3635
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.762267213779731e-06,
+      "loss": 0.7394,
+      "step": 3636
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7570676656447e-06,
+      "loss": 0.7615,
+      "step": 3637
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.751870071554035e-06,
+      "loss": 0.7792,
+      "step": 3638
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.746674433444887e-06,
+      "loss": 0.7581,
+      "step": 3639
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.741480753253672e-06,
+      "loss": 0.8013,
+      "step": 3640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.736289032916084e-06,
+      "loss": 0.8369,
+      "step": 3641
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.731099274367071e-06,
+      "loss": 0.7378,
+      "step": 3642
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.725911479540867e-06,
+      "loss": 0.7923,
+      "step": 3643
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.720725650370967e-06,
+      "loss": 0.7822,
+      "step": 3644
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.715541788790125e-06,
+      "loss": 0.8024,
+      "step": 3645
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.710359896730379e-06,
+      "loss": 0.7944,
+      "step": 3646
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.705179976123015e-06,
+      "loss": 0.7972,
+      "step": 3647
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7000020288985995e-06,
+      "loss": 0.7725,
+      "step": 3648
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.694826056986953e-06,
+      "loss": 0.8082,
+      "step": 3649
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.689652062317165e-06,
+      "loss": 0.2872,
+      "step": 3650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.684480046817591e-06,
+      "loss": 0.7306,
+      "step": 3651
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.679310012415838e-06,
+      "loss": 0.7235,
+      "step": 3652
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.674141961038783e-06,
+      "loss": 0.7588,
+      "step": 3653
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.668975894612566e-06,
+      "loss": 0.8083,
+      "step": 3654
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6638118150625774e-06,
+      "loss": 0.7955,
+      "step": 3655
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.658649724313481e-06,
+      "loss": 0.7378,
+      "step": 3656
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.653489624289181e-06,
+      "loss": 0.7479,
+      "step": 3657
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.648331516912857e-06,
+      "loss": 0.7778,
+      "step": 3658
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.643175404106938e-06,
+      "loss": 0.7572,
+      "step": 3659
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.638021287793113e-06,
+      "loss": 0.7784,
+      "step": 3660
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.632869169892319e-06,
+      "loss": 0.7467,
+      "step": 3661
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.627719052324757e-06,
+      "loss": 0.7882,
+      "step": 3662
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.622570937009879e-06,
+      "loss": 0.7503,
+      "step": 3663
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.617424825866391e-06,
+      "loss": 0.2466,
+      "step": 3664
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.612280720812256e-06,
+      "loss": 0.7571,
+      "step": 3665
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6071386237646805e-06,
+      "loss": 0.7931,
+      "step": 3666
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.601998536640126e-06,
+      "loss": 0.777,
+      "step": 3667
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5968604613543076e-06,
+      "loss": 0.7746,
+      "step": 3668
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.591724399822196e-06,
+      "loss": 0.8135,
+      "step": 3669
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.586590353957995e-06,
+      "loss": 0.7403,
+      "step": 3670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.581458325675171e-06,
+      "loss": 0.754,
+      "step": 3671
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.576328316886435e-06,
+      "loss": 0.7864,
+      "step": 3672
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.571200329503743e-06,
+      "loss": 0.7858,
+      "step": 3673
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.566074365438305e-06,
+      "loss": 0.7797,
+      "step": 3674
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5609504266005646e-06,
+      "loss": 0.7929,
+      "step": 3675
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.555828514900223e-06,
+      "loss": 0.8123,
+      "step": 3676
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.550708632246212e-06,
+      "loss": 0.2711,
+      "step": 3677
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.545590780546719e-06,
+      "loss": 0.7532,
+      "step": 3678
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.540474961709176e-06,
+      "loss": 0.8129,
+      "step": 3679
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.535361177640244e-06,
+      "loss": 0.8079,
+      "step": 3680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.530249430245838e-06,
+      "loss": 0.8194,
+      "step": 3681
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.525139721431109e-06,
+      "loss": 0.8087,
+      "step": 3682
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.520032053100453e-06,
+      "loss": 0.7916,
+      "step": 3683
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.514926427157492e-06,
+      "loss": 0.7368,
+      "step": 3684
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.509822845505103e-06,
+      "loss": 0.298,
+      "step": 3685
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.504721310045392e-06,
+      "loss": 0.7828,
+      "step": 3686
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.49962182267971e-06,
+      "loss": 0.8103,
+      "step": 3687
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.494524385308634e-06,
+      "loss": 0.7514,
+      "step": 3688
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.48942899983198e-06,
+      "loss": 0.668,
+      "step": 3689
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.484335668148804e-06,
+      "loss": 0.7701,
+      "step": 3690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.479244392157394e-06,
+      "loss": 0.7836,
+      "step": 3691
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.474155173755273e-06,
+      "loss": 0.7747,
+      "step": 3692
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.469068014839201e-06,
+      "loss": 0.7494,
+      "step": 3693
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.463982917305155e-06,
+      "loss": 0.7182,
+      "step": 3694
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.458899883048361e-06,
+      "loss": 0.7945,
+      "step": 3695
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.453818913963267e-06,
+      "loss": 0.8272,
+      "step": 3696
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.448740011943557e-06,
+      "loss": 0.7918,
+      "step": 3697
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.443663178882137e-06,
+      "loss": 0.8217,
+      "step": 3698
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4385884166711504e-06,
+      "loss": 0.2733,
+      "step": 3699
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.433515727201957e-06,
+      "loss": 0.7823,
+      "step": 3700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.428445112365158e-06,
+      "loss": 0.779,
+      "step": 3701
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.423376574050575e-06,
+      "loss": 0.7245,
+      "step": 3702
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4183101141472505e-06,
+      "loss": 0.7972,
+      "step": 3703
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.413245734543461e-06,
+      "loss": 0.7155,
+      "step": 3704
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.408183437126703e-06,
+      "loss": 0.8484,
+      "step": 3705
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4031232237837e-06,
+      "loss": 0.2467,
+      "step": 3706
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.398065096400399e-06,
+      "loss": 0.7329,
+      "step": 3707
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.39300905686196e-06,
+      "loss": 0.7952,
+      "step": 3708
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.38795510705278e-06,
+      "loss": 0.7921,
+      "step": 3709
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.382903248856464e-06,
+      "loss": 0.7315,
+      "step": 3710
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.377853484155848e-06,
+      "loss": 0.7558,
+      "step": 3711
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.372805814832976e-06,
+      "loss": 0.7485,
+      "step": 3712
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.36776024276912e-06,
+      "loss": 0.8148,
+      "step": 3713
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3627167698447704e-06,
+      "loss": 0.2495,
+      "step": 3714
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.357675397939632e-06,
+      "loss": 0.7667,
+      "step": 3715
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.35263612893263e-06,
+      "loss": 0.7508,
+      "step": 3716
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.347598964701898e-06,
+      "loss": 0.715,
+      "step": 3717
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.342563907124793e-06,
+      "loss": 0.8029,
+      "step": 3718
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.337530958077887e-06,
+      "loss": 0.7351,
+      "step": 3719
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.332500119436959e-06,
+      "loss": 0.7817,
+      "step": 3720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.327471393077009e-06,
+      "loss": 0.7402,
+      "step": 3721
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.322444780872245e-06,
+      "loss": 0.7595,
+      "step": 3722
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3174202846960864e-06,
+      "loss": 0.809,
+      "step": 3723
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3123979064211704e-06,
+      "loss": 0.7743,
+      "step": 3724
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.307377647919343e-06,
+      "loss": 0.8092,
+      "step": 3725
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.30235951106165e-06,
+      "loss": 0.7416,
+      "step": 3726
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.297343497718359e-06,
+      "loss": 0.7782,
+      "step": 3727
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.29232960975894e-06,
+      "loss": 0.8292,
+      "step": 3728
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.287317849052075e-06,
+      "loss": 0.8088,
+      "step": 3729
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.282308217465653e-06,
+      "loss": 0.7519,
+      "step": 3730
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.277300716866761e-06,
+      "loss": 0.8154,
+      "step": 3731
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.272295349121698e-06,
+      "loss": 0.7881,
+      "step": 3732
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.267292116095969e-06,
+      "loss": 0.7808,
+      "step": 3733
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.262291019654287e-06,
+      "loss": 0.752,
+      "step": 3734
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.257292061660556e-06,
+      "loss": 0.2556,
+      "step": 3735
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.252295243977895e-06,
+      "loss": 0.2449,
+      "step": 3736
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.247300568468622e-06,
+      "loss": 0.8148,
+      "step": 3737
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.242308036994254e-06,
+      "loss": 0.7493,
+      "step": 3738
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.237317651415515e-06,
+      "loss": 0.8295,
+      "step": 3739
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.232329413592321e-06,
+      "loss": 0.76,
+      "step": 3740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.22734332538379e-06,
+      "loss": 0.7464,
+      "step": 3741
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.222359388648248e-06,
+      "loss": 0.7369,
+      "step": 3742
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.217377605243202e-06,
+      "loss": 0.761,
+      "step": 3743
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.212397977025374e-06,
+      "loss": 0.7704,
+      "step": 3744
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2074205058506675e-06,
+      "loss": 0.7696,
+      "step": 3745
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.202445193574194e-06,
+      "loss": 0.7453,
+      "step": 3746
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1974720420502545e-06,
+      "loss": 0.7689,
+      "step": 3747
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1925010531323485e-06,
+      "loss": 0.7994,
+      "step": 3748
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.187532228673162e-06,
+      "loss": 0.2168,
+      "step": 3749
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.182565570524582e-06,
+      "loss": 0.7876,
+      "step": 3750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.177601080537683e-06,
+      "loss": 0.23,
+      "step": 3751
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.17263876056274e-06,
+      "loss": 0.7591,
+      "step": 3752
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.167678612449206e-06,
+      "loss": 0.7346,
+      "step": 3753
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.162720638045738e-06,
+      "loss": 0.7801,
+      "step": 3754
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.157764839200168e-06,
+      "loss": 0.7439,
+      "step": 3755
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.152811217759529e-06,
+      "loss": 0.7942,
+      "step": 3756
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.147859775570041e-06,
+      "loss": 0.831,
+      "step": 3757
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.142910514477113e-06,
+      "loss": 0.7448,
+      "step": 3758
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.137963436325329e-06,
+      "loss": 0.7761,
+      "step": 3759
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.133018542958474e-06,
+      "loss": 0.7015,
+      "step": 3760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1280758362195115e-06,
+      "loss": 0.7886,
+      "step": 3761
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.123135317950596e-06,
+      "loss": 0.7876,
+      "step": 3762
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.118196989993055e-06,
+      "loss": 0.8162,
+      "step": 3763
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.113260854187413e-06,
+      "loss": 0.7673,
+      "step": 3764
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.108326912373364e-06,
+      "loss": 0.7716,
+      "step": 3765
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.103395166389795e-06,
+      "loss": 0.7922,
+      "step": 3766
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0984656180747755e-06,
+      "loss": 0.7796,
+      "step": 3767
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.093538269265544e-06,
+      "loss": 0.8033,
+      "step": 3768
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.088613121798528e-06,
+      "loss": 0.772,
+      "step": 3769
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.083690177509336e-06,
+      "loss": 0.7491,
+      "step": 3770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.078769438232753e-06,
+      "loss": 0.8072,
+      "step": 3771
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.073850905802742e-06,
+      "loss": 0.7612,
+      "step": 3772
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.068934582052439e-06,
+      "loss": 0.8041,
+      "step": 3773
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.064020468814169e-06,
+      "loss": 0.767,
+      "step": 3774
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0591085679194145e-06,
+      "loss": 0.7174,
+      "step": 3775
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.054198881198854e-06,
+      "loss": 0.7465,
+      "step": 3776
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.049291410482325e-06,
+      "loss": 0.7647,
+      "step": 3777
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.044386157598845e-06,
+      "loss": 0.7824,
+      "step": 3778
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.039483124376609e-06,
+      "loss": 0.7418,
+      "step": 3779
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.034582312642977e-06,
+      "loss": 0.8019,
+      "step": 3780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.029683724224491e-06,
+      "loss": 0.8427,
+      "step": 3781
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0247873609468506e-06,
+      "loss": 0.821,
+      "step": 3782
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.019893224634936e-06,
+      "loss": 0.7887,
+      "step": 3783
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.015001317112797e-06,
+      "loss": 0.7137,
+      "step": 3784
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.010111640203652e-06,
+      "loss": 0.7566,
+      "step": 3785
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.005224195729886e-06,
+      "loss": 0.719,
+      "step": 3786
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.000338985513046e-06,
+      "loss": 0.7733,
+      "step": 3787
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.995456011373859e-06,
+      "loss": 0.764,
+      "step": 3788
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.990575275132212e-06,
+      "loss": 0.7826,
+      "step": 3789
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.985696778607163e-06,
+      "loss": 0.8067,
+      "step": 3790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.980820523616924e-06,
+      "loss": 0.7533,
+      "step": 3791
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.97594651197888e-06,
+      "loss": 0.7603,
+      "step": 3792
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9710747455095786e-06,
+      "loss": 0.7731,
+      "step": 3793
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.966205226024733e-06,
+      "loss": 0.7899,
+      "step": 3794
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.961337955339218e-06,
+      "loss": 0.772,
+      "step": 3795
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.956472935267062e-06,
+      "loss": 0.7554,
+      "step": 3796
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.951610167621467e-06,
+      "loss": 0.7933,
+      "step": 3797
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9467496542147844e-06,
+      "loss": 0.7638,
+      "step": 3798
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.941891396858537e-06,
+      "loss": 0.7145,
+      "step": 3799
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.937035397363393e-06,
+      "loss": 0.7534,
+      "step": 3800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9321816575391925e-06,
+      "loss": 0.7917,
+      "step": 3801
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.927330179194925e-06,
+      "loss": 0.8086,
+      "step": 3802
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.922480964138741e-06,
+      "loss": 0.7213,
+      "step": 3803
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9176340141779476e-06,
+      "loss": 0.7445,
+      "step": 3804
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9127893311190035e-06,
+      "loss": 0.7947,
+      "step": 3805
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.907946916767524e-06,
+      "loss": 0.7178,
+      "step": 3806
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9031067729282875e-06,
+      "loss": 0.8483,
+      "step": 3807
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.89826890140521e-06,
+      "loss": 0.2811,
+      "step": 3808
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.893433304001376e-06,
+      "loss": 0.7811,
+      "step": 3809
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.88859998251901e-06,
+      "loss": 0.8024,
+      "step": 3810
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.883768938759498e-06,
+      "loss": 0.7385,
+      "step": 3811
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.878940174523371e-06,
+      "loss": 0.7449,
+      "step": 3812
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.874113691610318e-06,
+      "loss": 0.705,
+      "step": 3813
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.869289491819166e-06,
+      "loss": 0.7762,
+      "step": 3814
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.864467576947899e-06,
+      "loss": 0.7873,
+      "step": 3815
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.859647948793649e-06,
+      "loss": 0.7932,
+      "step": 3816
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8548306091526975e-06,
+      "loss": 0.7895,
+      "step": 3817
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.850015559820465e-06,
+      "loss": 0.7798,
+      "step": 3818
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.845202802591528e-06,
+      "loss": 0.7986,
+      "step": 3819
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.840392339259599e-06,
+      "loss": 0.7601,
+      "step": 3820
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.835584171617543e-06,
+      "loss": 0.7655,
+      "step": 3821
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.83077830145737e-06,
+      "loss": 0.7414,
+      "step": 3822
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.825974730570231e-06,
+      "loss": 0.7787,
+      "step": 3823
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.821173460746415e-06,
+      "loss": 0.8098,
+      "step": 3824
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.816374493775362e-06,
+      "loss": 0.771,
+      "step": 3825
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.811577831445651e-06,
+      "loss": 0.8012,
+      "step": 3826
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8067834755450018e-06,
+      "loss": 0.7768,
+      "step": 3827
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.80199142786027e-06,
+      "loss": 0.8084,
+      "step": 3828
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7972016901774613e-06,
+      "loss": 0.8049,
+      "step": 3829
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7924142642817053e-06,
+      "loss": 0.7988,
+      "step": 3830
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.787629151957285e-06,
+      "loss": 0.7885,
+      "step": 3831
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.782846354987616e-06,
+      "loss": 0.7246,
+      "step": 3832
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7780658751552445e-06,
+      "loss": 0.8039,
+      "step": 3833
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7732877142418623e-06,
+      "loss": 0.7521,
+      "step": 3834
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.768511874028291e-06,
+      "loss": 0.8038,
+      "step": 3835
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7637383562944916e-06,
+      "loss": 0.7851,
+      "step": 3836
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.75896716281956e-06,
+      "loss": 0.7466,
+      "step": 3837
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.754198295381717e-06,
+      "loss": 0.7493,
+      "step": 3838
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.749431755758325e-06,
+      "loss": 0.7462,
+      "step": 3839
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7446675457258795e-06,
+      "loss": 0.8383,
+      "step": 3840
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7399056670600033e-06,
+      "loss": 0.7939,
+      "step": 3841
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7351461215354467e-06,
+      "loss": 0.7596,
+      "step": 3842
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7303889109260993e-06,
+      "loss": 0.7544,
+      "step": 3843
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7256340370049772e-06,
+      "loss": 0.746,
+      "step": 3844
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7208815015442233e-06,
