diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,50137 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 8352,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-08,
+      "loss": 3.9012,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-08,
+      "loss": 3.826,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 3.8346,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-08,
+      "loss": 3.8875,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 3.6975,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 3.9202,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4e-07,
+      "loss": 3.9103,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6e-07,
+      "loss": 3.9422,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-07,
+      "loss": 3.7993,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 3.7747,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2e-07,
+      "loss": 3.9226,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 3.9575,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6e-07,
+      "loss": 3.867,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8e-07,
+      "loss": 3.9544,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 3.8482,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2e-07,
+      "loss": 3.9101,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 3.8893,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-07,
+      "loss": 3.7823,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8e-07,
+      "loss": 3.9653,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 3.909,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 3.7943,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4e-07,
+      "loss": 3.8983,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6000000000000004e-07,
+      "loss": 3.7456,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 3.8321,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 3.8506,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.2e-07,
+      "loss": 3.8112,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.4e-07,
+      "loss": 3.7831,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.6e-07,
+      "loss": 3.8869,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 3.9131,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 3.8623,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.200000000000001e-07,
+      "loss": 3.8535,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.4e-07,
+      "loss": 3.8561,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.6e-07,
+      "loss": 3.9798,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 3.7784,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 3.8178,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.2e-07,
+      "loss": 3.863,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.4e-07,
+      "loss": 3.829,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.6e-07,
+      "loss": 3.8015,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.8e-07,
+      "loss": 3.8594,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 3.7031,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.200000000000001e-07,
+      "loss": 3.8661,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.400000000000001e-07,
+      "loss": 3.8924,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.6e-07,
+      "loss": 3.7648,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.8e-07,
+      "loss": 3.781,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 3.8825,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 3.8735,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.400000000000001e-07,
+      "loss": 3.9077,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 3.918,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 3.8777,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 3.7372,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.02e-06,
+      "loss": 3.8489,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.04e-06,
+      "loss": 3.8265,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.06e-06,
+      "loss": 3.7491,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.08e-06,
+      "loss": 3.802,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1e-06,
+      "loss": 3.8788,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.12e-06,
+      "loss": 3.778,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.14e-06,
+      "loss": 3.7504,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1600000000000001e-06,
+      "loss": 3.7636,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 3.8603,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 3.8469,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2200000000000002e-06,
+      "loss": 3.8409,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2400000000000002e-06,
+      "loss": 3.7411,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.26e-06,
+      "loss": 3.7741,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.28e-06,
+      "loss": 3.7418,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3e-06,
+      "loss": 3.9089,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.32e-06,
+      "loss": 3.9168,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.34e-06,
+      "loss": 3.7965,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 3.8305,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 3.7861,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 3.7846,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.42e-06,
+      "loss": 3.6873,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.44e-06,
+      "loss": 3.8001,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.46e-06,
+      "loss": 3.8096,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.48e-06,
+      "loss": 3.9349,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5e-06,
+      "loss": 3.8528,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.52e-06,
+      "loss": 3.839,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.54e-06,
+      "loss": 3.8031,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.56e-06,
+      "loss": 3.8147,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 3.829,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 3.8047,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6200000000000002e-06,
+      "loss": 3.9245,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6400000000000002e-06,
+      "loss": 3.8139,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6600000000000002e-06,
+      "loss": 3.7993,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 3.8273,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 3.7376,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.72e-06,
+      "loss": 3.8553,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.74e-06,
+      "loss": 3.7334,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.76e-06,
+      "loss": 3.8115,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 3.8428,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 3.8625,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8200000000000002e-06,
+      "loss": 3.7768,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8400000000000002e-06,
+      "loss": 3.8158,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8600000000000002e-06,
+      "loss": 3.8057,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 3.8087,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 3.8338,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 3.7724,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.94e-06,
+      "loss": 3.8251,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9600000000000003e-06,
+      "loss": 3.681,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.98e-06,
+      "loss": 3.7942,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 3.815,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.02e-06,
+      "loss": 3.8154,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.04e-06,
+      "loss": 3.7421,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.06e-06,
+      "loss": 3.8041,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.08e-06,
+      "loss": 3.6387,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 3.776,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.12e-06,
+      "loss": 3.7258,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1400000000000003e-06,
+      "loss": 3.817,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.16e-06,
+      "loss": 3.8503,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 3.8279,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2e-06,
+      "loss": 3.8281,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2200000000000003e-06,
+      "loss": 3.8261,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.24e-06,
+      "loss": 3.7969,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2600000000000004e-06,
+      "loss": 3.6749,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.28e-06,
+      "loss": 3.9268,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 3.8233,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3200000000000002e-06,
+      "loss": 3.7592,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3400000000000005e-06,
+      "loss": 3.8424,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3600000000000003e-06,
+      "loss": 3.7684,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.38e-06,
+      "loss": 3.7392,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 3.7265,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.42e-06,
+      "loss": 3.8559,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4400000000000004e-06,
+      "loss": 3.7969,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.46e-06,
+      "loss": 3.8251,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 3.8086,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-06,
+      "loss": 3.6924,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.52e-06,
+      "loss": 3.7633,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5400000000000002e-06,
+      "loss": 3.7818,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.56e-06,
+      "loss": 3.8137,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 3.7288,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6e-06,
+      "loss": 3.7256,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6200000000000003e-06,
+      "loss": 3.9652,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.64e-06,
+      "loss": 3.7739,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6600000000000004e-06,
+      "loss": 3.8233,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.68e-06,
+      "loss": 3.6884,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 3.7522,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 3.8247,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7400000000000004e-06,