+      "loss": 0.7904,
+      "step": 3845
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.716131306315117e-06,
+      "loss": 0.7765,
+      "step": 3846
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.71138345308805e-06,
+      "loss": 0.7481,
+      "step": 3847
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.706637943632555e-06,
+      "loss": 0.6832,
+      "step": 3848
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7018947797172864e-06,
+      "loss": 0.2427,
+      "step": 3849
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6971539631100274e-06,
+      "loss": 0.7608,
+      "step": 3850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.692415495577677e-06,
+      "loss": 0.7517,
+      "step": 3851
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.687679378886273e-06,
+      "loss": 0.751,
+      "step": 3852
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6829456148009603e-06,
+      "loss": 0.7489,
+      "step": 3853
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6782142050860204e-06,
+      "loss": 0.7476,
+      "step": 3854
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6734851515048576e-06,
+      "loss": 0.7635,
+      "step": 3855
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.668758455819984e-06,
+      "loss": 0.8155,
+      "step": 3856
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6640341197930475e-06,
+      "loss": 0.7792,
+      "step": 3857
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.65931214518481e-06,
+      "loss": 0.7485,
+      "step": 3858
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6545925337551557e-06,
+      "loss": 0.7798,
+      "step": 3859
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6498752872630904e-06,
+      "loss": 0.7784,
+      "step": 3860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.645160407466727e-06,
+      "loss": 0.8318,
+      "step": 3861
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6404478961233123e-06,
+      "loss": 0.7853,
+      "step": 3862
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.635737754989195e-06,
+      "loss": 0.8016,
+      "step": 3863
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.631029985819854e-06,
+      "loss": 0.8034,
+      "step": 3864
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.626324590369873e-06,
+      "loss": 0.7649,
+      "step": 3865
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.621621570392958e-06,
+      "loss": 0.7495,
+      "step": 3866
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.616920927641928e-06,
+      "loss": 0.7853,
+      "step": 3867
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6122226638687154e-06,
+      "loss": 0.7305,
+      "step": 3868
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6075267808243697e-06,
+      "loss": 0.8174,
+      "step": 3869
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.602833280259044e-06,
+      "loss": 0.7782,
+      "step": 3870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5981421639220115e-06,
+      "loss": 0.7848,
+      "step": 3871
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5934534335616577e-06,
+      "loss": 0.8032,
+      "step": 3872
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.588767090925469e-06,
+      "loss": 0.7862,
+      "step": 3873
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5840831377600548e-06,
+      "loss": 0.8695,
+      "step": 3874
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.579401575811121e-06,
+      "loss": 0.7954,
+      "step": 3875
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.574722406823493e-06,
+      "loss": 0.7842,
+      "step": 3876
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.570045632541099e-06,
+      "loss": 0.7923,
+      "step": 3877
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5653712547069798e-06,
+      "loss": 0.8255,
+      "step": 3878
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5606992750632717e-06,
+      "loss": 0.2351,
+      "step": 3879
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5560296953512296e-06,
+      "loss": 0.7965,
+      "step": 3880
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.551362517311208e-06,
+      "loss": 0.7724,
+      "step": 3881
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.546697742682668e-06,
+      "loss": 0.7613,
+      "step": 3882
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5420353732041756e-06,
+      "loss": 0.7504,
+      "step": 3883
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5373754106133973e-06,
+      "loss": 0.7553,
+      "step": 3884
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5327178566470998e-06,
+      "loss": 0.7649,
+      "step": 3885
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5280627130411615e-06,
+      "loss": 0.7843,
+      "step": 3886
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5234099815305566e-06,
+      "loss": 0.7676,
+      "step": 3887
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.518759663849365e-06,
+      "loss": 0.7704,
+      "step": 3888
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5141117617307554e-06,
+      "loss": 0.7405,
+      "step": 3889
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5094662769070076e-06,
+      "loss": 0.6865,
+      "step": 3890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5048232111094984e-06,
+      "loss": 0.7664,
+      "step": 3891
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5001825660687037e-06,
+      "loss": 0.7557,
+      "step": 3892
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.495544343514188e-06,
+      "loss": 0.7782,
+      "step": 3893
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4909085451746238e-06,
+      "loss": 0.7851,
+      "step": 3894
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.48627517277778e-06,
+      "loss": 0.7918,
+      "step": 3895
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.481644228050509e-06,
+      "loss": 0.7922,
+      "step": 3896
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4770157127187766e-06,
+      "loss": 0.7526,
+      "step": 3897
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4723896285076244e-06,
+      "loss": 0.7899,
+      "step": 3898
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.467765977141202e-06,
+      "loss": 0.7616,
+      "step": 3899
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.463144760342746e-06,
+      "loss": 0.7693,
+      "step": 3900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4585259798345882e-06,
+      "loss": 0.7562,
+      "step": 3901
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.453909637338153e-06,
+      "loss": 0.7963,
+      "step": 3902
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4492957345739497e-06,
+      "loss": 0.756,
+      "step": 3903
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4446842732615835e-06,
+      "loss": 0.7542,
+      "step": 3904
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4400752551197557e-06,
+      "loss": 0.7161,
+      "step": 3905
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.435468681866242e-06,
+      "loss": 0.7509,
+      "step": 3906
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.430864555217921e-06,
+      "loss": 0.7695,
+      "step": 3907
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4262628768907493e-06,
+      "loss": 0.739,
+      "step": 3908
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.421663648599777e-06,
+      "loss": 0.8172,
+      "step": 3909
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4170668720591407e-06,
+      "loss": 0.8215,
+      "step": 3910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4124725489820643e-06,
+      "loss": 0.7833,
+      "step": 3911
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4078806810808494e-06,
+      "loss": 0.7904,
+      "step": 3912
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.403291270066892e-06,
+      "loss": 0.7993,
+      "step": 3913
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3987043176506675e-06,
+      "loss": 0.7523,
+      "step": 3914
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3941198255417395e-06,
+      "loss": 0.8072,
+      "step": 3915
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3895377954487452e-06,
+      "loss": 0.806,
+      "step": 3916
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.384958229079416e-06,
+      "loss": 0.748,
+      "step": 3917
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.380381128140554e-06,
+      "loss": 0.7676,
+      "step": 3918
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3758064943380496e-06,
+      "loss": 0.7439,
+      "step": 3919
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3712343293768766e-06,
+      "loss": 0.8057,
+      "step": 3920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3666646349610767e-06,
+      "loss": 0.8036,
+      "step": 3921
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.362097412793781e-06,
+      "loss": 0.763,
+      "step": 3922
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.357532664577198e-06,
+      "loss": 0.2301,
+      "step": 3923
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.352970392012609e-06,
+      "loss": 0.7829,
+      "step": 3924
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.348410596800382e-06,
+      "loss": 0.7659,
+      "step": 3925
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3438532806399493e-06,
+      "loss": 0.7793,
+      "step": 3926
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3392984452298307e-06,
+      "loss": 0.7444,
+      "step": 3927
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.33474609226761e-06,
+      "loss": 0.777,
+      "step": 3928
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3301962234499594e-06,
+      "loss": 0.7586,
+      "step": 3929
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.325648840472612e-06,
+      "loss": 0.776,
+      "step": 3930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3211039450303827e-06,
+      "loss": 0.8193,
+      "step": 3931
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.316561538817157e-06,
+      "loss": 0.7831,
+      "step": 3932
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3120216235258927e-06,
+      "loss": 0.7209,
+      "step": 3933
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.307484200848624e-06,
+      "loss": 0.742,
+      "step": 3934
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3029492724764435e-06,
+      "loss": 0.8211,
+      "step": 3935
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2984168400995252e-06,
+      "loss": 0.7603,
+      "step": 3936
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.293886905407111e-06,
+      "loss": 0.771,
+      "step": 3937
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.289359470087512e-06,
+      "loss": 0.7953,
+      "step": 3938
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.284834535828104e-06,
+      "loss": 0.7503,
+      "step": 3939
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2803121043153296e-06,
+      "loss": 0.7565,
+      "step": 3940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2757921772347056e-06,
+      "loss": 0.7549,
+      "step": 3941
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2712747562708115e-06,
+      "loss": 0.7466,
+      "step": 3942
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2667598431072957e-06,
+      "loss": 0.7765,
+      "step": 3943
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2622474394268645e-06,
+      "loss": 0.7106,
+      "step": 3944
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.257737546911296e-06,
+      "loss": 0.7418,
+      "step": 3945
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.25323016724143e-06,
+      "loss": 0.757,
+      "step": 3946
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2487253020971708e-06,
+      "loss": 0.2756,
+      "step": 3947
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2442229531574867e-06,
+      "loss": 0.2718,
+      "step": 3948
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2397231221004e-06,
+      "loss": 0.7405,
+      "step": 3949
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2352258106030076e-06,
+      "loss": 0.7308,
+      "step": 3950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.230731020341453e-06,
+      "loss": 0.7238,
+      "step": 3951
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2262387529909533e-06,
+      "loss": 0.7645,
+      "step": 3952
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2217490102257797e-06,
+      "loss": 0.7155,
+      "step": 3953
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2172617937192575e-06,
+      "loss": 0.7458,
+      "step": 3954
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2127771051437786e-06,
+      "loss": 0.7808,
+      "step": 3955
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.208294946170789e-06,
+      "loss": 0.7716,
+      "step": 3956
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2038153184707953e-06,
+      "loss": 0.2522,
+      "step": 3957
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1993382237133507e-06,
+      "loss": 0.7324,
+      "step": 3958
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1948636635670772e-06,
+      "loss": 0.7689,
+      "step": 3959
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.190391639699647e-06,
+      "loss": 0.786,
+      "step": 3960
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.185922153777781e-06,
+      "loss": 0.7192,
+      "step": 3961
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1814552074672642e-06,
+      "loss": 0.7458,
+      "step": 3962
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1769908024329276e-06,
+      "loss": 0.7568,
+      "step": 3963
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1725289403386582e-06,
+      "loss": 0.8389,
+      "step": 3964
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1680696228473962e-06,
+      "loss": 0.792,
+      "step": 3965
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1636128516211316e-06,
+      "loss": 0.7516,
+      "step": 3966
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.15915862832091e-06,
+      "loss": 0.6803,
+      "step": 3967
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1547069546068153e-06,
+      "loss": 0.7499,
+      "step": 3968
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.150257832137995e-06,
+      "loss": 0.7332,
+      "step": 3969
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1458112625726413e-06,
+      "loss": 0.2651,
+      "step": 3970
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.141367247567988e-06,
+      "loss": 0.7642,
+      "step": 3971
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.136925788780328e-06,
+      "loss": 0.8655,
+      "step": 3972
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.132486887864992e-06,
+      "loss": 0.756,
+      "step": 3973
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.128050546476361e-06,
+      "loss": 0.7312,
+      "step": 3974
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.123616766267864e-06,
+      "loss": 0.7598,
+      "step": 3975
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1191855488919776e-06,
+      "loss": 0.7953,
+      "step": 3976
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1147568960002128e-06,
+      "loss": 0.7481,
+      "step": 3977
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.110330809243134e-06,
+      "loss": 0.7726,
+      "step": 3978
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.105907290270348e-06,
+      "loss": 0.7391,
+      "step": 3979
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.101486340730502e-06,
+      "loss": 0.7692,
+      "step": 3980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0970679622712907e-06,
+      "loss": 0.7548,
+      "step": 3981
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.092652156539443e-06,
+      "loss": 0.7532,
+      "step": 3982
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.088238925180729e-06,
+      "loss": 0.7762,
+      "step": 3983
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.083828269839968e-06,
+      "loss": 0.7476,
+      "step": 3984
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0794201921610155e-06,
+      "loss": 0.7136,
+      "step": 3985
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.07501469378676e-06,
+      "loss": 0.8116,
+      "step": 3986
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0706117763591346e-06,
+      "loss": 0.7753,
+      "step": 3987
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0662114415191125e-06,
+      "loss": 0.7715,
+      "step": 3988
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0618136909066996e-06,
+      "loss": 0.7722,
+      "step": 3989
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0574185261609445e-06,
+      "loss": 0.8131,
+      "step": 3990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.053025948919922e-06,
+      "loss": 0.7503,
+      "step": 3991
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0486359608207494e-06,
+      "loss": 0.7531,
+      "step": 3992
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0442485634995844e-06,
+      "loss": 0.7824,
+      "step": 3993
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.039863758591608e-06,
+      "loss": 0.6711,
+      "step": 3994
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.035481547731036e-06,
+      "loss": 0.729,
+      "step": 3995
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.031101932551127e-06,
+      "loss": 0.7823,
+      "step": 3996
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.026724914684164e-06,
+      "loss": 0.7393,
+      "step": 3997
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.022350495761466e-06,
+      "loss": 0.2577,
+      "step": 3998
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0179786774133845e-06,
+      "loss": 0.8141,
+      "step": 3999
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0136094612692923e-06,
+      "loss": 0.745,
+      "step": 4000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0092428489576033e-06,
+      "loss": 0.8153,
+      "step": 4001
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0048788421057562e-06,
+      "loss": 0.7415,
+      "step": 4002
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0005174423402217e-06,
+      "loss": 0.8329,
+      "step": 4003
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9961586512864947e-06,
+      "loss": 0.7677,
+      "step": 4004
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.991802470569096e-06,
+      "loss": 0.7304,
+      "step": 4005
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9874489018115806e-06,
+      "loss": 0.7697,
+      "step": 4006
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.983097946636525e-06,
+      "loss": 0.7879,
+      "step": 4007
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9787496066655376e-06,
+      "loss": 0.7523,
+      "step": 4008
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.97440388351924e-06,
+      "loss": 0.7115,
+      "step": 4009
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9700607788172898e-06,
+      "loss": 0.7531,
+      "step": 4010
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9657202941783656e-06,
+      "loss": 0.7915,
+      "step": 4011
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.961382431220169e-06,
+      "loss": 0.7642,
+      "step": 4012
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9570471915594267e-06,
+      "loss": 0.8133,
+      "step": 4013
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9527145768118803e-06,
+      "loss": 0.7983,
+      "step": 4014
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9483845885923035e-06,
+      "loss": 0.7699,
+      "step": 4015
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9440572285144787e-06,
+      "loss": 0.7661,
+      "step": 4016
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.939732498191221e-06,
+      "loss": 0.6795,
+      "step": 4017
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9354103992343618e-06,
+      "loss": 0.7735,
+      "step": 4018
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9310909332547454e-06,
+      "loss": 0.7897,
+      "step": 4019
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.926774101862241e-06,
+      "loss": 0.7967,
+      "step": 4020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9224599066657366e-06,
+      "loss": 0.7856,
+      "step": 4021
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.918148349273137e-06,
+      "loss": 0.8028,
+      "step": 4022
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9138394312913577e-06,
+      "loss": 0.778,
+      "step": 4023
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.909533154326336e-06,
+      "loss": 0.764,
+      "step": 4024
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9052295199830306e-06,
+      "loss": 0.7785,
+      "step": 4025
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9009285298654e-06,
+      "loss": 0.2746,
+      "step": 4026
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8966301855764343e-06,
+      "loss": 0.7846,
+      "step": 4027
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8923344887181213e-06,
+      "loss": 0.7357,
+      "step": 4028
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8880414408914748e-06,
+      "loss": 0.7663,
+      "step": 4029
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8837510436965154e-06,
+      "loss": 0.7833,
+      "step": 4030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8794632987322767e-06,
+      "loss": 0.2404,
+      "step": 4031
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8751782075968082e-06,
+      "loss": 0.246,
+      "step": 4032
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8708957718871598e-06,
+      "loss": 0.7609,
+      "step": 4033
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.866615993199402e-06,
+      "loss": 0.7711,
+      "step": 4034
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8623388731286097e-06,
+      "loss": 0.8212,
+      "step": 4035
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8580644132688717e-06,
+      "loss": 0.7626,
+      "step": 4036
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8537926152132788e-06,
+      "loss": 0.8064,
+      "step": 4037
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8495234805539316e-06,
+      "loss": 0.7008,
+      "step": 4038
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.84525701088194e-06,
+      "loss": 0.7498,
+      "step": 4039
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8409932077874226e-06,
+      "loss": 0.7811,
+      "step": 4040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8367320728595027e-06,
+      "loss": 0.7598,
+      "step": 4041
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8324736076863034e-06,
+      "loss": 0.7939,
+      "step": 4042
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.82821781385496e-06,
+      "loss": 0.7546,
+      "step": 4043
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8239646929516096e-06,
+      "loss": 0.7555,
+      "step": 4044
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8197142465613924e-06,
+      "loss": 0.7868,
+      "step": 4045
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.815466476268457e-06,
+      "loss": 0.8291,
+      "step": 4046
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.811221383655948e-06,
+      "loss": 0.774,
+      "step": 4047
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.806978970306008e-06,
+      "loss": 0.8212,
+      "step": 4048
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.802739237799793e-06,
+      "loss": 0.7924,
+      "step": 4049
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7985021877174555e-06,
+      "loss": 0.6797,
+      "step": 4050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7942678216381425e-06,
+      "loss": 0.7818,
+      "step": 4051
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.790036141140008e-06,
+      "loss": 0.7563,
+      "step": 4052
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7858071478002003e-06,
+      "loss": 0.8325,
+      "step": 4053
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7815808431948687e-06,
+      "loss": 0.7869,
+      "step": 4054
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7773572288991646e-06,
+      "loss": 0.7883,
+      "step": 4055
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.773136306487223e-06,
+      "loss": 0.751,
+      "step": 4056
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.76891807753219e-06,
+      "loss": 0.6653,
+      "step": 4057
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7647025436062037e-06,
+      "loss": 0.7448,
+      "step": 4058
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7604897062803935e-06,
+      "loss": 0.7144,
+      "step": 4059
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7562795671248856e-06,
+      "loss": 0.7817,
+      "step": 4060
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7520721277088023e-06,
+      "loss": 0.745,
+      "step": 4061
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7478673896002594e-06,
+      "loss": 0.7682,
+      "step": 4062
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.743665354366367e-06,
+      "loss": 0.7845,
+      "step": 4063
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7394660235732286e-06,
+      "loss": 0.7869,
+      "step": 4064
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7352693987859314e-06,
+      "loss": 0.8071,
+      "step": 4065
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7310754815685627e-06,
+      "loss": 0.7624,
+      "step": 4066
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7268842734841993e-06,
+      "loss": 0.8079,
+      "step": 4067
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.72269577609491e-06,
+      "loss": 0.7246,
+      "step": 4068
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7185099909617453e-06,
+      "loss": 0.2514,
+      "step": 4069
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.714326919644754e-06,
+      "loss": 0.8124,
+      "step": 4070
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7101465637029655e-06,
+      "loss": 0.2604,
+      "step": 4071
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.705968924694403e-06,
+      "loss": 0.7256,
+      "step": 4072
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7017940041760803e-06,
+      "loss": 0.7789,
+      "step": 4073
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6976218037039846e-06,
+      "loss": 0.8177,
+      "step": 4074
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6934523248331014e-06,
+      "loss": 0.7301,
+      "step": 4075
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.689285569117399e-06,
+      "loss": 0.8021,
+      "step": 4076
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6851215381098304e-06,
+      "loss": 0.805,
+      "step": 4077
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.680960233362335e-06,
+      "loss": 0.7773,
+      "step": 4078
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.676801656425827e-06,
+      "loss": 0.7941,
+      "step": 4079
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6726458088502184e-06,
+      "loss": 0.7537,
+      "step": 4080
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6684926921843903e-06,
+      "loss": 0.7622,
+      "step": 4081
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.664342307976213e-06,
+      "loss": 0.7222,
+      "step": 4082
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6601946577725423e-06,
+      "loss": 0.696,
+      "step": 4083
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6560497431192044e-06,
+      "loss": 0.75,
+      "step": 4084
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6519075655610126e-06,
+      "loss": 0.7611,
+      "step": 4085
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6477681266417622e-06,
+      "loss": 0.7928,
+      "step": 4086
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6436314279042265e-06,
+      "loss": 0.7767,
+      "step": 4087
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.639497470890149e-06,
+      "loss": 0.2769,
+      "step": 4088
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6353662571402616e-06,
+      "loss": 0.8786,
+      "step": 4089
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.631237788194274e-06,
+      "loss": 0.7839,
+      "step": 4090
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6271120655908633e-06,
+      "loss": 0.8189,
+      "step": 4091
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6229890908676956e-06,
+      "loss": 0.788,
+      "step": 4092
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.618868865561399e-06,
+      "loss": 0.7403,
+      "step": 4093
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6147513912075873e-06,
+      "loss": 0.7847,
+      "step": 4094
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.610636669340847e-06,
+      "loss": 0.7634,
+      "step": 4095
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.606524701494737e-06,
+      "loss": 0.7255,
+      "step": 4096
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6024154892017938e-06,
+      "loss": 0.7862,
+      "step": 4097
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.598309033993517e-06,
+      "loss": 0.7653,
+      "step": 4098
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5942053374003885e-06,
+      "loss": 0.7101,
+      "step": 4099
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.590104400951858e-06,
+      "loss": 0.7346,
+      "step": 4100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.586006226176352e-06,
+      "loss": 0.8019,
+      "step": 4101
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5819108146012594e-06,
+      "loss": 0.8037,
+      "step": 4102
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.577818167752939e-06,
+      "loss": 0.7987,
+      "step": 4103
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.573728287156726e-06,
+      "loss": 0.7788,
+      "step": 4104
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.569641174336922e-06,
+      "loss": 0.7591,
+      "step": 4105
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5655568308168e-06,
+      "loss": 0.7653,
+      "step": 4106
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5614752581185907e-06,
+      "loss": 0.7811,
+      "step": 4107
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5573964577635037e-06,
+      "loss": 0.7618,
+      "step": 4108
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5533204312717073e-06,
+      "loss": 0.765,
+      "step": 4109
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.549247180162343e-06,
+      "loss": 0.7932,
+      "step": 4110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.545176705953516e-06,
+      "loss": 0.738,
+      "step": 4111
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5411090101622873e-06,
+      "loss": 0.8347,
+      "step": 4112
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.537044094304697e-06,
+      "loss": 0.7083,
+      "step": 4113
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.532981959895736e-06,
+      "loss": 0.7409,
+      "step": 4114
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.52892260844937e-06,
+      "loss": 0.7431,
+      "step": 4115
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5248660414785166e-06,
+      "loss": 0.7789,
+      "step": 4116
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5208122604950637e-06,
+      "loss": 0.7658,
+      "step": 4117
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5167612670098573e-06,
+      "loss": 0.7975,
+      "step": 4118
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.512713062532707e-06,
+      "loss": 0.2747,
+      "step": 4119
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.508667648572384e-06,
+      "loss": 0.7922,
+      "step": 4120
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.504625026636609e-06,
+      "loss": 0.751,
+      "step": 4121
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.500585198232075e-06,
+      "loss": 0.7973,
+      "step": 4122
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.496548164864432e-06,
+      "loss": 0.7293,
+      "step": 4123
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.492513928038276e-06,
+      "loss": 0.7601,
+      "step": 4124
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.488482489257179e-06,
+      "loss": 0.7547,
+      "step": 4125
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4844538500236527e-06,
+      "loss": 0.7953,
+      "step": 4126
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4804280118391776e-06,
+      "loss": 0.8075,
+      "step": 4127
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4764049762041874e-06,
+      "loss": 0.6976,
+      "step": 4128
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4723847446180706e-06,
+      "loss": 0.2435,
+      "step": 4129
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4683673185791666e-06,
+      "loss": 0.7751,
+      "step": 4130
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4643526995847743e-06,
+      "loss": 0.7394,
+      "step": 4131
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4603408891311452e-06,
+      "loss": 0.8126,
+      "step": 4132
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.456331888713487e-06,
+      "loss": 0.7416,
+      "step": 4133
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4523256998259502e-06,
+      "loss": 0.8233,
+      "step": 4134
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4483223239616506e-06,
+      "loss": 0.7685,
+      "step": 4135
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4443217626126424e-06,
+      "loss": 0.7987,
+      "step": 4136
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4403240172699426e-06,
+      "loss": 0.7402,
+      "step": 4137
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4363290894235138e-06,
+      "loss": 0.7506,
+      "step": 4138
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4323369805622633e-06,
+      "loss": 0.2157,
+      "step": 4139
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4283476921740567e-06,
+      "loss": 0.7703,
+      "step": 4140
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4243612257457027e-06,
+      "loss": 0.7522,
+      "step": 4141
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4203775827629615e-06,
+      "loss": 0.7756,
+      "step": 4142
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.416396764710541e-06,
+      "loss": 0.7243,
+      "step": 4143
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4124187730720916e-06,
+      "loss": 0.7489,
+      "step": 4144
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4084436093302156e-06,
+      "loss": 0.7607,
+      "step": 4145
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4044712749664555e-06,
+      "loss": 0.7522,
+      "step": 4146
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4005017714613055e-06,
+      "loss": 0.7703,
+      "step": 4147
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.396535100294205e-06,
+      "loss": 0.7507,
+      "step": 4148
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3925712629435307e-06,
+      "loss": 0.7645,
+      "step": 4149
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.388610260886607e-06,
+      "loss": 0.7667,
+      "step": 4150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.384652095599704e-06,
+      "loss": 0.7438,
+      "step": 4151
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3806967685580364e-06,
+      "loss": 0.7976,
+      "step": 4152
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.37674428123575e-06,
+      "loss": 0.7581,
+      "step": 4153
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3727946351059427e-06,
+      "loss": 0.7045,
+      "step": 4154
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.368847831640649e-06,
+      "loss": 0.7632,
+      "step": 4155
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.364903872310851e-06,
+      "loss": 0.7439,
+      "step": 4156
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.360962758586459e-06,
+      "loss": 0.7127,
+      "step": 4157
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3570244919363285e-06,
+      "loss": 0.7536,
+      "step": 4158
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.353089073828255e-06,
+      "loss": 0.8035,
+      "step": 4159
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.349156505728972e-06,
+      "loss": 0.7216,
+      "step": 4160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3452267891041513e-06,
+      "loss": 0.7853,
+      "step": 4161
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.341299925418402e-06,
+      "loss": 0.7195,
+      "step": 4162
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3373759161352637e-06,
+      "loss": 0.2342,
+      "step": 4163
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.333454762717221e-06,
+      "loss": 0.7546,
+      "step": 4164
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.329536466625689e-06,
+      "loss": 0.7611,
+      "step": 4165
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3256210293210235e-06,
+      "loss": 0.801,
+      "step": 4166
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3217084522625043e-06,
+      "loss": 0.8035,
+      "step": 4167
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.317798736908358e-06,
+      "loss": 0.7885,
+      "step": 4168
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.31389188471573e-06,
+      "loss": 0.8012,
+      "step": 4169
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.309987897140712e-06,
+      "loss": 0.7988,
+      "step": 4170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3060867756383243e-06,
+      "loss": 0.8006,
+      "step": 4171
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.302188521662512e-06,
+      "loss": 0.7689,
+      "step": 4172
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.29829313666616e-06,
+      "loss": 0.8152,
+      "step": 4173
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.294400622101082e-06,
+      "loss": 0.7387,
+      "step": 4174
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.290510979418018e-06,
+      "loss": 0.7964,
+      "step": 4175
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2866242100666456e-06,
+      "loss": 0.2517,
+      "step": 4176
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2827403154955596e-06,
+      "loss": 0.7919,
+      "step": 4177
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2788592971522973e-06,
+      "loss": 0.242,
+      "step": 4178
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.274981156483309e-06,
+      "loss": 0.7655,
+      "step": 4179
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.271105894933987e-06,
+      "loss": 0.73,
+      "step": 4180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.267233513948638e-06,
+      "loss": 0.7803,
+      "step": 4181
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.263364014970505e-06,
+      "loss": 0.2455,
+      "step": 4182
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2594973994417503e-06,
+      "loss": 0.7881,
+      "step": 4183
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2556336688034664e-06,
+      "loss": 0.7921,
+      "step": 4184
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2517728244956704e-06,
+      "loss": 0.788,
+      "step": 4185
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.247914867957295e-06,
+      "loss": 0.7789,
+      "step": 4186
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.244059800626207e-06,
+      "loss": 0.7968,
+      "step": 4187
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2402076239391956e-06,
+      "loss": 0.7466,
+      "step": 4188
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2363583393319633e-06,
+      "loss": 0.7903,
+      "step": 4189
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2325119482391466e-06,
+      "loss": 0.7901,
+      "step": 4190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2286684520942937e-06,
+      "loss": 0.8004,
+      "step": 4191
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2248278523298796e-06,
+      "loss": 0.7055,
+      "step": 4192
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2209901503772977e-06,
+      "loss": 0.8099,
+      "step": 4193
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.217155347666866e-06,
+      "loss": 0.2556,
+      "step": 4194
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.213323445627813e-06,
+      "loss": 0.2639,
+      "step": 4195
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2094944456882937e-06,
+      "loss": 0.7206,
+      "step": 4196
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2056683492753774e-06,
+      "loss": 0.7841,
+      "step": 4197
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2018451578150545e-06,
+      "loss": 0.7827,
+      "step": 4198
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.198024872732233e-06,
+      "loss": 0.7943,
+      "step": 4199
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.194207495450733e-06,
+      "loss": 0.7315,
+      "step": 4200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.190393027393292e-06,
+      "loss": 0.7117,
+      "step": 4201
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1865814699815646e-06,
+      "loss": 0.261,
+      "step": 4202
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1827728246361257e-06,
+      "loss": 0.7507,
+      "step": 4203
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.178967092776455e-06,
+      "loss": 0.7743,
+      "step": 4204
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1751642758209524e-06,
+      "loss": 0.81,
+      "step": 4205
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.171364375186932e-06,
+      "loss": 0.7362,
+      "step": 4206
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.167567392290617e-06,
+      "loss": 0.8169,
+      "step": 4207
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.163773328547151e-06,
+      "loss": 0.7531,
+      "step": 4208
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.159982185370576e-06,
+      "loss": 0.7742,
+      "step": 4209
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1561939641738572e-06,
+      "loss": 0.7386,
+      "step": 4210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1524086663688716e-06,
+      "loss": 0.7691,
+      "step": 4211
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.148626293366394e-06,
+      "loss": 0.749,
+      "step": 4212
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.144846846576124e-06,
+      "loss": 0.7842,
+      "step": 4213
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.141070327406659e-06,
+      "loss": 0.7641,
+      "step": 4214
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.137296737265513e-06,
+      "loss": 0.782,
+      "step": 4215
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.133526077559105e-06,
+      "loss": 0.7744,
+      "step": 4216
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1297583496927645e-06,
+      "loss": 0.7676,
+      "step": 4217
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1259935550707222e-06,
+      "loss": 0.7724,
+      "step": 4218
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.122231695096121e-06,
+      "loss": 0.7561,
+      "step": 4219
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1184727711710097e-06,
+      "loss": 0.7906,
+      "step": 4220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.114716784696342e-06,
+      "loss": 0.7679,
+      "step": 4221
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1109637370719748e-06,
+      "loss": 0.8049,
+      "step": 4222
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.107213629696675e-06,
+      "loss": 0.6989,
+      "step": 4223
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1034664639681047e-06,
+      "loss": 0.7573,
+      "step": 4224
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.099722241282838e-06,
+      "loss": 0.7688,
+      "step": 4225
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0959809630363483e-06,
+      "loss": 0.752,
+      "step": 4226
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.092242630623016e-06,
+      "loss": 0.75,
+      "step": 4227
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0885072454361143e-06,
+      "loss": 0.8127,
+      "step": 4228
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0847748088678276e-06,
+      "loss": 0.7393,
+      "step": 4229
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0810453223092364e-06,
+      "loss": 0.7666,
+      "step": 4230
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.077318787150325e-06,
+      "loss": 0.7898,
+      "step": 4231
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0735952047799724e-06,
+      "loss": 0.7893,
+      "step": 4232
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.069874576585963e-06,
+      "loss": 0.8019,
+      "step": 4233
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.066156903954973e-06,
+      "loss": 0.8017,
+      "step": 4234
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.062442188272584e-06,
+      "loss": 0.7167,
+      "step": 4235
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.058730430923277e-06,
+      "loss": 0.8255,
+      "step": 4236
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0550216332904195e-06,
+      "loss": 0.7672,
+      "step": 4237
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0513157967562868e-06,
+      "loss": 0.7748,
+      "step": 4238
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.047612922702045e-06,
+      "loss": 0.779,
+      "step": 4239
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0439130125077588e-06,
+      "loss": 0.7664,
+      "step": 4240
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0402160675523895e-06,
+      "loss": 0.7694,
+      "step": 4241
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.036522089213786e-06,
+      "loss": 0.7854,
+      "step": 4242
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0328310788687023e-06,
+      "loss": 0.8251,
+      "step": 4243
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0291430378927735e-06,
+      "loss": 0.7825,
+      "step": 4244
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0254579676605426e-06,
+      "loss": 0.7849,
+      "step": 4245
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0217758695454294e-06,
+      "loss": 0.7538,
+      "step": 4246
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0180967449197586e-06,
+      "loss": 0.7898,
+      "step": 4247
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0144205951547423e-06,
+      "loss": 0.7652,
+      "step": 4248
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0107474216204847e-06,
+      "loss": 0.824,
+      "step": 4249
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0070772256859816e-06,
+      "loss": 0.7518,
+      "step": 4250
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0034100087191132e-06,
+      "loss": 0.7547,
+      "step": 4251
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9997457720866554e-06,
+      "loss": 0.7413,
+      "step": 4252
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.996084517154271e-06,
+      "loss": 0.7388,
+      "step": 4253
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.992426245286516e-06,
+      "loss": 0.7596,
+      "step": 4254
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9887709578468283e-06,
+      "loss": 0.7542,
+      "step": 4255
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9851186561975333e-06,
+      "loss": 0.8182,
+      "step": 4256
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9814693416998475e-06,
+      "loss": 0.8284,
+      "step": 4257
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9778230157138744e-06,
+      "loss": 0.2544,
+      "step": 4258
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9741796795986047e-06,
+      "loss": 0.7902,
+      "step": 4259
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.970539334711905e-06,
+      "loss": 0.7755,
+      "step": 4260
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9669019824105386e-06,
+      "loss": 0.7812,
+      "step": 4261
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.963267624050149e-06,
+      "loss": 0.7413,
+      "step": 4262
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9596362609852637e-06,
+      "loss": 0.7456,
+      "step": 4263