+      "loss": 3.757,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7600000000000003e-06,
+      "loss": 3.7434,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 3.7949,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 3.7857,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.82e-06,
+      "loss": 3.7667,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.84e-06,
+      "loss": 3.8499,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.86e-06,
+      "loss": 3.7775,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.88e-06,
+      "loss": 3.9018,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9e-06,
+      "loss": 3.7932,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.92e-06,
+      "loss": 3.8472,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 3.711,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.96e-06,
+      "loss": 3.7226,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 3.7707,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3e-06,
+      "loss": 3.7903,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0200000000000003e-06,
+      "loss": 3.8011,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.04e-06,
+      "loss": 3.8059,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0600000000000003e-06,
+      "loss": 3.7927,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.08e-06,
+      "loss": 3.768,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 3.5627,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.12e-06,
+      "loss": 3.8006,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1400000000000004e-06,
+      "loss": 3.7341,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1600000000000002e-06,
+      "loss": 3.7408,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 3.8441,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 3.7352,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2200000000000005e-06,
+      "loss": 3.728,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2400000000000003e-06,
+      "loss": 3.8174,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2600000000000006e-06,
+      "loss": 3.7911,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 3.6994,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 3.8212,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 3.8535,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3400000000000006e-06,
+      "loss": 3.9056,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 3.8047,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 3.8627,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 3.7443,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4200000000000007e-06,
+      "loss": 3.7661,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.44e-06,
+      "loss": 3.8005,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.46e-06,
+      "loss": 3.7889,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.48e-06,
+      "loss": 3.7514,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5e-06,
+      "loss": 3.8066,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.52e-06,
+      "loss": 3.8278,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.54e-06,
+      "loss": 3.763,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5600000000000002e-06,
+      "loss": 3.7754,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.58e-06,
+      "loss": 3.8164,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 3.7684,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.62e-06,
+      "loss": 3.8981,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6400000000000003e-06,
+      "loss": 3.7645,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.66e-06,
+      "loss": 3.7886,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 3.9392,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7e-06,
+      "loss": 3.7438,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7200000000000004e-06,
+      "loss": 3.8896,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.74e-06,
+      "loss": 3.799,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 3.8776,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 3.8734,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 3.7173,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.820000000000001e-06,
+      "loss": 3.8938,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 3.7611,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.86e-06,
+      "loss": 3.8431,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.88e-06,
+      "loss": 3.7073,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 3.8086,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 3.8451,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.94e-06,
+      "loss": 3.7564,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.96e-06,
+      "loss": 3.6668,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 3.8837,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 3.7732,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 3.8311,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.04e-06,
+      "loss": 3.8211,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.060000000000001e-06,
+      "loss": 3.7715,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.08e-06,
+      "loss": 3.8364,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1e-06,
+      "loss": 3.839,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.12e-06,
+      "loss": 3.7512,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.14e-06,
+      "loss": 3.8144,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.16e-06,
+      "loss": 3.8163,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.18e-06,
+      "loss": 3.7243,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 3.7203,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.22e-06,
+      "loss": 3.7735,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.24e-06,
+      "loss": 3.8583,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.26e-06,
+      "loss": 3.7177,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2800000000000005e-06,
+      "loss": 3.8945,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3e-06,
+      "loss": 3.7882,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.32e-06,
+      "loss": 3.8083,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.34e-06,
+      "loss": 3.6941,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.360000000000001e-06,
+      "loss": 3.8994,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.38e-06,
+      "loss": 3.7058,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4e-06,
+      "loss": 3.7686,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.42e-06,
+      "loss": 3.8213,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.440000000000001e-06,
+      "loss": 3.8949,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4600000000000005e-06,
+      "loss": 3.7542,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.48e-06,
+      "loss": 3.7217,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5e-06,
+      "loss": 3.8696,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.520000000000001e-06,
+      "loss": 3.6869,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.540000000000001e-06,
+      "loss": 3.7618,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.56e-06,
+      "loss": 3.734,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.58e-06,
+      "loss": 3.8116,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 3.6536,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.620000000000001e-06,
+      "loss": 3.6994,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 3.6915,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.66e-06,
+      "loss": 3.7601,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.680000000000001e-06,
+      "loss": 3.7837,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7e-06,
+      "loss": 3.7719,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7200000000000005e-06,
+      "loss": 3.7187,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.74e-06,
+      "loss": 3.6884,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.76e-06,
+      "loss": 3.726,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.78e-06,
+      "loss": 3.7876,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 3.7837,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8200000000000004e-06,
+      "loss": 3.9085,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.84e-06,
+      "loss": 3.7648,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.86e-06,
+      "loss": 3.8413,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 3.7507,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 3.8183,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.92e-06,
+      "loss": 3.7177,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.94e-06,
+      "loss": 3.8496,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 3.7641,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 3.8312,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5e-06,
+      "loss": 3.7127,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.02e-06,
+      "loss": 3.676,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.04e-06,
+      "loss": 3.7749,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.060000000000001e-06,
+      "loss": 3.838,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.0800000000000005e-06,
+      "loss": 3.8832,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.1e-06,
+      "loss": 3.8069,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.12e-06,
+      "loss": 3.7332,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.140000000000001e-06,
+      "loss": 3.7152,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.1600000000000006e-06,
+      "loss": 3.6925,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.18e-06,
+      "loss": 3.8659,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.2e-06,
+      "loss": 3.8052,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.220000000000001e-06,
+      "loss": 3.8298,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.240000000000001e-06,
+      "loss": 3.8399,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.2600000000000005e-06,
+      "loss": 3.7832,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.28e-06,
+      "loss": 3.8718,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.300000000000001e-06,
+      "loss": 3.7362,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.320000000000001e-06,
+      "loss": 3.7483,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.3400000000000005e-06,
+      "loss": 3.7802,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.36e-06,
+      "loss": 3.7312,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 3.7141,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 3.7551,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.420000000000001e-06,