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9560078945692972e-06,
+      "loss": 0.7548,
+      "step": 4264
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9523825261545383e-06,
+      "loss": 0.7864,
+      "step": 4265
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.94876015709217e-06,
+      "loss": 0.7811,
+      "step": 4266
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9451407887322447e-06,
+      "loss": 0.7284,
+      "step": 4267
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.941524422423712e-06,
+      "loss": 0.7264,
+      "step": 4268
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9379110595143846e-06,
+      "loss": 0.8444,
+      "step": 4269
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9343007013509717e-06,
+      "loss": 0.258,
+      "step": 4270
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9306933492790536e-06,
+      "loss": 0.7809,
+      "step": 4271
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9270890046430933e-06,
+      "loss": 0.7875,
+      "step": 4272
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9234876687864367e-06,
+      "loss": 0.8025,
+      "step": 4273
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9198893430512976e-06,
+      "loss": 0.7435,
+      "step": 4274
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9162940287787758e-06,
+      "loss": 0.7487,
+      "step": 4275
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9127017273088523e-06,
+      "loss": 0.2709,
+      "step": 4276
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909112439980374e-06,
+      "loss": 0.2558,
+      "step": 4277
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9055261681310755e-06,
+      "loss": 0.7459,
+      "step": 4278
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9019429130975575e-06,
+      "loss": 0.2413,
+      "step": 4279
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8983626762153052e-06,
+      "loss": 0.7344,
+      "step": 4280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8947854588186733e-06,
+      "loss": 0.7664,
+      "step": 4281
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8912112622408974e-06,
+      "loss": 0.7199,
+      "step": 4282
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8876400878140776e-06,
+      "loss": 0.749,
+      "step": 4283
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.884071936869195e-06,
+      "loss": 0.7461,
+      "step": 4284
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8805068107361025e-06,
+      "loss": 0.7369,
+      "step": 4285
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8769447107435267e-06,
+      "loss": 0.6949,
+      "step": 4286
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8733856382190606e-06,
+      "loss": 0.7981,
+      "step": 4287
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8698295944891764e-06,
+      "loss": 0.7961,
+      "step": 4288
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8662765808792104e-06,
+      "loss": 0.6924,
+      "step": 4289
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8627265987133746e-06,
+      "loss": 0.7257,
+      "step": 4290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8591796493147508e-06,
+      "loss": 0.8165,
+      "step": 4291
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.855635734005291e-06,
+      "loss": 0.7841,
+      "step": 4292
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8520948541058103e-06,
+      "loss": 0.7765,
+      "step": 4293
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8485570109359997e-06,
+      "loss": 0.7427,
+      "step": 4294
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.845022205814414e-06,
+      "loss": 0.8039,
+      "step": 4295
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.841490440058482e-06,
+      "loss": 0.7304,
+      "step": 4296
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8379617149844897e-06,
+      "loss": 0.2775,
+      "step": 4297
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8344360319076004e-06,
+      "loss": 0.8068,
+      "step": 4298
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.830913392141832e-06,
+      "loss": 0.7929,
+      "step": 4299
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.827393797000079e-06,
+      "loss": 0.7885,
+      "step": 4300
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.823877247794098e-06,
+      "loss": 0.7748,
+      "step": 4301
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8203637458345048e-06,
+      "loss": 0.7252,
+      "step": 4302
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8168532924307869e-06,
+      "loss": 0.7411,
+      "step": 4303
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8133458888912924e-06,
+      "loss": 0.7005,
+      "step": 4304
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8098415365232325e-06,
+      "loss": 0.7541,
+      "step": 4305
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8063402366326832e-06,
+      "loss": 0.7718,
+      "step": 4306
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8028419905245787e-06,
+      "loss": 0.711,
+      "step": 4307
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7993467995027193e-06,
+      "loss": 0.7598,
+      "step": 4308
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7958546648697662e-06,
+      "loss": 0.7257,
+      "step": 4309
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7923655879272395e-06,
+      "loss": 0.8203,
+      "step": 4310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7888795699755169e-06,
+      "loss": 0.7431,
+      "step": 4311
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7853966123138423e-06,
+      "loss": 0.7689,
+      "step": 4312
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7819167162403173e-06,
+      "loss": 0.7816,
+      "step": 4313
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7784398830519002e-06,
+      "loss": 0.7952,
+      "step": 4314
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.774966114044412e-06,
+      "loss": 0.7536,
+      "step": 4315
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7714954105125227e-06,
+      "loss": 0.733,
+      "step": 4316
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7680277737497708e-06,
+      "loss": 0.7954,
+      "step": 4317
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.764563205048545e-06,
+      "loss": 0.7565,
+      "step": 4318
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7611017057000944e-06,
+      "loss": 0.7368,
+      "step": 4319
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7576432769945196e-06,
+      "loss": 0.7829,
+      "step": 4320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7541879202207811e-06,
+      "loss": 0.8127,
+      "step": 4321
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7507356366666907e-06,
+      "loss": 0.7849,
+      "step": 4322
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7472864276189162e-06,
+      "loss": 0.7638,
+      "step": 4323
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7438402943629852e-06,
+      "loss": 0.7384,
+      "step": 4324
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7403972381832679e-06,
+      "loss": 0.7387,
+      "step": 4325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7369572603629958e-06,
+      "loss": 0.7429,
+      "step": 4326
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7335203621842512e-06,
+      "loss": 0.2834,
+      "step": 4327
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.730086544927968e-06,
+      "loss": 0.7677,
+      "step": 4328
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.726655809873935e-06,
+      "loss": 0.7817,
+      "step": 4329
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7232281583007847e-06,
+      "loss": 0.7274,
+      "step": 4330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.71980359148601e-06,
+      "loss": 0.7091,
+      "step": 4331
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7163821107059443e-06,
+      "loss": 0.7703,
+      "step": 4332
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7129637172357805e-06,
+      "loss": 0.7838,
+      "step": 4333
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.709548412349551e-06,
+      "loss": 0.7318,
+      "step": 4334
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7061361973201452e-06,
+      "loss": 0.6999,
+      "step": 4335
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702727073419298e-06,
+      "loss": 0.8051,
+      "step": 4336
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.699321041917591e-06,
+      "loss": 0.7772,
+      "step": 4337
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6959181040844574e-06,
+      "loss": 0.818,
+      "step": 4338
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6925182611881707e-06,
+      "loss": 0.7845,
+      "step": 4339
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.689121514495855e-06,
+      "loss": 0.7681,
+      "step": 4340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6857278652734843e-06,
+      "loss": 0.7036,
+      "step": 4341
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6823373147858679e-06,
+      "loss": 0.7392,
+      "step": 4342
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6789498642966718e-06,
+      "loss": 0.2619,
+      "step": 4343
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6755655150683958e-06,
+      "loss": 0.7278,
+      "step": 4344
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6721842683623911e-06,
+      "loss": 0.753,
+      "step": 4345
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6688061254388521e-06,
+      "loss": 0.7561,
+      "step": 4346
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6654310875568157e-06,
+      "loss": 0.7068,
+      "step": 4347
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6620591559741562e-06,
+      "loss": 0.7792,
+      "step": 4348
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6586903319475979e-06,
+      "loss": 0.7782,
+      "step": 4349
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6553246167327042e-06,
+      "loss": 0.7289,
+      "step": 4350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6519620115838776e-06,
+      "loss": 0.7422,
+      "step": 4351
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6486025177543674e-06,
+      "loss": 0.2499,
+      "step": 4352
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6452461364962559e-06,
+      "loss": 0.728,
+      "step": 4353
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6418928690604663e-06,
+      "loss": 0.7398,
+      "step": 4354
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.638542716696766e-06,
+      "loss": 0.7264,
+      "step": 4355
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6351956806537595e-06,
+      "loss": 0.7667,
+      "step": 4356
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.631851762178891e-06,
+      "loss": 0.7081,
+      "step": 4357
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.628510962518436e-06,
+      "loss": 0.7675,
+      "step": 4358
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.625173282917517e-06,
+      "loss": 0.6751,
+      "step": 4359
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6218387246200874e-06,
+      "loss": 0.76,
+      "step": 4360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6185072888689436e-06,
+      "loss": 0.7707,
+      "step": 4361
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.615178976905708e-06,
+      "loss": 0.7538,
+      "step": 4362
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6118537899708465e-06,
+      "loss": 0.7745,
+      "step": 4363
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6085317293036617e-06,
+      "loss": 0.7194,
+      "step": 4364
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6052127961422826e-06,
+      "loss": 0.796,
+      "step": 4365
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6018969917236838e-06,
+      "loss": 0.7932,
+      "step": 4366
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5985843172836612e-06,
+      "loss": 0.7927,
+      "step": 4367
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5952747740568554e-06,
+      "loss": 0.8047,
+      "step": 4368
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5919683632767325e-06,
+      "loss": 0.7038,
+      "step": 4369
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.588665086175596e-06,
+      "loss": 0.713,
+      "step": 4370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5853649439845819e-06,
+      "loss": 0.7247,
+      "step": 4371
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.58206793793365e-06,
+      "loss": 0.7743,
+      "step": 4372
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5787740692515985e-06,
+      "loss": 0.8057,
+      "step": 4373
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5754833391660585e-06,
+      "loss": 0.7937,
+      "step": 4374
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5721957489034834e-06,
+      "loss": 0.8014,
+      "step": 4375
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5689112996891576e-06,
+      "loss": 0.7429,
+      "step": 4376
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.565629992747202e-06,
+      "loss": 0.7213,
+      "step": 4377
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.562351829300559e-06,
+      "loss": 0.7576,
+      "step": 4378
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5590768105710052e-06,
+      "loss": 0.7431,
+      "step": 4379
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5558049377791428e-06,
+      "loss": 0.786,
+      "step": 4380
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5525362121443965e-06,
+      "loss": 0.7917,
+      "step": 4381
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.549270634885025e-06,
+      "loss": 0.7411,
+      "step": 4382
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5460082072181115e-06,
+      "loss": 0.751,
+      "step": 4383
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5427489303595677e-06,
+      "loss": 0.7824,
+      "step": 4384
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.539492805524122e-06,
+      "loss": 0.7454,
+      "step": 4385
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5362398339253403e-06,
+      "loss": 0.7607,
+      "step": 4386
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5329900167756018e-06,
+      "loss": 0.7877,
+      "step": 4387
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5297433552861186e-06,
+      "loss": 0.7708,
+      "step": 4388
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.526499850666925e-06,
+      "loss": 0.7797,
+      "step": 4389
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5232595041268738e-06,
+      "loss": 0.7524,
+      "step": 4390
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.520022316873646e-06,
+      "loss": 0.7509,
+      "step": 4391
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.516788290113743e-06,
+      "loss": 0.8056,
+      "step": 4392
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5135574250524898e-06,
+      "loss": 0.7845,
+      "step": 4393
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.510329722894033e-06,
+      "loss": 0.7745,
+      "step": 4394
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5071051848413365e-06,
+      "loss": 0.7824,
+      "step": 4395
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5038838120961908e-06,
+      "loss": 0.7347,
+      "step": 4396
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5006656058591984e-06,
+      "loss": 0.7739,
+      "step": 4397
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4974505673297912e-06,
+      "loss": 0.7693,
+      "step": 4398
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4942386977062185e-06,
+      "loss": 0.7462,
+      "step": 4399
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4910299981855402e-06,
+      "loss": 0.6818,
+      "step": 4400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4878244699636434e-06,
+      "loss": 0.7421,
+      "step": 4401
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4846221142352313e-06,
+      "loss": 0.7331,
+      "step": 4402
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4814229321938256e-06,
+      "loss": 0.7564,
+      "step": 4403
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4782269250317593e-06,
+      "loss": 0.758,
+      "step": 4404
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4750340939401885e-06,
+      "loss": 0.735,
+      "step": 4405
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4718444401090838e-06,
+      "loss": 0.7694,
+      "step": 4406
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4686579647272337e-06,
+      "loss": 0.7589,
+      "step": 4407
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4654746689822374e-06,
+      "loss": 0.7528,
+      "step": 4408
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4622945540605083e-06,
+      "loss": 0.7301,
+      "step": 4409
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4591176211472791e-06,
+      "loss": 0.7726,
+      "step": 4410
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4559438714265973e-06,
+      "loss": 0.8063,
+      "step": 4411
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.452773306081322e-06,
+      "loss": 0.7455,
+      "step": 4412
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4496059262931206e-06,
+      "loss": 0.7598,
+      "step": 4413
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4464417332424796e-06,
+      "loss": 0.7018,
+      "step": 4414
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.443280728108697e-06,
+      "loss": 0.7409,
+      "step": 4415
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4401229120698812e-06,
+      "loss": 0.7547,
+      "step": 4416
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4369682863029533e-06,
+      "loss": 0.759,
+      "step": 4417
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4338168519836437e-06,
+      "loss": 0.7836,
+      "step": 4418
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.430668610286492e-06,
+      "loss": 0.7089,
+      "step": 4419
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4275235623848505e-06,
+      "loss": 0.7329,
+      "step": 4420
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4243817094508817e-06,
+      "loss": 0.7783,
+      "step": 4421
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4212430526555588e-06,
+      "loss": 0.2678,
+      "step": 4422
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.418107593168656e-06,
+      "loss": 0.7693,
+      "step": 4423
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4149753321587646e-06,
+      "loss": 0.7649,
+      "step": 4424
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4118462707932789e-06,
+      "loss": 0.7228,
+      "step": 4425
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4087204102384056e-06,
+      "loss": 0.8133,
+      "step": 4426
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4055977516591502e-06,
+      "loss": 0.8131,
+      "step": 4427
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4024782962193317e-06,
+      "loss": 0.7719,
+      "step": 4428
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3993620450815747e-06,
+      "loss": 0.7773,
+      "step": 4429
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3962489994073047e-06,
+      "loss": 0.7549,
+      "step": 4430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3931391603567601e-06,
+      "loss": 0.7696,
+      "step": 4431
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.390032529088976e-06,
+      "loss": 0.7533,
+      "step": 4432
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3869291067617962e-06,
+      "loss": 0.7755,
+      "step": 4433
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3838288945318712e-06,
+      "loss": 0.7747,
+      "step": 4434
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3807318935546498e-06,
+      "loss": 0.8178,
+      "step": 4435
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.377638104984389e-06,
+      "loss": 0.7752,
+      "step": 4436
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3745475299741428e-06,
+      "loss": 0.7326,
+      "step": 4437
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3714601696757713e-06,
+      "loss": 0.7665,
+      "step": 4438
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.368376025239938e-06,
+      "loss": 0.7799,
+      "step": 4439
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3652950978161029e-06,
+      "loss": 0.8034,
+      "step": 4440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3622173885525315e-06,
+      "loss": 0.7582,
+      "step": 4441
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3591428985962852e-06,
+      "loss": 0.7358,
+      "step": 4442
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3560716290932307e-06,
+      "loss": 0.8034,
+      "step": 4443
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3530035811880315e-06,
+      "loss": 0.7385,
+      "step": 4444
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.349938756024154e-06,
+      "loss": 0.7124,
+      "step": 4445
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3468771547438564e-06,
+      "loss": 0.7273,
+      "step": 4446
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3438187784882007e-06,
+      "loss": 0.7946,
+      "step": 4447
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.340763628397047e-06,
+      "loss": 0.6323,
+      "step": 4448
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3377117056090505e-06,
+      "loss": 0.2792,
+      "step": 4449
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3346630112616676e-06,
+      "loss": 0.7755,
+      "step": 4450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3316175464911463e-06,
+      "loss": 0.7256,
+      "step": 4451
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3285753124325318e-06,
+      "loss": 0.8173,
+      "step": 4452
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3255363102196683e-06,
+      "loss": 0.8047,
+      "step": 4453
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3225005409851966e-06,
+      "loss": 0.7404,
+      "step": 4454
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3194680058605447e-06,
+      "loss": 0.7112,
+      "step": 4455
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3164387059759443e-06,
+      "loss": 0.2365,
+      "step": 4456
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3134126424604155e-06,
+      "loss": 0.8143,
+      "step": 4457
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.310389816441776e-06,
+      "loss": 0.7215,
+      "step": 4458
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3073702290466362e-06,
+      "loss": 0.8754,
+      "step": 4459
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3043538814003942e-06,
+      "loss": 0.7641,
+      "step": 4460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.301340774627249e-06,
+      "loss": 0.7812,
+      "step": 4461
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2983309098501828e-06,
+      "loss": 0.8197,
+      "step": 4462
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2953242881909766e-06,
+      "loss": 0.7747,
+      "step": 4463
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2923209107702028e-06,
+      "loss": 0.7904,
+      "step": 4464
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2893207787072182e-06,
+      "loss": 0.7232,
+      "step": 4465
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.286323893120176e-06,
+      "loss": 0.2499,
+      "step": 4466
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2833302551260163e-06,
+      "loss": 0.8063,
+      "step": 4467
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.280339865840473e-06,
+      "loss": 0.7629,
+      "step": 4468
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2773527263780626e-06,
+      "loss": 0.7638,
+      "step": 4469
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2743688378520946e-06,
+      "loss": 0.8517,
+      "step": 4470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2713882013746682e-06,
+      "loss": 0.7449,
+      "step": 4471
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2684108180566678e-06,
+      "loss": 0.7496,
+      "step": 4472
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2654366890077675e-06,
+      "loss": 0.776,
+      "step": 4473
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2624658153364222e-06,
+      "loss": 0.783,
+      "step": 4474
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2594981981498823e-06,
+      "loss": 0.7475,
+      "step": 4475
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2565338385541792e-06,
+      "loss": 0.8133,
+      "step": 4476
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2535727376541352e-06,
+      "loss": 0.6995,
+      "step": 4477
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2506148965533492e-06,
+      "loss": 0.2356,
+      "step": 4478
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2476603163542145e-06,
+      "loss": 0.8034,
+      "step": 4479
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.244708998157903e-06,
+      "loss": 0.7499,
+      "step": 4480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2417609430643751e-06,
+      "loss": 0.75,
+      "step": 4481
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.238816152172373e-06,
+      "loss": 0.7601,
+      "step": 4482
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2358746265794186e-06,
+      "loss": 0.2813,
+      "step": 4483
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.232936367381825e-06,
+      "loss": 0.7245,
+      "step": 4484
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2300013756746788e-06,
+      "loss": 0.7261,
+      "step": 4485
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2270696525518567e-06,
+      "loss": 0.8099,
+      "step": 4486
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2241411991060137e-06,
+      "loss": 0.6953,
+      "step": 4487
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2212160164285836e-06,
+      "loss": 0.7653,
+      "step": 4488
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2182941056097864e-06,
+      "loss": 0.761,
+      "step": 4489
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2153754677386188e-06,
+      "loss": 0.7401,
+      "step": 4490
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2124601039028626e-06,
+      "loss": 0.7153,
+      "step": 4491
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2095480151890715e-06,
+      "loss": 0.7162,
+      "step": 4492
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2066392026825835e-06,
+      "loss": 0.7994,
+      "step": 4493
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2037336674675182e-06,
+      "loss": 0.7671,
+      "step": 4494
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2008314106267671e-06,
+      "loss": 0.7041,
+      "step": 4495
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1979324332420062e-06,
+      "loss": 0.2319,
+      "step": 4496
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1950367363936821e-06,
+      "loss": 0.7875,
+      "step": 4497
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1921443211610251e-06,
+      "loss": 0.7361,
+      "step": 4498
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1892551886220405e-06,
+      "loss": 0.2392,
+      "step": 4499
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1863693398535115e-06,
+      "loss": 0.7823,
+      "step": 4500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1834867759309953e-06,
+      "loss": 0.7812,
+      "step": 4501
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1806074979288218e-06,
+      "loss": 0.7781,
+      "step": 4502
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1777315069201034e-06,
+      "loss": 0.7696,
+      "step": 4503
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.174858803976725e-06,
+      "loss": 0.7664,
+      "step": 4504
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1719893901693401e-06,
+      "loss": 0.7224,
+      "step": 4505
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1691232665673868e-06,
+      "loss": 0.7425,
+      "step": 4506
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.166260434239066e-06,
+      "loss": 0.2453,
+      "step": 4507
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1634008942513598e-06,
+      "loss": 0.7229,
+      "step": 4508
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.16054464767002e-06,
+      "loss": 0.7946,
+      "step": 4509
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.157691695559574e-06,
+      "loss": 0.7592,
+      "step": 4510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1548420389833148e-06,
+      "loss": 0.751,
+      "step": 4511
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1519956790033138e-06,
+      "loss": 0.7604,
+      "step": 4512
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1491526166804101e-06,
+      "loss": 0.7828,
+      "step": 4513
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1463128530742173e-06,
+      "loss": 0.7742,
+      "step": 4514
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1434763892431166e-06,
+      "loss": 0.7413,
+      "step": 4515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1406432262442591e-06,
+      "loss": 0.7895,
+      "step": 4516
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1378133651335644e-06,
+      "loss": 0.72,
+      "step": 4517
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1349868069657265e-06,
+      "loss": 0.8134,
+      "step": 4518
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1321635527942055e-06,
+      "loss": 0.799,
+      "step": 4519
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1293436036712279e-06,
+      "loss": 0.8072,
+      "step": 4520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1265269606477914e-06,
+      "loss": 0.7437,
+      "step": 4521
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1237136247736623e-06,
+      "loss": 0.7591,
+      "step": 4522
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1209035970973702e-06,
+      "loss": 0.7803,
+      "step": 4523
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1180968786662194e-06,
+      "loss": 0.7832,
+      "step": 4524
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1152934705262696e-06,
+      "loss": 0.8232,
+      "step": 4525
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1124933737223553e-06,
+      "loss": 0.7978,
+      "step": 4526
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1096965892980772e-06,
+      "loss": 0.7742,
+      "step": 4527
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1069031182957935e-06,
+      "loss": 0.8032,
+      "step": 4528
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.104112961756637e-06,
+      "loss": 0.772,
+      "step": 4529
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1013261207204973e-06,
+      "loss": 0.7363,
+      "step": 4530
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0985425962260342e-06,
+      "loss": 0.7385,
+      "step": 4531
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0957623893106694e-06,
+      "loss": 0.7228,
+      "step": 4532
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0929855010105883e-06,
+      "loss": 0.7635,
+      "step": 4533
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0902119323607363e-06,
+      "loss": 0.7914,
+      "step": 4534
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0874416843948255e-06,
+      "loss": 0.7993,
+      "step": 4535
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0846747581453298e-06,
+      "loss": 0.7329,
+      "step": 4536
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0819111546434858e-06,
+      "loss": 0.7517,
+      "step": 4537
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0791508749192881e-06,
+      "loss": 0.7916,
+      "step": 4538
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.076393920001496e-06,
+      "loss": 0.7615,
+      "step": 4539
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0736402909176268e-06,
+      "loss": 0.7116,
+      "step": 4540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0708899886939606e-06,
+      "loss": 0.7307,
+      "step": 4541
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0681430143555393e-06,
+      "loss": 0.7401,
+      "step": 4542
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0653993689261577e-06,
+      "loss": 0.7758,
+      "step": 4543
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0626590534283776e-06,
+      "loss": 0.7785,
+      "step": 4544
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0599220688835142e-06,
+      "loss": 0.7879,
+      "step": 4545
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0571884163116464e-06,
+      "loss": 0.77,
+      "step": 4546
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0544580967316076e-06,
+      "loss": 0.7465,
+      "step": 4547
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0517311111609864e-06,
+      "loss": 0.7375,
+      "step": 4548
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.049007460616137e-06,
+      "loss": 0.7487,
+      "step": 4549
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0462871461121605e-06,
+      "loss": 0.7296,
+      "step": 4550
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.043570168662924e-06,
+      "loss": 0.8472,
+      "step": 4551
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0408565292810468e-06,
+      "loss": 0.7636,
+      "step": 4552
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0381462289778998e-06,
+      "loss": 0.7737,
+      "step": 4553
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0354392687636184e-06,
+      "loss": 0.7496,
+      "step": 4554
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0327356496470865e-06,
+      "loss": 0.8537,
+      "step": 4555
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0300353726359458e-06,
+      "loss": 0.7792,
+      "step": 4556
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0273384387365892e-06,
+      "loss": 0.7503,
+      "step": 4557
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0246448489541671e-06,
+      "loss": 0.7614,
+      "step": 4558
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0219546042925842e-06,
+      "loss": 0.7354,
+      "step": 4559
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.019267705754493e-06,
+      "loss": 0.7953,
+      "step": 4560
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0165841543413058e-06,
+      "loss": 0.6849,
+      "step": 4561
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.01390395105318e-06,
+      "loss": 0.7859,
+      "step": 4562
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0112270968890303e-06,
+      "loss": 0.7351,
+      "step": 4563
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.008553592846524e-06,
+      "loss": 0.7675,
+      "step": 4564
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0058834399220762e-06,
+      "loss": 0.7702,
+      "step": 4565
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0032166391108579e-06,
+      "loss": 0.7418,
+      "step": 4566
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0005531914067822e-06,
+      "loss": 0.7942,
+      "step": 4567
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.978930978025214e-07,
+      "loss": 0.7167,
+      "step": 4568
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.95236359289493e-07,
+      "loss": 0.779,
+      "step": 4569
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.925829768578666e-07,
+      "loss": 0.7784,
+      "step": 4570
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.899329514965593e-07,
+      "loss": 0.7596,
+      "step": 4571
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.872862841932341e-07,
+      "loss": 0.7636,
+      "step": 4572
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.84642975934309e-07,
+      "loss": 0.7578,
+      "step": 4573
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.820030277049452e-07,
+      "loss": 0.73,
+      "step": 4574
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.793664404890558e-07,
+      "loss": 0.8039,
+      "step": 4575
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.767332152692942e-07,
+      "loss": 0.7631,
+      "step": 4576
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.741033530270683e-07,
+      "loss": 0.762,
+      "step": 4577
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.71476854742528e-07,
+      "loss": 0.8165,
+      "step": 4578
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.688537213945726e-07,
+      "loss": 0.7909,
+      "step": 4579
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.662339539608456e-07,
+      "loss": 0.7828,
+      "step": 4580
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.636175534177338e-07,
+      "loss": 0.7591,
+      "step": 4581
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.61004520740375e-07,
+      "loss": 0.7603,
+      "step": 4582
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.583948569026447e-07,
+      "loss": 0.7834,
+      "step": 4583
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.55788562877169e-07,
+      "loss": 0.7958,
+      "step": 4584
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.531856396353123e-07,
+      "loss": 0.7687,
+      "step": 4585
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.505860881471885e-07,
+      "loss": 0.7772,
+      "step": 4586
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.479899093816503e-07,
+      "loss": 0.7176,
+      "step": 4587
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.45397104306297e-07,
+      "loss": 0.7758,
+      "step": 4588
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.428076738874692e-07,
+      "loss": 0.7302,
+      "step": 4589
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.402216190902469e-07,
+      "loss": 0.7603,
+      "step": 4590
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.376389408784559e-07,
+      "loss": 0.2483,
+      "step": 4591
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.350596402146639e-07,
+      "loss": 0.6943,
+      "step": 4592
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.324837180601743e-07,
+      "loss": 0.7438,
+      "step": 4593
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.299111753750378e-07,
+      "loss": 0.7289,
+      "step": 4594
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.273420131180399e-07,
+      "loss": 0.781,
+      "step": 4595
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.247762322467114e-07,
+      "loss": 0.7511,
+      "step": 4596
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.222138337173204e-07,
+      "loss": 0.7184,
+      "step": 4597
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.196548184848753e-07,
+      "loss": 0.793,
+      "step": 4598
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.170991875031199e-07,
+      "loss": 0.7812,
+      "step": 4599
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.145469417245423e-07,
+      "loss": 0.7862,
+      "step": 4600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.119980821003638e-07,
+      "loss": 0.7798,
+      "step": 4601
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.0945260958055e-07,
+      "loss": 0.7889,
+      "step": 4602
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.069105251137955e-07,
+      "loss": 0.75,
+      "step": 4603
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.043718296475412e-07,
+      "loss": 0.773,
+      "step": 4604
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.01836524127957e-07,
+      "loss": 0.7552,
+      "step": 4605
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.993046094999536e-07,
+      "loss": 0.7456,
+      "step": 4606
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.967760867071784e-07,
+      "loss": 0.7526,
+      "step": 4607
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.942509566920155e-07,
+      "loss": 0.8028,
+      "step": 4608
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.917292203955774e-07,
+      "loss": 0.7714,
+      "step": 4609
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.892108787577214e-07,
+      "loss": 0.6791,
+      "step": 4610
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.866959327170321e-07,
+      "loss": 0.6983,
+      "step": 4611
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.841843832108354e-07,
+      "loss": 0.2181,
+      "step": 4612
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.816762311751825e-07,
+      "loss": 0.7168,
+      "step": 4613
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.791714775448679e-07,
+      "loss": 0.7389,
+      "step": 4614
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.766701232534113e-07,
+      "loss": 0.7493,
+      "step": 4615
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.741721692330696e-07,
+      "loss": 0.7804,
+      "step": 4616
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.716776164148355e-07,
+      "loss": 0.8187,
+      "step": 4617
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.691864657284265e-07,
+      "loss": 0.7439,
+      "step": 4618
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.666987181022979e-07,
+      "loss": 0.7476,
+      "step": 4619
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.642143744636345e-07,
+      "loss": 0.8126,
+      "step": 4620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.617334357383533e-07,
+      "loss": 0.7732,
+      "step": 4621
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.592559028511017e-07,
+      "loss": 0.8393,
+      "step": 4622
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.567817767252562e-07,
+      "loss": 0.7834,
+      "step": 4623
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.543110582829272e-07,
+      "loss": 0.7461,
+      "step": 4624
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.518437484449537e-07,
+      "loss": 0.7421,
+      "step": 4625
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.493798481309035e-07,
+      "loss": 0.7728,
+      "step": 4626
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.469193582590707e-07,
+      "loss": 0.7378,
+      "step": 4627
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.44462279746483e-07,
+      "loss": 0.7593,
+      "step": 4628
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.420086135088979e-07,
+      "loss": 0.7402,
+      "step": 4629
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.395583604607948e-07,
+      "loss": 0.7576,
+      "step": 4630
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.371115215153891e-07,
+      "loss": 0.712,
+      "step": 4631
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.346680975846144e-07,
+      "loss": 0.7923,
+      "step": 4632
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.322280895791379e-07,
+      "loss": 0.7111,
+      "step": 4633
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.297914984083533e-07,
+      "loss": 0.7313,
+      "step": 4634
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.273583249803818e-07,
+      "loss": 0.7163,
+      "step": 4635
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.249285702020637e-07,
+      "loss": 0.2467,
+      "step": 4636
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.225022349789757e-07,
+      "loss": 0.7878,
+      "step": 4637
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.200793202154089e-07,
+      "loss": 0.7816,
+      "step": 4638
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.176598268143887e-07,
+      "loss": 0.8019,
+      "step": 4639
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.152437556776626e-07,
+      "loss": 0.7348,
+      "step": 4640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.128311077056982e-07,
+      "loss": 0.8255,
+      "step": 4641
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.10421883797694e-07,
+      "loss": 0.7696,
+      "step": 4642
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.080160848515683e-07,
+      "loss": 0.776,
+      "step": 4643
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.056137117639651e-07,
+      "loss": 0.7334,
+      "step": 4644
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.032147654302503e-07,
+      "loss": 0.7747,
+      "step": 4645
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.00819246744512e-07,
+      "loss": 0.8398,
+      "step": 4646
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.984271565995627e-07,
+      "loss": 0.7104,
+      "step": 4647
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.960384958869327e-07,
+      "loss": 0.7764,
+      "step": 4648
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.93653265496882e-07,
+      "loss": 0.7284,
+      "step": 4649
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.91271466318384e-07,
+      "loss": 0.8036,
+      "step": 4650
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.888930992391386e-07,
+      "loss": 0.7743,
+      "step": 4651
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.865181651455633e-07,
+      "loss": 0.743,
+      "step": 4652
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.841466649228003e-07,
+      "loss": 0.7681,
+      "step": 4653
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.817785994547078e-07,
+      "loss": 0.7326,
+      "step": 4654
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.794139696238645e-07,
+      "loss": 0.7483,
+      "step": 4655
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.77052776311571e-07,
+      "loss": 0.7643,
+      "step": 4656
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.746950203978453e-07,
+      "loss": 0.7711,
+      "step": 4657
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.723407027614227e-07,
+      "loss": 0.7743,
+      "step": 4658
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.699898242797632e-07,
+      "loss": 0.7743,
+      "step": 4659
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.67642385829036e-07,
+      "loss": 0.7234,
+      "step": 4660
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.652983882841347e-07,
+      "loss": 0.7566,
+      "step": 4661
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.629578325186693e-07,
+      "loss": 0.7191,
+      "step": 4662
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.606207194049675e-07,
+      "loss": 0.7532,
+      "step": 4663
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.582870498140704e-07,
+      "loss": 0.7878,
+      "step": 4664
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.5595682461574e-07,
+      "loss": 0.8015,
+      "step": 4665
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.536300446784517e-07,
+      "loss": 0.7673,
+      "step": 4666
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.513067108693983e-07,
+      "loss": 0.7246,
+      "step": 4667
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.489868240544906e-07,
+      "loss": 0.2979,
+      "step": 4668
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.466703850983492e-07,
+      "loss": 0.803,
+      "step": 4669
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.443573948643101e-07,
+      "loss": 0.7864,
+      "step": 4670
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.420478542144305e-07,
+      "loss": 0.7027,
+      "step": 4671
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.397417640094751e-07,
+      "loss": 0.7008,
+      "step": 4672
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.374391251089275e-07,
+      "loss": 0.727,
+      "step": 4673
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.351399383709801e-07,
+      "loss": 0.7355,
+      "step": 4674
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.32844204652543e-07,
+      "loss": 0.8019,
+      "step": 4675
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.305519248092374e-07,
+      "loss": 0.26,
+      "step": 4676
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.282630996953999e-07,
+      "loss": 0.7497,
+      "step": 4677
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.259777301640724e-07,
+      "loss": 0.826,
+      "step": 4678
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.236958170670161e-07,
+      "loss": 0.7926,
+      "step": 4679
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.214173612547037e-07,
+      "loss": 0.7635,
+      "step": 4680
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.191423635763118e-07,
+      "loss": 0.7785,
+      "step": 4681
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.168708248797395e-07,