+      "loss": 3.8991,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 3.7591,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.460000000000001e-06,
+      "loss": 3.896,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.480000000000001e-06,
+      "loss": 3.7681,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 3.7558,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.5200000000000005e-06,
+      "loss": 3.7139,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.540000000000001e-06,
+      "loss": 3.787,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.560000000000001e-06,
+      "loss": 3.8428,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 3.7902,
+      "step": 279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 3.8238,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.620000000000001e-06,
+      "loss": 3.6849,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.64e-06,
+      "loss": 3.77,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.66e-06,
+      "loss": 3.7725,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.68e-06,
+      "loss": 3.7715,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.7e-06,
+      "loss": 3.8749,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.72e-06,
+      "loss": 3.723,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.74e-06,
+      "loss": 3.7304,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.76e-06,
+      "loss": 3.7806,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.78e-06,
+      "loss": 3.7708,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.8e-06,
+      "loss": 3.8191,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.82e-06,
+      "loss": 3.7519,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.84e-06,
+      "loss": 3.7793,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.86e-06,
+      "loss": 3.8136,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 3.7467,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9e-06,
+      "loss": 3.7748,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.92e-06,
+      "loss": 3.6701,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.94e-06,
+      "loss": 3.8166,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9600000000000005e-06,
+      "loss": 3.7067,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.98e-06,
+      "loss": 3.6907,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6e-06,
+      "loss": 3.7716,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.02e-06,
+      "loss": 3.7565,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.040000000000001e-06,
+      "loss": 3.7321,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 3.8042,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.08e-06,
+      "loss": 3.7521,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.1e-06,
+      "loss": 3.8451,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.120000000000001e-06,
+      "loss": 3.7261,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.1400000000000005e-06,
+      "loss": 3.7631,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.16e-06,
+      "loss": 3.798,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.18e-06,
+      "loss": 3.782,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 3.6279,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.220000000000001e-06,
+      "loss": 3.7595,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.24e-06,
+      "loss": 3.8129,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.26e-06,
+      "loss": 3.6852,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.280000000000001e-06,
+      "loss": 3.881,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 3.7273,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.3200000000000005e-06,
+      "loss": 3.8242,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.34e-06,
+      "loss": 3.8384,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.360000000000001e-06,
+      "loss": 3.8346,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 3.7583,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 3.7467,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.42e-06,
+      "loss": 3.6747,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.440000000000001e-06,
+      "loss": 3.8129,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.460000000000001e-06,
+      "loss": 3.8244,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.480000000000001e-06,
+      "loss": 3.803,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 3.7685,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.520000000000001e-06,
+      "loss": 3.6483,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.540000000000001e-06,
+      "loss": 3.7487,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 3.869,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 3.6489,
+      "step": 329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 3.7984,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.620000000000001e-06,
+      "loss": 3.7108,
+      "step": 331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.640000000000001e-06,
+      "loss": 3.7666,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 3.8598,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.680000000000001e-06,
+      "loss": 3.8838,
+      "step": 334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 3.766,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 3.8307,
+      "step": 336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.740000000000001e-06,
+      "loss": 3.6697,
+      "step": 337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.760000000000001e-06,
+      "loss": 3.7457,
+      "step": 338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 3.7852,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 3.8357,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.820000000000001e-06,
+      "loss": 3.736,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.8400000000000014e-06,
+      "loss": 3.7328,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.860000000000001e-06,
+      "loss": 3.7202,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.88e-06,
+      "loss": 3.726,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.9e-06,
+      "loss": 3.7178,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.92e-06,
+      "loss": 3.7612,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.9400000000000005e-06,
+      "loss": 3.7723,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.96e-06,
+      "loss": 3.8617,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.98e-06,
+      "loss": 3.7863,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7e-06,
+      "loss": 3.7527,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 3.6565,
+      "step": 351
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.04e-06,
+      "loss": 3.8247,
+      "step": 352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.06e-06,
+      "loss": 3.8321,
+      "step": 353
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.08e-06,
+      "loss": 3.8355,
+      "step": 354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.100000000000001e-06,
+      "loss": 3.7383,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.1200000000000004e-06,
+      "loss": 3.7902,
+      "step": 356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.14e-06,
+      "loss": 3.7573,
+      "step": 357
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.16e-06,
+      "loss": 3.7076,
+      "step": 358
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 3.787,
+      "step": 359
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 3.863,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.22e-06,
+      "loss": 3.7535,
+      "step": 361
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.24e-06,
+      "loss": 3.8664,
+      "step": 362
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.260000000000001e-06,
+      "loss": 3.8059,
+      "step": 363
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.280000000000001e-06,
+      "loss": 3.7723,
+      "step": 364
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.3e-06,
+      "loss": 3.7766,
+      "step": 365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.32e-06,
+      "loss": 3.7882,
+      "step": 366
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.340000000000001e-06,
+      "loss": 3.7713,
+      "step": 367
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.360000000000001e-06,
+      "loss": 3.8248,
+      "step": 368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 3.8573,
+      "step": 369
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.4e-06,
+      "loss": 3.6932,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.420000000000001e-06,
+      "loss": 3.7725,
+      "step": 371
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.440000000000001e-06,
+      "loss": 3.7848,
+      "step": 372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.4600000000000006e-06,
+      "loss": 3.5942,
+      "step": 373
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.48e-06,
+      "loss": 3.8098,
+      "step": 374
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 3.7918,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.520000000000001e-06,
+      "loss": 3.8509,
+      "step": 376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.540000000000001e-06,
+      "loss": 3.7866,
+      "step": 377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 3.7287,
+      "step": 378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.58e-06,
+      "loss": 3.7783,
+      "step": 379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 3.6774,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.620000000000001e-06,
+      "loss": 3.7231,
+      "step": 381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.640000000000001e-06,
+      "loss": 3.7661,
+      "step": 382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.660000000000001e-06,
+      "loss": 3.7667,
+      "step": 383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 3.7615,
+      "step": 384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.7e-06,
+      "loss": 3.7512,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.72e-06,
+      "loss": 3.8102,
+      "step": 386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.74e-06,
+      "loss": 3.7407,
+      "step": 387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.76e-06,
+      "loss": 3.7862,
+      "step": 388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.78e-06,
+      "loss": 3.7377,
+      "step": 389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 3.6598,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.820000000000001e-06,
+      "loss": 3.6543,
+      "step": 391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.840000000000001e-06,
+      "loss": 3.8151,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 3.7528,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.88e-06,
+      "loss": 3.7736,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.9e-06,
+      "loss": 3.7685,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.92e-06,
+      "loss": 3.7336,
+      "step": 396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.94e-06,
+      "loss": 3.8301,
+      "step": 397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.960000000000002e-06,
+      "loss": 3.7547,
+      "step": 398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 3.735,
+      "step": 399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 3.72,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.020000000000001e-06,
+      "loss": 3.793,
+      "step": 401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 3.6947,
+      "step": 402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.06e-06,
+      "loss": 3.8443,
+      "step": 403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.08e-06,
+      "loss": 3.9183,
+      "step": 404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.1e-06,
+      "loss": 3.8144,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.120000000000002e-06,
+      "loss": 3.8662,
+      "step": 406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.14e-06,
+      "loss": 3.7503,
+      "step": 407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.16e-06,
+      "loss": 3.7408,
+      "step": 408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.18e-06,
+      "loss": 3.8975,
+      "step": 409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.2e-06,
+      "loss": 3.6836,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 3.7313,
+      "step": 411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.24e-06,
+      "loss": 3.818,
+      "step": 412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.26e-06,
+      "loss": 3.8352,
+      "step": 413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.28e-06,
+      "loss": 3.7318,
+      "step": 414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.3e-06,
+      "loss": 3.6935,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.32e-06,
+      "loss": 3.6974,
+      "step": 416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.34e-06,
+      "loss": 3.7522,
+      "step": 417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.36e-06,
+      "loss": 3.7027,
+      "step": 418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 3.8926,
+      "step": 419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 3.8059,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.42e-06,
+      "loss": 3.6664,
+      "step": 421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.44e-06,
+      "loss": 3.7155,
+      "step": 422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.46e-06,
+      "loss": 3.7702,
+      "step": 423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.48e-06,
+      "loss": 3.7422,
+      "step": 424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.5e-06,
+      "loss": 3.8407,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.52e-06,
+      "loss": 3.7124,
+      "step": 426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.540000000000001e-06,
+      "loss": 3.8564,
+      "step": 427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.560000000000001e-06,
+      "loss": 3.8286,
+      "step": 428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 3.6486,
+      "step": 429
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.6e-06,
+      "loss": 3.7785,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.62e-06,
+      "loss": 3.8003,
+      "step": 431
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.64e-06,
+      "loss": 3.792,
+      "step": 432
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.66e-06,
+      "loss": 3.7059,
+      "step": 433
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.68e-06,
+      "loss": 3.8339,
+      "step": 434
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.700000000000001e-06,
+      "loss": 3.6909,
+      "step": 435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.720000000000001e-06,
+      "loss": 3.8111,
+      "step": 436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.740000000000001e-06,
+      "loss": 3.7607,
+      "step": 437
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.76e-06,
+      "loss": 3.7825,
+      "step": 438
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.78e-06,
+      "loss": 3.7865,
+      "step": 439
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.8e-06,
+      "loss": 3.7218,
+      "step": 440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.82e-06,
+      "loss": 3.7289,
+      "step": 441
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.84e-06,
+      "loss": 3.7588,
+      "step": 442
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.860000000000002e-06,
+      "loss": 3.7855,
+      "step": 443
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.880000000000001e-06,
+      "loss": 3.7666,
+      "step": 444
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 3.7288,
+      "step": 445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.920000000000001e-06,
+      "loss": 3.7591,
+      "step": 446
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.94e-06,
+      "loss": 3.7688,
+      "step": 447
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.96e-06,
+      "loss": 3.6911,
+      "step": 448
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.98e-06,
+      "loss": 3.7837,
+      "step": 449
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9e-06,
+      "loss": 3.7613,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.020000000000002e-06,
+      "loss": 3.6352,
+      "step": 451
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.040000000000002e-06,
+      "loss": 3.7101,
+      "step": 452
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.060000000000001e-06,
+      "loss": 3.8283,
+      "step": 453
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.080000000000001e-06,
+      "loss": 3.8973,
+      "step": 454
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 3.7982,
+      "step": 455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.12e-06,
+      "loss": 3.6799,
+      "step": 456
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.14e-06,
+      "loss": 3.8307,
+      "step": 457
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.16e-06,
+      "loss": 3.7604,
+      "step": 458
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 3.7864,
+      "step": 459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 3.7481,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.220000000000002e-06,
+      "loss": 3.6667,
+      "step": 461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.240000000000001e-06,
+      "loss": 3.7827,
+      "step": 462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.260000000000001e-06,
+      "loss": 3.789,
+      "step": 463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 3.7419,
+      "step": 464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.3e-06,
+      "loss": 3.703,
+      "step": 465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.32e-06,
+      "loss": 3.763,
+      "step": 466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.340000000000002e-06,
+      "loss": 3.8032,
+      "step": 467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.360000000000002e-06,
+      "loss": 3.7533,
+      "step": 468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.38e-06,
+      "loss": 3.8361,
+      "step": 469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.4e-06,
+      "loss": 3.7695,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.42e-06,
+      "loss": 3.7548,
+      "step": 471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.440000000000001e-06,
+      "loss": 3.7543,
+      "step": 472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.460000000000001e-06,
+      "loss": 3.783,
+      "step": 473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.48e-06,
+      "loss": 3.7328,
+      "step": 474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.5e-06,
+      "loss": 3.8312,
+      "step": 475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.52e-06,
+      "loss": 3.7154,
+      "step": 476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.54e-06,
+      "loss": 3.7563,
+      "step": 477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.56e-06,
+      "loss": 3.7237,
+      "step": 478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.58e-06,
+      "loss": 3.8245,
+      "step": 479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 3.7352,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.620000000000001e-06,
+      "loss": 3.7525,
+      "step": 481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.640000000000001e-06,
+      "loss": 3.8083,
+      "step": 482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.66e-06,
+      "loss": 3.7884,
+      "step": 483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.68e-06,
+      "loss": 3.7871,
+      "step": 484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.7e-06,
+      "loss": 3.704,
+      "step": 485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.72e-06,
+      "loss": 3.6649,
+      "step": 486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.74e-06,
+      "loss": 3.8165,
+      "step": 487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 3.7975,
+      "step": 488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 3.6787,
+      "step": 489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 3.71,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.820000000000001e-06,
+      "loss": 3.7441,
+      "step": 491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.84e-06,
+      "loss": 3.826,
+      "step": 492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.86e-06,
+      "loss": 3.6905,
+      "step": 493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.88e-06,
+      "loss": 3.7733,
+      "step": 494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9e-06,
+      "loss": 3.7182,
+      "step": 495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 3.8161,
+      "step": 496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.940000000000001e-06,
+      "loss": 3.6731,
+      "step": 497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 3.8146,
+      "step": 498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 3.647,
+      "step": 499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1e-05,
+      "loss": 3.8151,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999999599798082e-06,
+      "loss": 3.7304,
+      "step": 501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99999839919239e-06,