+      "loss": 0.7469,
+      "step": 4682
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.146027460115868e-07,
+      "loss": 0.7326,
+      "step": 4683
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.123381278171681e-07,
+      "loss": 0.7307,
+      "step": 4684
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.100769711405109e-07,
+      "loss": 0.7263,
+      "step": 4685
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.078192768243486e-07,
+      "loss": 0.783,
+      "step": 4686
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.055650457101249e-07,
+      "loss": 0.7757,
+      "step": 4687
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.03314278637992e-07,
+      "loss": 0.8377,
+      "step": 4688
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.010669764468148e-07,
+      "loss": 0.7912,
+      "step": 4689
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.988231399741641e-07,
+      "loss": 0.8112,
+      "step": 4690
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.965827700563188e-07,
+      "loss": 0.7722,
+      "step": 4691
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.943458675282632e-07,
+      "loss": 0.7651,
+      "step": 4692
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.921124332236961e-07,
+      "loss": 0.7208,
+      "step": 4693
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.898824679750182e-07,
+      "loss": 0.8317,
+      "step": 4694
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.876559726133403e-07,
+      "loss": 0.7653,
+      "step": 4695
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.854329479684796e-07,
+      "loss": 0.7565,
+      "step": 4696
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.832133948689556e-07,
+      "loss": 0.7746,
+      "step": 4697
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.80997314141999e-07,
+      "loss": 0.7774,
+      "step": 4698
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.787847066135445e-07,
+      "loss": 0.8029,
+      "step": 4699
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.765755731082335e-07,
+      "loss": 0.755,
+      "step": 4700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.743699144494098e-07,
+      "loss": 0.7519,
+      "step": 4701
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.721677314591257e-07,
+      "loss": 0.8207,
+      "step": 4702
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.699690249581336e-07,
+      "loss": 0.7607,
+      "step": 4703
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.677737957658937e-07,
+      "loss": 0.7139,
+      "step": 4704
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.655820447005723e-07,
+      "loss": 0.7469,
+      "step": 4705
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.633937725790319e-07,
+      "loss": 0.7389,
+      "step": 4706
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.612089802168465e-07,
+      "loss": 0.7658,
+      "step": 4707
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.59027668428287e-07,
+      "loss": 0.8281,
+      "step": 4708
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.568498380263322e-07,
+      "loss": 0.7875,
+      "step": 4709
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.546754898226615e-07,
+      "loss": 0.7712,
+      "step": 4710
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.525046246276523e-07,
+      "loss": 0.7369,
+      "step": 4711
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.503372432503907e-07,
+      "loss": 0.751,
+      "step": 4712
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.481733464986595e-07,
+      "loss": 0.7823,
+      "step": 4713
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.460129351789468e-07,
+      "loss": 0.8042,
+      "step": 4714
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.438560100964353e-07,
+      "loss": 0.7496,
+      "step": 4715
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.417025720550141e-07,
+      "loss": 0.7396,
+      "step": 4716
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.395526218572723e-07,
+      "loss": 0.7541,
+      "step": 4717
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.374061603044967e-07,
+      "loss": 0.7205,
+      "step": 4718
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.352631881966775e-07,
+      "loss": 0.8184,
+      "step": 4719
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.33123706332498e-07,
+      "loss": 0.7571,
+      "step": 4720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.309877155093458e-07,
+      "loss": 0.7538,
+      "step": 4721
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.288552165233053e-07,
+      "loss": 0.7612,
+      "step": 4722
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.26726210169164e-07,
+      "loss": 0.7536,
+      "step": 4723
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.246006972404017e-07,
+      "loss": 0.8436,
+      "step": 4724
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.22478678529197e-07,
+      "loss": 0.7751,
+      "step": 4725
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.203601548264271e-07,
+      "loss": 0.7883,
+      "step": 4726
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.182451269216705e-07,
+      "loss": 0.7743,
+      "step": 4727
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.161335956031989e-07,
+      "loss": 0.734,
+      "step": 4728
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.140255616579794e-07,
+      "loss": 0.7487,
+      "step": 4729
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.119210258716778e-07,
+      "loss": 0.7139,
+      "step": 4730
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.098199890286583e-07,
+      "loss": 0.7485,
+      "step": 4731
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.077224519119762e-07,
+      "loss": 0.7451,
+      "step": 4732
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.056284153033875e-07,
+      "loss": 0.8079,
+      "step": 4733
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.035378799833402e-07,
+      "loss": 0.7574,
+      "step": 4734
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.014508467309765e-07,
+      "loss": 0.7659,
+      "step": 4735
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.993673163241365e-07,
+      "loss": 0.8074,
+      "step": 4736
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.972872895393522e-07,
+      "loss": 0.716,
+      "step": 4737
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.952107671518537e-07,
+      "loss": 0.7946,
+      "step": 4738
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.931377499355596e-07,
+      "loss": 0.7909,
+      "step": 4739
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.91068238663085e-07,
+      "loss": 0.7698,
+      "step": 4740
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.890022341057389e-07,
+      "loss": 0.8157,
+      "step": 4741
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.86939737033524e-07,
+      "loss": 0.7502,
+      "step": 4742
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.848807482151309e-07,
+      "loss": 0.7815,
+      "step": 4743
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.828252684179481e-07,
+      "loss": 0.7888,
+      "step": 4744
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.80773298408056e-07,
+      "loss": 0.761,
+      "step": 4745
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.787248389502221e-07,
+      "loss": 0.7956,
+      "step": 4746
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.766798908079107e-07,
+      "loss": 0.7552,
+      "step": 4747
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.746384547432738e-07,
+      "loss": 0.793,
+      "step": 4748
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.726005315171568e-07,
+      "loss": 0.7496,
+      "step": 4749
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.705661218890945e-07,
+      "loss": 0.7955,
+      "step": 4750
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.685352266173138e-07,
+      "loss": 0.712,
+      "step": 4751
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.665078464587304e-07,
+      "loss": 0.7811,
+      "step": 4752
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.644839821689485e-07,
+      "loss": 0.7587,
+      "step": 4753
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.62463634502266e-07,
+      "loss": 0.2247,
+      "step": 4754
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.604468042116673e-07,
+      "loss": 0.7645,
+      "step": 4755
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.584334920488255e-07,
+      "loss": 0.753,
+      "step": 4756
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.564236987641047e-07,
+      "loss": 0.229,
+      "step": 4757
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.544174251065549e-07,
+      "loss": 0.7716,
+      "step": 4758
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.524146718239143e-07,
+      "loss": 0.8021,
+      "step": 4759
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.504154396626138e-07,
+      "loss": 0.7505,
+      "step": 4760
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.484197293677673e-07,
+      "loss": 0.7333,
+      "step": 4761
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.464275416831766e-07,
+      "loss": 0.7448,
+      "step": 4762
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.44438877351331e-07,
+      "loss": 0.736,
+      "step": 4763
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.424537371134086e-07,
+      "loss": 0.7485,
+      "step": 4764
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.404721217092723e-07,
+      "loss": 0.7687,
+      "step": 4765
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.384940318774701e-07,
+      "loss": 0.7589,
+      "step": 4766
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.365194683552388e-07,
+      "loss": 0.8093,
+      "step": 4767
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.345484318784977e-07,
+      "loss": 0.8054,
+      "step": 4768
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.325809231818546e-07,
+      "loss": 0.7317,
+      "step": 4769
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.30616942998603e-07,
+      "loss": 0.7376,
+      "step": 4770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.286564920607162e-07,
+      "loss": 0.7443,
+      "step": 4771
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.266995710988576e-07,
+      "loss": 0.7925,
+      "step": 4772
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.247461808423726e-07,
+      "loss": 0.7081,
+      "step": 4773
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.227963220192911e-07,
+      "loss": 0.772,
+      "step": 4774
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.208499953563284e-07,
+      "loss": 0.7935,
+      "step": 4775
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.189072015788777e-07,
+      "loss": 0.7308,
+      "step": 4776
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.169679414110241e-07,
+      "loss": 0.7168,
+      "step": 4777
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.150322155755271e-07,
+      "loss": 0.833,
+      "step": 4778
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.131000247938367e-07,
+      "loss": 0.7863,
+      "step": 4779
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.111713697860787e-07,
+      "loss": 0.796,
+      "step": 4780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.092462512710649e-07,
+      "loss": 0.7551,
+      "step": 4781
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.073246699662893e-07,
+      "loss": 0.2425,
+      "step": 4782
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.054066265879254e-07,
+      "loss": 0.7204,
+      "step": 4783
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.034921218508304e-07,
+      "loss": 0.8004,
+      "step": 4784
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.015811564685402e-07,
+      "loss": 0.7833,
+      "step": 4785
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.99673731153274e-07,
+      "loss": 0.7793,
+      "step": 4786
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.977698466159286e-07,
+      "loss": 0.7016,
+      "step": 4787
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.958695035660865e-07,
+      "loss": 0.7993,
+      "step": 4788
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.939727027120055e-07,
+      "loss": 0.7465,
+      "step": 4789
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.920794447606214e-07,
+      "loss": 0.2497,
+      "step": 4790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.901897304175552e-07,
+      "loss": 0.8205,
+      "step": 4791
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.883035603871045e-07,
+      "loss": 0.7557,
+      "step": 4792
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.864209353722471e-07,
+      "loss": 0.7891,
+      "step": 4793
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.845418560746373e-07,
+      "loss": 0.7712,
+      "step": 4794
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.826663231946094e-07,
+      "loss": 0.7725,
+      "step": 4795
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.807943374311753e-07,
+      "loss": 0.784,
+      "step": 4796
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.789258994820268e-07,
+      "loss": 0.7333,
+      "step": 4797
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.770610100435313e-07,
+      "loss": 0.773,
+      "step": 4798
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.751996698107342e-07,
+      "loss": 0.7531,
+      "step": 4799
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.733418794773581e-07,
+      "loss": 0.7458,
+      "step": 4800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.714876397358026e-07,
+      "loss": 0.7602,
+      "step": 4801
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.696369512771437e-07,
+      "loss": 0.7431,
+      "step": 4802
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.677898147911364e-07,
+      "loss": 0.7874,
+      "step": 4803
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.659462309662066e-07,
+      "loss": 0.7869,
+      "step": 4804
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.641062004894614e-07,
+      "loss": 0.7698,
+      "step": 4805
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6226972404668e-07,
+      "loss": 0.7951,
+      "step": 4806
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6043680232232045e-07,
+      "loss": 0.7591,
+      "step": 4807
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5860743599951186e-07,
+      "loss": 0.7321,
+      "step": 4808
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5678162576005993e-07,
+      "loss": 0.7002,
+      "step": 4809
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.549593722844492e-07,
+      "loss": 0.7756,
+      "step": 4810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.531406762518287e-07,
+      "loss": 0.7463,
+      "step": 4811
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.513255383400339e-07,
+      "loss": 0.7952,
+      "step": 4812
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.495139592255626e-07,
+      "loss": 0.7426,
+      "step": 4813
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4770593958359365e-07,
+      "loss": 0.7974,
+      "step": 4814
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4590148008797794e-07,
+      "loss": 0.7672,
+      "step": 4815
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.441005814112387e-07,
+      "loss": 0.7997,
+      "step": 4816
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4230324422457226e-07,
+      "loss": 0.7373,
+      "step": 4817
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.405094691978451e-07,
+      "loss": 0.7596,
+      "step": 4818
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3871925699960015e-07,
+      "loss": 0.7756,
+      "step": 4819
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.369326082970515e-07,
+      "loss": 0.7822,
+      "step": 4820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.351495237560821e-07,
+      "loss": 0.808,
+      "step": 4821
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.333700040412514e-07,
+      "loss": 0.7738,
+      "step": 4822
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3159404981578446e-07,
+      "loss": 0.248,
+      "step": 4823
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.298216617415818e-07,
+      "loss": 0.6959,
+      "step": 4824
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2805284047921503e-07,
+      "loss": 0.7536,
+      "step": 4825
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.262875866879246e-07,
+      "loss": 0.8067,
+      "step": 4826
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2452590102561974e-07,
+      "loss": 0.7471,
+      "step": 4827
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.227677841488842e-07,
+      "loss": 0.7461,
+      "step": 4828
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2101323671296825e-07,
+      "loss": 0.7778,
+      "step": 4829
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.192622593717932e-07,
+      "loss": 0.7307,
+      "step": 4830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.175148527779493e-07,
+      "loss": 0.7557,
+      "step": 4831
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.157710175826979e-07,
+      "loss": 0.7016,
+      "step": 4832
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.140307544359634e-07,
+      "loss": 0.7374,
+      "step": 4833
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1229406398634595e-07,
+      "loss": 0.8247,
+      "step": 4834
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1056094688111224e-07,
+      "loss": 0.7746,
+      "step": 4835
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0883140376619334e-07,
+      "loss": 0.7722,
+      "step": 4836
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0710543528619137e-07,
+      "loss": 0.7502,
+      "step": 4837
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0538304208437853e-07,
+      "loss": 0.7031,
+      "step": 4838
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0366422480269006e-07,
+      "loss": 0.2305,
+      "step": 4839
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.019489840817314e-07,
+      "loss": 0.7739,
+      "step": 4840
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0023732056077235e-07,
+      "loss": 0.7276,
+      "step": 4841
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9852923487775143e-07,
+      "loss": 0.7982,
+      "step": 4842
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9682472766927607e-07,
+      "loss": 0.8056,
+      "step": 4843
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9512379957061367e-07,
+      "loss": 0.765,
+      "step": 4844
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9342645121570153e-07,
+      "loss": 0.8108,
+      "step": 4845
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9173268323714373e-07,
+      "loss": 0.7647,
+      "step": 4846
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.900424962662086e-07,
+      "loss": 0.265,
+      "step": 4847
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8835589093283e-07,
+      "loss": 0.7452,
+      "step": 4848
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.866728678656073e-07,
+      "loss": 0.758,
+      "step": 4849
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8499342769180214e-07,
+      "loss": 0.779,
+      "step": 4850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.83317571037346e-07,
+      "loss": 0.7517,
+      "step": 4851
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.816452985268293e-07,
+      "loss": 0.7283,
+      "step": 4852
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.799766107835112e-07,
+      "loss": 0.7663,
+      "step": 4853
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.78311508429311e-07,
+      "loss": 0.7462,
+      "step": 4854
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7664999208481547e-07,
+      "loss": 0.7501,
+      "step": 4855
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.749920623692693e-07,
+      "loss": 0.7527,
+      "step": 4856
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.733377199005872e-07,
+      "loss": 0.7257,
+      "step": 4857
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.716869652953436e-07,
+      "loss": 0.7511,
+      "step": 4858
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.700397991687743e-07,
+      "loss": 0.7546,
+      "step": 4859
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.683962221347792e-07,
+      "loss": 0.7942,
+      "step": 4860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.667562348059217e-07,
+      "loss": 0.7608,
+      "step": 4861
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.651198377934251e-07,
+      "loss": 0.7621,
+      "step": 4862
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.63487031707177e-07,
+      "loss": 0.7668,
+      "step": 4863
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.618578171557241e-07,
+      "loss": 0.7829,
+      "step": 4864
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6023219474627833e-07,
+      "loss": 0.7707,
+      "step": 4865
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.586101650847051e-07,
+      "loss": 0.7601,
+      "step": 4866
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5699172877554087e-07,
+      "loss": 0.7357,
+      "step": 4867
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5537688642197533e-07,
+      "loss": 0.7196,
+      "step": 4868
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5376563862586256e-07,
+      "loss": 0.7786,
+      "step": 4869
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.521579859877144e-07,
+      "loss": 0.7582,
+      "step": 4870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5055392910670395e-07,
+      "loss": 0.7027,
+      "step": 4871
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4895346858066723e-07,
+      "loss": 0.8023,
+      "step": 4872
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.473566050060939e-07,
+      "loss": 0.7637,
+      "step": 4873
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4576333897813674e-07,
+      "loss": 0.7632,
+      "step": 4874
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.441736710906085e-07,
+      "loss": 0.7599,
+      "step": 4875
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.425876019359764e-07,
+      "loss": 0.2478,
+      "step": 4876
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.410051321053731e-07,
+      "loss": 0.2234,
+      "step": 4877
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3942626218858353e-07,
+      "loss": 0.7824,
+      "step": 4878
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.378509927740536e-07,
+      "loss": 0.7429,
+      "step": 4879
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.362793244488882e-07,
+      "loss": 0.7406,
+      "step": 4880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.347112577988498e-07,
+      "loss": 0.684,
+      "step": 4881
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3314679340835877e-07,
+      "loss": 0.8008,
+      "step": 4882
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.315859318604875e-07,
+      "loss": 0.7435,
+      "step": 4883
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.30028673736974e-07,
+      "loss": 0.7681,
+      "step": 4884
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.284750196182085e-07,
+      "loss": 0.7648,
+      "step": 4885
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2692497008324e-07,
+      "loss": 0.8109,
+      "step": 4886
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.253785257097719e-07,
+      "loss": 0.7658,
+      "step": 4887
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2383568707416434e-07,
+      "loss": 0.2407,
+      "step": 4888
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.22296454751434e-07,
+      "loss": 0.7405,
+      "step": 4889
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.207608293152553e-07,
+      "loss": 0.8159,
+      "step": 4890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1922881133795827e-07,
+      "loss": 0.8545,
+      "step": 4891