+      "loss": 3.6749,
+      "step": 502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999996398183117e-06,
+      "loss": 3.809,
+      "step": 503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999993596770583e-06,
+      "loss": 3.7708,
+      "step": 504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999989994955236e-06,
+      "loss": 3.7405,
+      "step": 505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999985592737656e-06,
+      "loss": 3.7319,
+      "step": 506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999980390118543e-06,
+      "loss": 3.7865,
+      "step": 507
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999974387098732e-06,
+      "loss": 3.8395,
+      "step": 508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999967583679185e-06,
+      "loss": 3.7087,
+      "step": 509
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999959979860988e-06,
+      "loss": 3.7578,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999951575645361e-06,
+      "loss": 3.6811,
+      "step": 511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999942371033649e-06,
+      "loss": 3.833,
+      "step": 512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999932366027326e-06,
+      "loss": 3.7434,
+      "step": 513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99992156062799e-06,
+      "loss": 3.6252,
+      "step": 514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999909954837375e-06,
+      "loss": 3.7564,
+      "step": 515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999897548657335e-06,
+      "loss": 3.8616,
+      "step": 516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99988434208986e-06,
+      "loss": 3.8619,
+      "step": 517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999870335137063e-06,
+      "loss": 3.768,
+      "step": 518
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999855527801184e-06,
+      "loss": 3.7743,
+      "step": 519
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999839920084596e-06,
+      "loss": 3.7402,
+      "step": 520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999823511989795e-06,
+      "loss": 3.8434,
+      "step": 521
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99980630351941e-06,
+      "loss": 3.7691,
+      "step": 522
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999788294676195e-06,
+      "loss": 3.7337,
+      "step": 523
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999769485463033e-06,
+      "loss": 3.7757,
+      "step": 524
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999749875882934e-06,
+      "loss": 3.7336,
+      "step": 525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999729465939036e-06,
+      "loss": 3.7516,
+      "step": 526
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99970825563461e-06,
+      "loss": 3.7153,
+      "step": 527
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999686244973048e-06,
+      "loss": 3.6561,
+      "step": 528
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999663433957878e-06,
+      "loss": 3.746,
+      "step": 529
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999639822592747e-06,
+      "loss": 3.7758,
+      "step": 530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999615410881433e-06,
+      "loss": 3.7724,
+      "step": 531
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999590198827851e-06,
+      "loss": 3.718,
+      "step": 532
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99956418643603e-06,
+      "loss": 3.6746,
+      "step": 533
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99953737371014e-06,
+      "loss": 3.7448,
+      "step": 534
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999509760654468e-06,
+      "loss": 3.7897,
+      "step": 535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999481347273437e-06,
+      "loss": 3.691,
+      "step": 536
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999452133571595e-06,
+      "loss": 3.8667,
+      "step": 537
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99942211955362e-06,
+      "loss": 3.7471,
+      "step": 538
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999391305224313e-06,
+      "loss": 3.7554,
+      "step": 539
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99935969058861e-06,
+      "loss": 3.7135,
+      "step": 540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999327275651572e-06,
+      "loss": 3.6947,
+      "step": 541
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999294060418385e-06,
+      "loss": 3.6356,
+      "step": 542
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99926004489437e-06,
+      "loss": 3.8723,
+      "step": 543
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999225229084967e-06,
+      "loss": 3.8393,
+      "step": 544
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999189612995757e-06,
+      "loss": 3.7499,
+      "step": 545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999153196632435e-06,
+      "loss": 3.7577,
+      "step": 546
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999115980000831e-06,
+      "loss": 3.8125,
+      "step": 547
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999077963106906e-06,
+      "loss": 3.8297,
+      "step": 548
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999039145956744e-06,
+      "loss": 3.6906,
+      "step": 549
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99899952855656e-06,
+      "loss": 3.6871,
+      "step": 550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998959110912693e-06,
+      "loss": 3.6398,
+      "step": 551
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998917893031615e-06,
+      "loss": 3.8656,
+      "step": 552
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998875874919925e-06,
+      "loss": 3.7794,
+      "step": 553
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998833056584349e-06,
+      "loss": 3.676,
+      "step": 554
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99878943803174e-06,
+      "loss": 3.6798,
+      "step": 555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99874501926908e-06,
+      "loss": 3.7295,
+      "step": 556
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998699800303485e-06,
+      "loss": 3.7017,
+      "step": 557
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998653781142188e-06,
+      "loss": 3.7906,
+      "step": 558
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998606961792555e-06,
+      "loss": 3.9784,
+      "step": 559
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998559342262086e-06,
+      "loss": 3.7859,
+      "step": 560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9985109225584e-06,
+      "loss": 3.8223,
+      "step": 561
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998461702689249e-06,
+      "loss": 3.7204,
+      "step": 562
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998411682662513e-06,
+      "loss": 3.7921,
+      "step": 563
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998360862486199e-06,
+      "loss": 3.7294,
+      "step": 564
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998309242168443e-06,
+      "loss": 3.7662,
+      "step": 565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998256821717504e-06,
+      "loss": 3.7737,
+      "step": 566
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99820360114178e-06,
+      "loss": 3.7778,
+      "step": 567
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998149580449787e-06,
+      "loss": 3.8653,
+      "step": 568
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998094759650174e-06,
+      "loss": 3.792,
+      "step": 569
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998039138751716e-06,
+      "loss": 3.7762,
+      "step": 570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997982717763314e-06,
+      "loss": 3.6998,
+      "step": 571
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997925496694006e-06,
+      "loss": 3.8061,
+      "step": 572
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997867475552948e-06,
+      "loss": 3.8363,
+      "step": 573
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997808654349428e-06,
+      "loss": 3.7158,
+      "step": 574
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997749033092866e-06,
+      "loss": 3.7689,
+      "step": 575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9976886117928e-06,
+      "loss": 3.7385,
+      "step": 576
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997627390458905e-06,
+      "loss": 3.7412,
+      "step": 577
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997565369100983e-06,
+      "loss": 3.7583,
+      "step": 578
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99750254772896e-06,
+      "loss": 3.6986,
+      "step": 579
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997438926352896e-06,
+      "loss": 3.8152,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997374504982973e-06,
+      "loss": 3.7643,
+      "step": 581
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997309283629504e-06,
+      "loss": 3.7054,
+      "step": 582
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997243262302928e-06,
+      "loss": 3.7737,
+      "step": 583
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997176441013817e-06,
+      "loss": 3.8593,
+      "step": 584
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997108819772865e-06,
+      "loss": 3.8143,
+      "step": 585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997040398590898e-06,
+      "loss": 3.8923,
+      "step": 586
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99697117747887e-06,
+      "loss": 3.7419,
+      "step": 587
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99690115644786e-06,
+      "loss": 3.8084,
+      "step": 588
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996830335509077e-06,
+      "loss": 3.7224,
+      "step": 589
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996758714673862e-06,
+      "loss": 3.7705,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996686293953674e-06,
+      "loss": 3.7793,
+      "step": 591
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996613073360111e-06,
+      "loss": 3.765,
+      "step": 592
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996539052904893e-06,
+      "loss": 3.794,
+      "step": 593
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996464232599869e-06,
+      "loss": 3.8119,
+      "step": 594
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996388612457014e-06,
+      "loss": 3.7791,
+      "step": 595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996312192488436e-06,
+      "loss": 3.8588,
+      "step": 596
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996234972706369e-06,
+      "loss": 3.6757,
+      "step": 597
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996156953123172e-06,
+      "loss": 3.734,
+      "step": 598
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996078133751336e-06,
+      "loss": 3.8464,
+      "step": 599
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995998514603478e-06,
+      "loss": 3.7081,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995918095692342e-06,
+      "loss": 3.6299,
+      "step": 601
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995836877030806e-06,
+      "loss": 3.7386,
+      "step": 602
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995754858631868e-06,
+      "loss": 3.7168,
+      "step": 603
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995672040508656e-06,
+      "loss": 3.7595,
+      "step": 604
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995588422674432e-06,
+      "loss": 3.7581,
+      "step": 605
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995504005142578e-06,
+      "loss": 3.7893,
+      "step": 606
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995418787926609e-06,
+      "loss": 3.7934,
+      "step": 607
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995332771040167e-06,
+      "loss": 3.6586,
+      "step": 608
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995245954497022e-06,
+      "loss": 3.7783,
+      "step": 609
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99515833831107e-06,
+      "loss": 3.7567,
+      "step": 610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995069922496336e-06,
+      "loss": 3.6486,
+      "step": 611
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994980707066979e-06,
+      "loss": 3.6797,
+      "step": 612
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994890692037274e-06,
+      "loss": 3.6823,
+      "step": 613
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994799877421635e-06,
+      "loss": 3.8082,
+      "step": 614
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994708263234599e-06,
+      "loss": 3.7731,
+      "step": 615
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99461584949083e-06,
+      "loss": 3.7915,
+      "step": 616
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994522636205121e-06,
+      "loss": 3.8476,
+      "step": 617
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994428623392399e-06,
+      "loss": 3.7142,
+      "step": 618
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994333811067707e-06,
+      "loss": 3.7858,
+      "step": 619
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994238199246226e-06,
+      "loss": 3.8165,
+      "step": 620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994141787943262e-06,
+      "loss": 3.6955,
+      "step": 621
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994044577174247e-06,
+      "loss": 3.6854,
+      "step": 622
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993946566954742e-06,
+      "loss": 3.7402,
+      "step": 623
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993847757300441e-06,
+      "loss": 3.8017,
+      "step": 624
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993748148227156e-06,
+      "loss": 3.7397,
+      "step": 625
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993647739750836e-06,
+      "loss": 3.8262,
+      "step": 626
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.993546531887554e-06,
+      "loss": 3.8381,
+      "step": 627
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.993444524653509e-06,
+      "loss": 3.8592,
+      "step": 628
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.993341718065034e-06,
+      "loss": 3.7147,
+      "step": 629
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.993238112138584e-06,
+      "loss": 3.6616,
+      "step": 630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.993133706890744e-06,
+      "loss": 3.6506,
+      "step": 631
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.99302850233823e-06,
+      "loss": 3.7831,
+      "step": 632
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992922498497879e-06,
+      "loss": 3.8253,
+      "step": 633
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992815695386665e-06,
+      "loss": 3.7485,
+      "step": 634
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992708093021682e-06,
+      "loss": 3.6393,
+      "step": 635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992599691420154e-06,
+      "loss": 3.8023,
+      "step": 636
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992490490599437e-06,
+      "loss": 3.6286,
+      "step": 637
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.99238049057701e-06,
+      "loss": 3.7863,
+      "step": 638
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992269691370485e-06,
+      "loss": 3.6515,
+      "step": 639
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992158092997593e-06,
+      "loss": 3.6677,
+      "step": 640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992045695476205e-06,
+      "loss": 3.7665,
+      "step": 641
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991932498824309e-06,
+      "loss": 3.6928,
+      "step": 642
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991818503060027e-06,
+      "loss": 3.7508,
+      "step": 643
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.99170370820161e-06,
+      "loss": 3.6603,
+      "step": 644
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991588114267431e-06,
+      "loss": 3.6693,
+      "step": 645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991471721275997e-06,
+      "loss": 3.7998,
+      "step": 646
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.99135452924594e-06,
+      "loss": 3.6893,
+      "step": 647
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991236538196016e-06,
+      "loss": 3.6574,
+      "step": 648
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991117748145118e-06,
+      "loss": 3.7701,
+      "step": 649
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990998159112262e-06,
+      "loss": 3.7718,
+      "step": 650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990877771116588e-06,
+      "loss": 3.6894,
+      "step": 651
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.99075658417737e-06,
+      "loss": 3.8705,
+      "step": 652
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.99063459831401e-06,
+      "loss": 3.6244,
+      "step": 653
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.99051181354603e-06,
+      "loss": 3.6502,
+      "step": 654
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990388229893093e-06,
+      "loss": 3.6407,
+      "step": 655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990263847374976e-06,
+      "loss": 3.7809,
+      "step": 656
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990138666011593e-06,
+      "loss": 3.8987,
+      "step": 657
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990012685822983e-06,
+      "loss": 3.7676,
+      "step": 658
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98988590682931e-06,
+      "loss": 3.7932,
+      "step": 659
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989758329050874e-06,
+      "loss": 3.8242,
+      "step": 660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989629952508095e-06,
+      "loss": 3.7893,
+      "step": 661
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989500777221523e-06,
+      "loss": 3.8547,
+      "step": 662
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989370803211838e-06,
+      "loss": 3.7065,
+      "step": 663
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989240030499846e-06,
+      "loss": 3.7883,
+      "step": 664
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98910845910648e-06,
+      "loss": 3.6299,
+      "step": 665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988976089052802e-06,
+      "loss": 3.7821,
+      "step": 666
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988842920360005e-06,
+      "loss": 3.7152,
+      "step": 667
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988708953049402e-06,
+      "loss": 3.761,
+      "step": 668
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988574187142443e-06,
+      "loss": 3.743,
+      "step": 669
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988438622660699e-06,
+      "loss": 3.7687,
+      "step": 670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988302259625872e-06,
+      "loss": 3.7605,
+      "step": 671
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988165098059791e-06,
+      "loss": 3.6461,
+      "step": 672
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988027137984414e-06,
+      "loss": 3.7031,
+      "step": 673
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987888379421822e-06,
+      "loss": 3.5594,
+      "step": 674
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987748822394232e-06,
+      "loss": 3.7744,
+      "step": 675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987608466923982e-06,
+      "loss": 3.6419,
+      "step": 676
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987467313033542e-06,
+      "loss": 3.7626,
+      "step": 677
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987325360745506e-06,
+      "loss": 3.7438,
+      "step": 678
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987182610082598e-06,
+      "loss": 3.6608,
+      "step": 679
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987039061067672e-06,
+      "loss": 3.6745,
+      "step": 680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986894713723703e-06,
+      "loss": 3.8848,
+      "step": 681
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986749568073804e-06,
+      "loss": 3.7206,
+      "step": 682
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986603624141204e-06,
+      "loss": 3.7194,
+      "step": 683
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986456881949271e-06,
+      "loss": 3.6806,
+      "step": 684
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986309341521493e-06,
+      "loss": 3.7207,
+      "step": 685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986161002881489e-06,
+      "loss": 3.7959,
+      "step": 686
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986011866053005e-06,
+      "loss": 3.7066,
+      "step": 687
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985861931059914e-06,
+      "loss": 3.8465,
+      "step": 688
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985711197926218e-06,
+      "loss": 3.7559,
+      "step": 689
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985559666676047e-06,
+      "loss": 3.7531,
+      "step": 690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985407337333661e-06,