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.177004013905227e-07,
+      "loss": 0.7023,
+      "step": 4892
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1617560004259087e-07,
+      "loss": 0.261,
+      "step": 4893
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.146544078624569e-07,
+      "loss": 0.2336,
+      "step": 4894
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.131368254170708e-07,
+      "loss": 0.7204,
+      "step": 4895
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1162285327203443e-07,
+      "loss": 0.7645,
+      "step": 4896
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1011249199160765e-07,
+      "loss": 0.734,
+      "step": 4897
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.086057421387045e-07,
+      "loss": 0.744,
+      "step": 4898
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0710260427488926e-07,
+      "loss": 0.7329,
+      "step": 4899
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0560307896038367e-07,
+      "loss": 0.7722,
+      "step": 4900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0410716675406074e-07,
+      "loss": 0.7511,
+      "step": 4901
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0261486821344887e-07,
+      "loss": 0.7328,
+      "step": 4902
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.011261838947277e-07,
+      "loss": 0.8081,
+      "step": 4903
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.996411143527322e-07,
+      "loss": 0.6814,
+      "step": 4904
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.981596601409498e-07,
+      "loss": 0.8136,
+      "step": 4905
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.966818218115164e-07,
+      "loss": 0.7467,
+      "step": 4906
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9520759991522485e-07,
+      "loss": 0.7794,
+      "step": 4907
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.937369950015201e-07,
+      "loss": 0.7611,
+      "step": 4908
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9227000761849476e-07,
+      "loss": 0.7991,
+      "step": 4909
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9080663831289823e-07,
+      "loss": 0.7314,
+      "step": 4910
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.893468876301275e-07,
+      "loss": 0.8028,
+      "step": 4911
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.87890756114233e-07,
+      "loss": 0.7479,
+      "step": 4912
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.864382443079161e-07,
+      "loss": 0.8104,
+      "step": 4913
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.849893527525316e-07,
+      "loss": 0.8179,
+      "step": 4914
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.835440819880786e-07,
+      "loss": 0.7746,
+      "step": 4915
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8210243255321066e-07,
+      "loss": 0.7822,
+      "step": 4916
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8066440498523364e-07,
+      "loss": 0.7381,
+      "step": 4917
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.792299998201009e-07,
+      "loss": 0.7776,
+      "step": 4918
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7779921759241493e-07,
+      "loss": 0.7685,
+      "step": 4919
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.763720588354313e-07,
+      "loss": 0.7732,
+      "step": 4920
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.749485240810501e-07,
+      "loss": 0.7256,
+      "step": 4921
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.735286138598259e-07,
+      "loss": 0.7807,
+      "step": 4922
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7211232870095973e-07,
+      "loss": 0.7324,
+      "step": 4923
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.706996691323016e-07,
+      "loss": 0.7254,
+      "step": 4924
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.692906356803504e-07,
+      "loss": 0.7419,
+      "step": 4925
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6788522887025494e-07,
+      "loss": 0.7623,
+      "step": 4926
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6648344922581057e-07,
+      "loss": 0.7373,
+      "step": 4927
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6508529726946287e-07,
+      "loss": 0.7094,
+      "step": 4928
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.636907735223004e-07,
+      "loss": 0.7779,
+      "step": 4929
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6229987850406755e-07,
+      "loss": 0.7698,
+      "step": 4930
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.609126127331485e-07,
+      "loss": 0.7415,
+      "step": 4931
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.595289767265796e-07,
+      "loss": 0.7324,
+      "step": 4932
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5814897100004287e-07,
+      "loss": 0.2875,
+      "step": 4933
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5677259606786686e-07,
+      "loss": 0.7393,
+      "step": 4934
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.553998524430268e-07,
+      "loss": 0.6852,
+      "step": 4935
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.540307406371467e-07,
+      "loss": 0.8012,
+      "step": 4936
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.526652611604963e-07,
+      "loss": 0.7267,
+      "step": 4937
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5130341452198726e-07,
+      "loss": 0.7942,
+      "step": 4938
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4994520122918366e-07,
+      "loss": 0.704,
+      "step": 4939
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4859062178829053e-07,
+      "loss": 0.77,
+      "step": 4940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.472396767041629e-07,
+      "loss": 0.7602,
+      "step": 4941
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.45892366480297e-07,
+      "loss": 0.8016,
+      "step": 4942
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.445486916188367e-07,
+      "loss": 0.2499,
+      "step": 4943
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4320865262057016e-07,
+      "loss": 0.7323,
+      "step": 4944
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.418722499849302e-07,
+      "loss": 0.7126,
+      "step": 4945
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.405394842099973e-07,
+      "loss": 0.8029,
+      "step": 4946
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3921035579249297e-07,
+      "loss": 0.7376,
+      "step": 4947
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3788486522778431e-07,
+      "loss": 0.7287,
+      "step": 4948
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3656301300988172e-07,
+      "loss": 0.7434,
+      "step": 4949
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3524479963143998e-07,
+      "loss": 0.7039,
+      "step": 4950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3393022558376166e-07,
+      "loss": 0.763,
+      "step": 4951
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3261929135678486e-07,
+      "loss": 0.8272,
+      "step": 4952
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3131199743909983e-07,
+      "loss": 0.7377,
+      "step": 4953
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3000834431793128e-07,
+      "loss": 0.7747,
+      "step": 4954
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2870833247915502e-07,
+      "loss": 0.2336,
+      "step": 4955
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.274119624072857e-07,
+      "loss": 0.2495,
+      "step": 4956
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2611923458548012e-07,
+      "loss": 0.7746,
+      "step": 4957
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.248301494955396e-07,
+      "loss": 0.7786,
+      "step": 4958
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2354470761790757e-07,
+      "loss": 0.8043,
+      "step": 4959
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.222629094316686e-07,
+      "loss": 0.7093,
+      "step": 4960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2098475541454944e-07,
+      "loss": 0.7923,
+      "step": 4961
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.197102460429179e-07,
+      "loss": 0.7295,
+      "step": 4962
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1843938179178847e-07,
+      "loss": 0.7185,
+      "step": 4963
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1717216313480782e-07,
+      "loss": 0.7887,
+      "step": 4964
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.159085905442737e-07,
+      "loss": 0.7607,
+      "step": 4965
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.146486644911172e-07,
+      "loss": 0.7944,
+      "step": 4966
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.133923854449149e-07,
+      "loss": 0.2534,
+      "step": 4967
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.121397538738823e-07,
+      "loss": 0.7358,
+      "step": 4968
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1089077024487703e-07,
+      "loss": 0.7499,
+      "step": 4969
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0964543502339673e-07,
+      "loss": 0.2468,
+      "step": 4970
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0840374867357572e-07,
+      "loss": 0.756,
+      "step": 4971
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0716571165819378e-07,
+      "loss": 0.78,
+      "step": 4972
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0593132443866847e-07,
+      "loss": 0.7963,
+      "step": 4973
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0470058747505516e-07,
+      "loss": 0.7806,
+      "step": 4974
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0347350122605136e-07,
+      "loss": 0.7818,
+      "step": 4975
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0225006614899233e-07,
+      "loss": 0.7546,
+      "step": 4976
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0103028269985225e-07,
+      "loss": 0.8421,
+      "step": 4977
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9981415133324744e-07,
+      "loss": 0.7534,
+      "step": 4978
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.986016725024298e-07,
+      "loss": 0.7456,
+      "step": 4979
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9739284665929005e-07,
+      "loss": 0.7611,
+      "step": 4980
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9618767425436002e-07,
+      "loss": 0.734,
+      "step": 4981
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.94986155736806e-07,
+      "loss": 0.7236,
+      "step": 4982
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9378829155443758e-07,
+      "loss": 0.2488,
+      "step": 4983
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9259408215369867e-07,
+      "loss": 0.7665,
+      "step": 4984
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.914035279796711e-07,
+      "loss": 0.869,
+      "step": 4985
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9021662947607544e-07,
+      "loss": 0.7622,
+      "step": 4986
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8903338708526898e-07,
+      "loss": 0.8031,
+      "step": 4987
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8785380124824892e-07,
+      "loss": 0.7655,
+      "step": 4988
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8667787240464586e-07,
+      "loss": 0.7584,
+      "step": 4989
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8550560099272918e-07,
+      "loss": 0.7649,
+      "step": 4990
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8433698744940609e-07,
+      "loss": 0.7703,
+      "step": 4991
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.831720322102193e-07,
+      "loss": 0.7756,
+      "step": 4992
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8201073570934924e-07,
+      "loss": 0.7145,
+      "step": 4993
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8085309837960975e-07,
+      "loss": 0.7519,
+      "step": 4994
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7969912065245565e-07,
+      "loss": 0.7209,
+      "step": 4995
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7854880295797406e-07,
+      "loss": 0.7454,
+      "step": 4996
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7740214572488755e-07,
+      "loss": 0.72,
+      "step": 4997
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.762591493805599e-07,
+      "loss": 0.7611,
+      "step": 4998
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7511981435098246e-07,
+      "loss": 0.6797,
+      "step": 4999
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7398414106078786e-07,
+      "loss": 0.7039,
+      "step": 5000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7285212993324307e-07,
+      "loss": 0.8136,
+      "step": 5001
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7172378139024836e-07,
+      "loss": 0.7985,
+      "step": 5002
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.705990958523407e-07,
+      "loss": 0.254,
+      "step": 5003
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.694780737386903e-07,
+      "loss": 0.7459,
+      "step": 5004
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6836071546710297e-07,
+      "loss": 0.7963,
+      "step": 5005
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6724702145402005e-07,
+      "loss": 0.8026,
+      "step": 5006
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6613699211451507e-07,
+      "loss": 0.7836,
+      "step": 5007
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6503062786229708e-07,
+      "loss": 0.7775,
+      "step": 5008
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6392792910970733e-07,
+      "loss": 0.7728,
+      "step": 5009
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.628288962677249e-07,
+      "loss": 0.7981,
+      "step": 5010
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.617335297459577e-07,
+      "loss": 0.7558,
+      "step": 5011
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6064182995265132e-07,
+      "loss": 0.8001,
+      "step": 5012
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5955379729468256e-07,
+      "loss": 0.7513,
+      "step": 5013
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5846943217756039e-07,
+      "loss": 0.7385,
+      "step": 5014
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5738873500543038e-07,
+      "loss": 0.6845,
+      "step": 5015
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5631170618106928e-07,
+      "loss": 0.8463,
+      "step": 5016
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5523834610588374e-07,
+      "loss": 0.8525,
+      "step": 5017
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.541686551799193e-07,
+      "loss": 0.7248,
+      "step": 5018
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5310263380184708e-07,
+      "loss": 0.7701,
+      "step": 5019
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5204028236897705e-07,
+      "loss": 0.7852,
+      "step": 5020
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5098160127724694e-07,
+      "loss": 0.762,
+      "step": 5021
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4992659092122664e-07,
+      "loss": 0.7852,
+      "step": 5022
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4887525169412055e-07,
+      "loss": 0.7951,
+      "step": 5023
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.478275839877641e-07,
+      "loss": 0.736,
+      "step": 5024
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.467835881926216e-07,
+      "loss": 0.7724,
+      "step": 5025
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.457432646977941e-07,
+      "loss": 0.777,
+      "step": 5026
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4470661389100804e-07,
+      "loss": 0.7174,
+      "step": 5027
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.436736361586266e-07,
+      "loss": 0.7439,
+      "step": 5028
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4264433188563743e-07,
+      "loss": 0.8126,
+      "step": 5029
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4161870145566693e-07,
+      "loss": 0.7385,
+      "step": 5030
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.405967452509649e-07,
+      "loss": 0.7597,
+      "step": 5031
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3957846365241667e-07,
+      "loss": 0.7267,
+      "step": 5032
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3856385703953757e-07,
+      "loss": 0.7948,
+      "step": 5033
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3755292579046952e-07,
+      "loss": 0.7787,
+      "step": 5034
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3654567028199006e-07,
+      "loss": 0.8047,
+      "step": 5035
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3554209088950222e-07,
+      "loss": 0.7269,
+      "step": 5036
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3454218798704122e-07,
+      "loss": 0.8132,
+      "step": 5037
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.335459619472712e-07,
+      "loss": 0.8169,
+      "step": 5038
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.325534131414874e-07,
+      "loss": 0.2365,
+      "step": 5039
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3156454193961278e-07,
+      "loss": 0.7538,
+      "step": 5040
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3057934871019918e-07,
+      "loss": 0.8082,
+      "step": 5041
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2959783382042957e-07,
+      "loss": 0.7239,
+      "step": 5042
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2861999763611466e-07,
+      "loss": 0.7703,
+      "step": 5043
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2764584052169737e-07,
+      "loss": 0.7551,
+      "step": 5044
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2667536284024284e-07,
+      "loss": 0.725,
+      "step": 5045
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2570856495345174e-07,
+      "loss": 0.7678,
+      "step": 5046
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2474544722164806e-07,
+      "loss": 0.7779,
+      "step": 5047
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.23786010003788e-07,
+      "loss": 0.7552,
+      "step": 5048
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.228302536574544e-07,
+      "loss": 0.8032,
+      "step": 5049
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2187817853885675e-07,
+      "loss": 0.752,
+      "step": 5050
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.209297850028357e-07,
+      "loss": 0.7841,
+      "step": 5051
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1998507340285737e-07,
+      "loss": 0.728,
+      "step": 5052
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1904404409101678e-07,
+      "loss": 0.7907,
+      "step": 5053
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1810669741803671e-07,
+      "loss": 0.7561,
+      "step": 5054
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1717303373326438e-07,
+      "loss": 0.8005,
+      "step": 5055
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1624305338468034e-07,
+      "loss": 0.2688,
+      "step": 5056
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1531675671888621e-07,
+      "loss": 0.7493,
+      "step": 5057
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1439414408111471e-07,
+      "loss": 0.7687,
+      "step": 5058
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1347521581522413e-07,
+      "loss": 0.2356,
+      "step": 5059
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.125599722636983e-07,
+      "loss": 0.7611,
+      "step": 5060
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1164841376765211e-07,
+      "loss": 0.7256,
+      "step": 5061
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1074054066682049e-07,
+      "loss": 0.7888,
+      "step": 5062
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0983635329957054e-07,
+      "loss": 0.7324,
+      "step": 5063
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0893585200289269e-07,
+      "loss": 0.7879,
+      "step": 5064
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0803903711240405e-07,
+      "loss": 0.7457,
+      "step": 5065
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0714590896234944e-07,
+      "loss": 0.7634,
+      "step": 5066
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0625646788559707e-07,
+      "loss": 0.7766,
+      "step": 5067
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0537071421364286e-07,
+      "loss": 0.7289,
+      "step": 5068
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0448864827660721e-07,
+      "loss": 0.7154,
+      "step": 5069
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0361027040323824e-07,
+      "loss": 0.8032,
+      "step": 5070
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.027355809209063e-07,
+      "loss": 0.7333,
+      "step": 5071
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.018645801556084e-07,
+      "loss": 0.7112,
+      "step": 5072
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0099726843196933e-07,
+      "loss": 0.7435,
+      "step": 5073
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0013364607323383e-07,
+      "loss": 0.7375,
+      "step": 5074
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.927371340127667e-08,
+      "loss": 0.8052,
+      "step": 5075
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.84174707365948e-08,
+      "loss": 0.2728,
+      "step": 5076
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.756491839831073e-08,
+      "loss": 0.7275,
+      "step": 5077
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.671605670417139e-08,
+      "loss": 0.7723,
+      "step": 5078
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.587088597054595e-08,
+      "loss": 0.7855,
+      "step": 5079
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.502940651243242e-08,
+      "loss": 0.7682,
+      "step": 5080
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.419161864345106e-08,
+      "loss": 0.7577,
+      "step": 5081
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.33575226758443e-08,
+      "loss": 0.7694,
+      "step": 5082
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.252711892048128e-08,
+      "loss": 0.7757,
+      "step": 5083
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.170040768685329e-08,
+      "loss": 0.7531,
+      "step": 5084
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.08773892830761e-08,
+      "loss": 0.7473,
+      "step": 5085
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.0058064015891e-08,
+      "loss": 0.7703,
+      "step": 5086
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.924243219065931e-08,
+      "loss": 0.7272,
+      "step": 5087
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.843049411136784e-08,
+      "loss": 0.7428,
+      "step": 5088
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.762225008062675e-08,
+      "loss": 0.7803,
+      "step": 5089
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.681770039966842e-08,
+      "loss": 0.8194,
+      "step": 5090
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.601684536835076e-08,
+      "loss": 0.7145,
+      "step": 5091
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.521968528515057e-08,
+      "loss": 0.7826,
+      "step": 5092
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.44262204471713e-08,
+      "loss": 0.8225,
+      "step": 5093
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.363645115013863e-08,
+      "loss": 0.261,
+      "step": 5094
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.285037768839932e-08,
+      "loss": 0.2463,
+      "step": 5095
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.206800035492346e-08,
+      "loss": 0.7921,
+      "step": 5096
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.128931944130225e-08,
+      "loss": 0.7992,
+      "step": 5097
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.051433523775243e-08,
+      "loss": 0.7781,
+      "step": 5098
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.974304803311072e-08,
+      "loss": 0.7723,
+      "step": 5099
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.89754581148372e-08,
+      "loss": 0.7604,