+      "loss": 3.7854,
+      "step": 691
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985254209923441e-06,
+      "loss": 3.7579,
+      "step": 692
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9851002844699e-06,
+      "loss": 3.6614,
+      "step": 693
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984945560997681e-06,
+      "loss": 3.8036,
+      "step": 694
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984790039531552e-06,
+      "loss": 3.7305,
+      "step": 695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984633720096405e-06,
+      "loss": 3.754,
+      "step": 696
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984476602717268e-06,
+      "loss": 3.6544,
+      "step": 697
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98431868741929e-06,
+      "loss": 3.7068,
+      "step": 698
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984159974227751e-06,
+      "loss": 3.7474,
+      "step": 699
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984000463168059e-06,
+      "loss": 3.7566,
+      "step": 700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983840154265749e-06,
+      "loss": 3.7917,
+      "step": 701
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983679047546482e-06,
+      "loss": 3.8712,
+      "step": 702
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983517143036047e-06,
+      "loss": 3.7049,
+      "step": 703
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983354440760363e-06,
+      "loss": 3.7366,
+      "step": 704
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983190940745476e-06,
+      "loss": 3.7349,
+      "step": 705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983026643017559e-06,
+      "loss": 3.7663,
+      "step": 706
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.982861547602913e-06,
+      "loss": 3.8129,
+      "step": 707
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.982695654527966e-06,
+      "loss": 3.8059,
+      "step": 708
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.982528963819273e-06,
+      "loss": 3.6965,
+      "step": 709
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.982361475503522e-06,
+      "loss": 3.692,
+      "step": 710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.98219318960752e-06,
+      "loss": 3.8708,
+      "step": 711
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.98202410615821e-06,
+      "loss": 3.7409,
+      "step": 712
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.98185422518266e-06,
+      "loss": 3.8269,
+      "step": 713
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.98168354670806e-06,
+      "loss": 3.7052,
+      "step": 714
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981512070761733e-06,
+      "loss": 3.9767,
+      "step": 715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.98133979737113e-06,
+      "loss": 3.7109,
+      "step": 716
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981166726563831e-06,
+      "loss": 3.737,
+      "step": 717
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.98099285836754e-06,
+      "loss": 3.8233,
+      "step": 718
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.980818192810088e-06,
+      "loss": 3.8123,
+      "step": 719
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.98064272991944e-06,
+      "loss": 3.8526,
+      "step": 720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.98046646972368e-06,
+      "loss": 3.73,
+      "step": 721
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.980289412251023e-06,
+      "loss": 3.8342,
+      "step": 722
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.980111557529816e-06,
+      "loss": 3.6799,
+      "step": 723
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97993290558853e-06,
+      "loss": 3.7523,
+      "step": 724
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.979753456455762e-06,
+      "loss": 3.7157,
+      "step": 725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.979573210160238e-06,
+      "loss": 3.6784,
+      "step": 726
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.979392166730815e-06,
+      "loss": 3.7618,
+      "step": 727
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97921032619647e-06,
+      "loss": 3.7464,
+      "step": 728
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.979027688586316e-06,
+      "loss": 3.7697,
+      "step": 729
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.978844253929587e-06,
+      "loss": 3.7449,
+      "step": 730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97866002225565e-06,
+      "loss": 3.8451,
+      "step": 731
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.978474993593996e-06,
+      "loss": 3.6618,
+      "step": 732
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.978289167974244e-06,
+      "loss": 3.851,
+      "step": 733
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97810254542614e-06,
+      "loss": 3.6797,
+      "step": 734
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.977915125979562e-06,
+      "loss": 3.6114,
+      "step": 735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.977726909664509e-06,
+      "loss": 3.7564,
+      "step": 736
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.977537896511113e-06,
+      "loss": 3.6848,
+      "step": 737
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97734808654963e-06,
+      "loss": 3.7605,
+      "step": 738
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.977157479810445e-06,
+      "loss": 3.8293,
+      "step": 739
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97696607632407e-06,
+      "loss": 3.7953,
+      "step": 740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.976773876121148e-06,
+      "loss": 3.8408,
+      "step": 741
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.976580879232443e-06,
+      "loss": 3.7138,
+      "step": 742
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.976387085688852e-06,
+      "loss": 3.662,
+      "step": 743
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.976192495521397e-06,
+      "loss": 3.7988,
+      "step": 744
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.975997108761228e-06,
+      "loss": 3.8061,
+      "step": 745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.975800925439623e-06,
+      "loss": 3.7273,
+      "step": 746
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.975603945587987e-06,
+      "loss": 3.7159,
+      "step": 747
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.975406169237853e-06,
+      "loss": 3.8255,
+      "step": 748
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.975207596420882e-06,
+      "loss": 3.6994,
+      "step": 749
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.975008227168858e-06,
+      "loss": 3.7622,
+      "step": 750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.974808061513701e-06,
+      "loss": 3.6287,
+      "step": 751
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.974607099487452e-06,
+      "loss": 3.7368,
+      "step": 752
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97440534112228e-06,
+      "loss": 3.7837,
+      "step": 753
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.974202786450483e-06,
+      "loss": 3.7829,
+      "step": 754
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.973999435504487e-06,
+      "loss": 3.9057,
+      "step": 755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.973795288316843e-06,
+      "loss": 3.755,
+      "step": 756
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.973590344920232e-06,
+      "loss": 3.7342,
+      "step": 757
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.973384605347462e-06,
+      "loss": 3.7599,
+      "step": 758
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.973178069631468e-06,
+      "loss": 3.7456,
+      "step": 759
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.972970737805312e-06,
+      "loss": 3.6559,
+      "step": 760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.972762609902182e-06,
+      "loss": 3.7222,
+      "step": 761
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.9725536859554e-06,
+      "loss": 3.6752,
+      "step": 762
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.972343965998405e-06,
+      "loss": 3.7436,
+      "step": 763
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.972133450064774e-06,
+      "loss": 3.7333,
+      "step": 764
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.971922138188201e-06,
+      "loss": 3.8312,
+      "step": 765
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97171003040252e-06,
+      "loss": 3.8126,
+      "step": 766
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97149712674168e-06,
+      "loss": 3.76,
+      "step": 767
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.971283427239764e-06,
+      "loss": 3.8325,
+      "step": 768
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.971068931930983e-06,
+      "loss": 3.7374,
+      "step": 769
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97085364084967e-06,
+      "loss": 3.7691,
+      "step": 770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.970637554030295e-06,
+      "loss": 3.7247,
+      "step": 771
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.970420671507444e-06,
+      "loss": 3.7535,
+      "step": 772
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.970202993315836e-06,
+      "loss": 3.7646,
+      "step": 773
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.969984519490318e-06,
+      "loss": 3.6848,
+      "step": 774
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.969765250065866e-06,
+      "loss": 3.7556,
+      "step": 775
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.969545185077578e-06,
+      "loss": 3.7959,
+      "step": 776
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.969324324560683e-06,
+      "loss": 3.7326,
+      "step": 777
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.969102668550535e-06,
+      "loss": 3.7005,
+      "step": 778
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.968880217082619e-06,
+      "loss": 3.7389,
+      "step": 779
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.968656970192545e-06,
+      "loss": 3.7054,
+      "step": 780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.968432927916049e-06,
+      "loss": 3.6609,
+      "step": 781
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.968208090288998e-06,
+      "loss": 3.7976,
+      "step": 782
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.967982457347382e-06,
+      "loss": 3.8099,
+      "step": 783
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.967756029127321e-06,
+      "loss": 3.8733,
+      "step": 784