+      "step": 5100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.821156576901079e-08,
+      "loss": 0.8047,
+      "step": 5101
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.745137128033598e-08,
+      "loss": 0.7661,
+      "step": 5102
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.669487493213834e-08,
+      "loss": 0.7587,
+      "step": 5103
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.594207700636457e-08,
+      "loss": 0.7739,
+      "step": 5104
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.519297778358136e-08,
+      "loss": 0.7618,
+      "step": 5105
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.444757754297982e-08,
+      "loss": 0.7568,
+      "step": 5106
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.370587656237105e-08,
+      "loss": 0.7562,
+      "step": 5107
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.296787511818725e-08,
+      "loss": 0.793,
+      "step": 5108
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.223357348548287e-08,
+      "loss": 0.7147,
+      "step": 5109
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.15029719379312e-08,
+      "loss": 0.7572,
+      "step": 5110
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.077607074782777e-08,
+      "loss": 0.2468,
+      "step": 5111
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.005287018609031e-08,
+      "loss": 0.8068,
+      "step": 5112
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.933337052225653e-08,
+      "loss": 0.7127,
+      "step": 5113
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.861757202448305e-08,
+      "loss": 0.7781,
+      "step": 5114
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.79054749595498e-08,
+      "loss": 0.7604,
+      "step": 5115
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.719707959285449e-08,
+      "loss": 0.7399,
+      "step": 5116
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.649238618841702e-08,
+      "loss": 0.7672,
+      "step": 5117
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.579139500887843e-08,
+      "loss": 0.7218,
+      "step": 5118
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.509410631549529e-08,
+      "loss": 0.7855,
+      "step": 5119
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.440052036815081e-08,
+      "loss": 0.7188,
+      "step": 5120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.371063742534378e-08,
+      "loss": 0.741,
+      "step": 5121
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.302445774419408e-08,
+      "loss": 0.8163,
+      "step": 5122
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.234198158044158e-08,
+      "loss": 0.7369,
+      "step": 5123
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.166320918844504e-08,
+      "loss": 0.7392,
+      "step": 5124
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.09881408211832e-08,
+      "loss": 0.7604,
+      "step": 5125
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.031677673025593e-08,
+      "loss": 0.7568,
+      "step": 5126
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.964911716587974e-08,
+      "loss": 0.7438,
+      "step": 5127
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.898516237689223e-08,
+      "loss": 0.7664,
+      "step": 5128
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.83249126107488e-08,
+      "loss": 0.8149,
+      "step": 5129
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7668368113527054e-08,
+      "loss": 0.7777,
+      "step": 5130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7015529129919036e-08,
+      "loss": 0.7615,
+      "step": 5131
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.636639590324122e-08,
+      "loss": 0.7111,
+      "step": 5132
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5720968675423426e-08,
+      "loss": 0.2569,
+      "step": 5133
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5079247687017666e-08,
+      "loss": 0.28,
+      "step": 5134
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4441233177194855e-08,
+      "loss": 0.7358,
+      "step": 5135
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.380692538374144e-08,
+      "loss": 0.7405,
+      "step": 5136
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.317632454306609e-08,
+      "loss": 0.7926,
+      "step": 5137
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.254943089019304e-08,
+      "loss": 0.7765,
+      "step": 5138
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.19262446587665e-08,
+      "loss": 0.8168,
+      "step": 5139
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1306766081048456e-08,
+      "loss": 0.7121,
+      "step": 5140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0690995387918665e-08,
+      "loss": 0.7536,
+      "step": 5141
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.007893280887577e-08,
+      "loss": 0.721,
+      "step": 5142
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9470578572035076e-08,
+      "loss": 0.7647,
+      "step": 5143
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.886593290413078e-08,
+      "loss": 0.7662,
+      "step": 5144
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.826499603051593e-08,
+      "loss": 0.734,
+      "step": 5145
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7667768175159167e-08,
+      "loss": 0.7256,
+      "step": 5146
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.707424956064688e-08,
+      "loss": 0.7197,
+      "step": 5147
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.648444040818545e-08,
+      "loss": 0.789,
+      "step": 5148
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5898340937595706e-08,
+      "loss": 0.7488,
+      "step": 5149
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.531595136731848e-08,
+      "loss": 0.7539,
+      "step": 5150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.473727191441124e-08,
+      "loss": 0.826,
+      "step": 5151
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.416230279454592e-08,
+      "loss": 0.7981,
+      "step": 5152
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3591044222015545e-08,
+      "loss": 0.7863,
+      "step": 5153
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3023496409727585e-08,
+      "loss": 0.8,
+      "step": 5154
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.24596595692095e-08,
+      "loss": 0.7838,
+      "step": 5155
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1899533910600975e-08,
+      "loss": 0.7478,
+      "step": 5156
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.134311964266391e-08,
+      "loss": 0.7517,
+      "step": 5157
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0790416972771306e-08,
+      "loss": 0.7394,
+      "step": 5158
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.024142610691839e-08,
+      "loss": 0.7736,
+      "step": 5159
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9696147249713714e-08,
+      "loss": 0.6982,
+      "step": 5160
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9154580604382486e-08,
+      "loss": 0.7116,
+      "step": 5161
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.86167263727677e-08,
+      "loss": 0.7402,
+      "step": 5162
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.808258475532678e-08,
+      "loss": 0.765,
+      "step": 5163
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.755215595113493e-08,
+      "loss": 0.7573,
+      "step": 5164
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.702544015788512e-08,
+      "loss": 0.7499,
+      "step": 5165
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.650243757188254e-08,
+      "loss": 0.7472,
+      "step": 5166
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.598314838805239e-08,
+      "loss": 0.7421,
+      "step": 5167
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.546757279993207e-08,
+      "loss": 0.7196,
+      "step": 5168
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4955710999678984e-08,
+      "loss": 0.7551,
+      "step": 5169
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.444756317806275e-08,
+      "loss": 0.7622,
+      "step": 5170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3943129524471875e-08,
+      "loss": 0.7703,
+      "step": 5171
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3442410226907085e-08,
+      "loss": 0.8127,
+      "step": 5172
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.29454054719891e-08,
+      "loss": 0.7522,
+      "step": 5173
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.245211544494975e-08,
+      "loss": 0.7154,
+      "step": 5174
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1962540329640856e-08,
+      "loss": 0.7586,
+      "step": 5175
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1476680308525354e-08,
+      "loss": 0.7136,
+      "step": 5176
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.099453556268395e-08,
+      "loss": 0.7381,
+      "step": 5177
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0516106271814005e-08,
+      "loss": 0.8016,
+      "step": 5178
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0041392614224006e-08,
+      "loss": 0.7883,
+      "step": 5179
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9570394766842425e-08,
+      "loss": 0.7703,
+      "step": 5180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9103112905208843e-08,
+      "loss": 0.7042,
+      "step": 5181
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.86395472034795e-08,
+      "loss": 0.7861,
+      "step": 5182
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.817969783442731e-08,
+      "loss": 0.8115,
+      "step": 5183
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.772356496943629e-08,
+      "loss": 0.7717,
+      "step": 5184
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7271148778508227e-08,
+      "loss": 0.2567,
+      "step": 5185
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6822449430259357e-08,
+      "loss": 0.7302,
+      "step": 5186
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.637746709192035e-08,
+      "loss": 0.7326,
+      "step": 5187
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5936201929335214e-08,
+      "loss": 0.786,
+      "step": 5188
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5498654106965727e-08,
+      "loss": 0.7802,
+      "step": 5189
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5064823787884772e-08,
+      "loss": 0.7633,
+      "step": 5190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4634711133780796e-08,
+      "loss": 0.754,
+      "step": 5191
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4208316304958902e-08,
+      "loss": 0.7324,
+      "step": 5192
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3785639460335298e-08,
+      "loss": 0.7681,
+      "step": 5193
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.336668075744175e-08,
+      "loss": 0.7849,
+      "step": 5194
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.295144035242558e-08,
+      "loss": 0.7541,
+      "step": 5195
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2539918400046325e-08,
+      "loss": 0.7618,
+      "step": 5196
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.213211505367907e-08,
+      "loss": 0.7811,
+      "step": 5197
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1728030465312243e-08,
+      "loss": 0.7787,
+      "step": 5198
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1327664785547596e-08,
+      "loss": 0.7539,
+      "step": 5199
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0931018163602435e-08,
+      "loss": 0.2464,
+      "step": 5200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0538090747307395e-08,
+      "loss": 0.7445,
+      "step": 5201
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.014888268310755e-08,
+      "loss": 0.7221,
+      "step": 5202
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9763394116059097e-08,
+      "loss": 0.7906,
+      "step": 5203
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9381625189833776e-08,
+      "loss": 0.7386,
+      "step": 5204
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9003576046718873e-08,
+      "loss": 0.7815,
+      "step": 5205
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.86292468276128e-08,
+      "loss": 0.7861,
+      "step": 5206
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.82586376720284e-08,
+      "loss": 0.7491,
+      "step": 5207
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.789174871809185e-08,
+      "loss": 0.2515,
+      "step": 5208
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.752858010254377e-08,
+      "loss": 0.7271,
+      "step": 5209
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7169131960734774e-08,
+      "loss": 0.7913,
+      "step": 5210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.681340442663437e-08,
+      "loss": 0.7172,
+      "step": 5211
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6461397632820954e-08,
+      "loss": 0.2535,
+      "step": 5212
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.611311171048735e-08,
+      "loss": 0.7805,
+      "step": 5213
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5768546789440843e-08,
+      "loss": 0.7572,
+      "step": 5214
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5427702998100925e-08,
+      "loss": 0.7884,
+      "step": 5215
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5090580463500427e-08,
+      "loss": 0.2523,
+      "step": 5216
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4757179311284398e-08,
+      "loss": 0.7226,
+      "step": 5217
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4427499665712331e-08,
+      "loss": 0.7892,
+      "step": 5218
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4101541649654826e-08,
+      "loss": 0.7628,
+      "step": 5219
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3779305384598041e-08,
+      "loss": 0.7437,
+      "step": 5220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3460790990640349e-08,
+      "loss": 0.7111,
+      "step": 5221
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3145998586490128e-08,
+      "loss": 0.7209,
+      "step": 5222
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2834928289472415e-08,
+      "loss": 0.726,
+      "step": 5223
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2527580215522251e-08,
+      "loss": 0.791,
+      "step": 5224
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2223954479188005e-08,
+      "loss": 0.7255,
+      "step": 5225
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.19240511936336e-08,
+      "loss": 0.7636,
+      "step": 5226
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1627870470630742e-08,
+      "loss": 0.2684,
+      "step": 5227
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1335412420566682e-08,
+      "loss": 0.255,
+      "step": 5228
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.104667715244201e-08,
+      "loss": 0.7447,
+      "step": 5229
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0761664773867308e-08,
+      "loss": 0.8286,
+      "step": 5230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.048037539106761e-08,
+      "loss": 0.7418,
+      "step": 5231
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0202809108879053e-08,
+      "loss": 0.8166,
+      "step": 5232
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.928966030749997e-09,
+      "loss": 0.7636,
+      "step": 5233
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.658846258744358e-09,
+      "loss": 0.7814,
+      "step": 5234
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.392449893533828e-09,
+      "loss": 0.2292,
+      "step": 5235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.129777034405651e-09,
+      "loss": 0.8442,
+      "step": 5236
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.870827779258185e-09,
+      "loss": 0.78,
+      "step": 5237
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.615602224600894e-09,
+      "loss": 0.7874,
+      "step": 5238
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.36410046555658e-09,
+      "loss": 0.7434,
+      "step": 5239
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.11632259586137e-09,
+      "loss": 0.7123,
+      "step": 5240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.872268707861396e-09,
+      "loss": 0.741,
+      "step": 5241
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.631938892516122e-09,
+      "loss": 0.7514,
+      "step": 5242
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.395333239396118e-09,
+      "loss": 0.7741,
+      "step": 5243
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.162451836685291e-09,
+      "loss": 0.8115,
+      "step": 5244
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.9332947711786556e-09,
+      "loss": 0.8352,
+      "step": 5245
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.7078621282812285e-09,
+      "loss": 0.2487,
+      "step": 5246
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.4861539920146875e-09,
+      "loss": 0.7808,
+      "step": 5247
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.268170445007382e-09,
+      "loss": 0.7594,
+      "step": 5248
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.053911568504322e-09,
+      "loss": 0.7436,
+      "step": 5249
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.84337744235719e-09,
+      "loss": 0.7839,
+      "step": 5250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.636568145034327e-09,
+      "loss": 0.7211,
+      "step": 5251
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.433483753612967e-09,
+      "loss": 0.773,
+      "step": 5252
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.2341243437836755e-09,
+      "loss": 0.7826,
+      "step": 5253
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.038489989845907e-09,
+      "loss": 0.7556,
+      "step": 5254
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.84658076471356e-09,
+      "loss": 0.7314,
+      "step": 5255
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.658396739912752e-09,
+      "loss": 0.7722,
+      "step": 5256
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.473937985578492e-09,
+      "loss": 0.7482,
+      "step": 5257
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.293204570458009e-09,
+      "loss": 0.7858,
+      "step": 5258
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.116196561911867e-09,
+      "loss": 0.7606,
+      "step": 5259
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.942914025910627e-09,
+      "loss": 0.7734,
+      "step": 5260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.773357027038182e-09,
+      "loss": 0.7653,
+      "step": 5261
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6075256284862082e-09,
+      "loss": 0.836,
+      "step": 5262
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4454198920608196e-09,
+      "loss": 0.7624,
+      "step": 5263
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2870398781803538e-09,
+      "loss": 0.7395,
+      "step": 5264
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.132385645872038e-09,
+      "loss": 0.249,
+      "step": 5265
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9814572527753218e-09,
+      "loss": 0.7696,
+      "step": 5266
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8342547551429844e-09,
+      "loss": 0.776,
+      "step": 5267
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.690778207834477e-09,
+      "loss": 0.7207,
+      "step": 5268
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5510276643270214e-09,
+      "loss": 0.7423,
+      "step": 5269
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.415003176703401e-09,
+      "loss": 0.7513,
+      "step": 5270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2827047956608395e-09,
+      "loss": 0.2667,
+      "step": 5271
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.154132570506562e-09,
+      "loss": 0.7546,
+      "step": 5272
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0292865491600145e-09,
+      "loss": 0.7613,
+      "step": 5273
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9081667781517544e-09,
+      "loss": 0.7621,
+      "step": 5274
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7907733026223396e-09,
+      "loss": 0.7637,
+      "step": 5275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6771061663245492e-09,
+      "loss": 0.7662,
+      "step": 5276
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5671654116233836e-09,
+      "loss": 0.8618,
+      "step": 5277
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4609510794916237e-09,
+      "loss": 0.7494,
+      "step": 5278
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.358463209517602e-09,
+      "loss": 0.7747,
+      "step": 5279
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2597018398974315e-09,
+      "loss": 0.7496,
+      "step": 5280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1646670074405563e-09,
+      "loss": 0.7606,
+      "step": 5281
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0733587475642016e-09,
+      "loss": 0.7751,
+      "step": 5282
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.857770943022538e-10,
+      "loss": 0.7703,
+      "step": 5283
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.019220802930495e-10,
+      "loss": 0.7548,
+      "step": 5284
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.217937367926976e-10,
+      "loss": 0.7896,
+      "step": 5285
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.453920936628666e-10,
+      "loss": 0.7934,
+      "step": 5286
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.72717179378557e-10,
+      "loss": 0.7515,
+      "step": 5287
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.037690210269898e-10,
+      "loss": 0.6935,
+      "step": 5288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.385476443053872e-10,
+      "loss": 0.7791,
+      "step": 5289
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.77053073519862e-10,
+      "loss": 0.7492,
+      "step": 5290
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.192853315909684e-10,
+      "loss": 0.7553,
+      "step": 5291
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.652444400481514e-10,
+      "loss": 0.7314,
+      "step": 5292
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1493041903307707e-10,
+      "loss": 0.2722,
+      "step": 5293
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6834328729741234e-10,
+      "loss": 0.7814,
+      "step": 5294
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.25483062203935e-10,
+      "loss": 0.7356,
+      "step": 5295
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8634975972764424e-10,
+      "loss": 0.7726,
+      "step": 5296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5094339445242967e-10,
+      "loss": 0.7501,
+      "step": 5297
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1926397957440217e-10,
+      "loss": 0.7682,
+      "step": 5298
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.131152690189382e-11,
+      "loss": 0.7517,
+      "step": 5299
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.708604685101705e-11,
+      "loss": 0.7908,
+      "step": 5300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.658754845232594e-11,
+      "loss": 0.7039,
+      "step": 5301
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.981603934304467e-11,
+      "loss": 0.8125,
+      "step": 5302
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6771525777059538e-11,
+      "loss": 0.7915,
+      "step": 5303
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.454012613816731e-12,
+      "loss": 0.2625,
+      "step": 5304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.863503327204086e-12,
+      "loss": 0.4025,
+      "step": 5305
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.5315,
+      "step": 5306
+    },
+    {
+      "epoch": 1.0,
+      "step": 5306,
+      "total_flos": 2189936477601792.0,
+      "train_loss": 0.7786206637312231,
+      "train_runtime": 66524.7745,
+      "train_samples_per_second": 10.208,
+      "train_steps_per_second": 0.08
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 5306,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 50000,
+  "total_flos": 2189936477601792.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}