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.967528805665063e-06,
+      "loss": 3.8291,
+      "step": 785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96730078699698e-06,
+      "loss": 3.7565,
+      "step": 786
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.967071973159577e-06,
+      "loss": 3.6969,
+      "step": 787
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96684236418948e-06,
+      "loss": 3.7783,
+      "step": 788
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.966611960123444e-06,
+      "loss": 3.8081,
+      "step": 789
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.966380760998356e-06,
+      "loss": 3.6725,
+      "step": 790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.966148766851223e-06,
+      "loss": 3.6063,
+      "step": 791
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.965915977719186e-06,
+      "loss": 3.7198,
+      "step": 792
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.965682393639505e-06,
+      "loss": 3.7661,
+      "step": 793
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.965448014649576e-06,
+      "loss": 3.8066,
+      "step": 794
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.96521284078692e-06,
+      "loss": 3.7455,
+      "step": 795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.964976872089181e-06,
+      "loss": 3.7408,
+      "step": 796
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.964740108594132e-06,
+      "loss": 3.7277,
+      "step": 797
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.964502550339679e-06,
+      "loss": 3.6805,
+      "step": 798
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.964264197363846e-06,
+      "loss": 3.7099,
+      "step": 799
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.964025049704791e-06,
+      "loss": 3.6611,
+      "step": 800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.963785107400794e-06,
+      "loss": 3.7364,
+      "step": 801
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.96354437049027e-06,
+      "loss": 3.8606,
+      "step": 802
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.963302839011754e-06,
+      "loss": 3.8492,
+      "step": 803
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.96306051300391e-06,
+      "loss": 3.7921,
+      "step": 804
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.96281739250553e-06,
+      "loss": 3.5436,
+      "step": 805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.962573477555532e-06,
+      "loss": 3.6371,
+      "step": 806
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.962328768192965e-06,
+      "loss": 3.7415,
+      "step": 807
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.962083264456999e-06,
+      "loss": 3.6488,
+      "step": 808
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.961836966386937e-06,
+      "loss": 3.7974,
+      "step": 809
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.961589874022203e-06,
+      "loss": 3.8077,
+      "step": 810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.961341987402356e-06,
+      "loss": 3.7509,
+      "step": 811
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.961093306567076e-06,
+      "loss": 3.7063,
+      "step": 812
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.960843831556172e-06,
+      "loss": 3.8293,
+      "step": 813
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.96059356240958e-06,
+      "loss": 3.6667,
+      "step": 814
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.960342499167363e-06,
+      "loss": 3.7511,
+      "step": 815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.960090641869713e-06,
+      "loss": 3.8521,
+      "step": 816
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.959837990556946e-06,
+      "loss": 3.8014,
+      "step": 817
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.959584545269504e-06,
+      "loss": 3.6995,
+      "step": 818
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.959330306047966e-06,
+      "loss": 3.8043,
+      "step": 819
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.959075272933026e-06,
+      "loss": 3.7707,
+      "step": 820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.958819445965509e-06,
+      "loss": 3.7671,
+      "step": 821
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95856282518637e-06,
+      "loss": 3.715,
+      "step": 822
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.958305410636687e-06,
+      "loss": 3.7014,
+      "step": 823
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.958047202357669e-06,
+      "loss": 3.6039,
+      "step": 824
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95778820039065e-06,
+      "loss": 3.7353,
+      "step": 825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.957528404777092e-06,
+      "loss": 3.7695,
+      "step": 826
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95726781555858e-06,
+      "loss": 3.743,
+      "step": 827
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.957006432776835e-06,
+      "loss": 3.7273,
+      "step": 828
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.956744256473693e-06,
+      "loss": 3.6951,
+      "step": 829
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.956481286691127e-06,
+      "loss": 3.5916,
+      "step": 830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.956217523471233e-06,
+      "loss": 3.8143,
+      "step": 831
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.955952966856235e-06,
+      "loss": 3.6547,
+      "step": 832
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.955687616888481e-06,
+      "loss": 3.8692,
+      "step": 833
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95542147361045e-06,
+      "loss": 3.6806,
+      "step": 834
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.955154537064749e-06,
+      "loss": 3.793,
+      "step": 835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.954886807294104e-06,
+      "loss": 3.7679,
+      "step": 836
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.954618284341377e-06,
+      "loss": 3.7017,
+      "step": 837
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.954348968249552e-06,
+      "loss": 3.7663,
+      "step": 838
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.954078859061742e-06,
+      "loss": 3.8668,
+      "step": 839
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.953807956821186e-06,
+      "loss": 3.7601,
+      "step": 840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.953536261571252e-06,
+      "loss": 3.8367,
+      "step": 841
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95326377335543e-06,
+      "loss": 3.7522,
+      "step": 842
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.952990492217343e-06,
+      "loss": 3.7,
+      "step": 843
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.952716418200736e-06,
+      "loss": 3.7738,
+      "step": 844
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.952441551349484e-06,
+      "loss": 3.8522,
+      "step": 845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.952165891707588e-06,
+      "loss": 3.7394,
+      "step": 846
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.951889439319175e-06,
+      "loss": 3.7701,
+      "step": 847
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.9516121942285e-06,
+      "loss": 3.7491,
+      "step": 848
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.951334156479946e-06,
+      "loss": 3.7426,
+      "step": 849
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.951055326118017e-06,
+      "loss": 3.7433,
+      "step": 850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.950775703187354e-06,
+      "loss": 3.7795,
+      "step": 851
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.950495287732718e-06,
+      "loss": 3.6999,
+      "step": 852
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.950214079798994e-06,
+      "loss": 3.7677,
+      "step": 853
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.949932079431204e-06,
+      "loss": 3.7818,
+      "step": 854
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.949649286674485e-06,
+      "loss": 3.6835,
+      "step": 855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.949365701574111e-06,
+      "loss": 3.7415,
+      "step": 856
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.949081324175478e-06,
+      "loss": 3.6369,
+      "step": 857
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.948796154524108e-06,
+      "loss": 3.7241,
+      "step": 858
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94851019266565e-06,
+      "loss": 3.8096,
+      "step": 859
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.948223438645885e-06,
+      "loss": 3.821,
+      "step": 860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.947935892510711e-06,
+      "loss": 3.776,
+      "step": 861
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.947647554306165e-06,
+      "loss": 3.7941,
+      "step": 862
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.947358424078401e-06,
+      "loss": 3.7433,
+      "step": 863
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.947068501873702e-06,
+      "loss": 3.7805,
+      "step": 864
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.946777787738482e-06,
+      "loss": 3.8257,
+      "step": 865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.946486281719276e-06,
+      "loss": 3.8239,
+      "step": 866
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94619398386275e-06,
+      "loss": 3.7264,
+      "step": 867
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.945900894215697e-06,
+      "loss": 3.6964,
+      "step": 868
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.945607012825032e-06,
+      "loss": 3.7272,
+      "step": 869
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.945312339737799e-06,
+      "loss": 3.6891,
+      "step": 870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.945016875001171e-06,
+      "loss": 3.8009,
+      "step": 871
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94472061866245e-06,
+      "loss": 3.7248,
+      "step": 872
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.944423570769056e-06,
+      "loss": 3.7764,
+      "step": 873
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.944125731368542e-06,
+      "loss": 3.7366,
+      "step": 874
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.943827100508586e-06,
+      "loss": 3.696,
+      "step": 875
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.943527678236993e-06,
+      "loss": 3.8127,
+      "step": 876
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.943227464601697e-06,
+      "loss": 3.5793,
+      "step": 877
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.942926459650752e-06,
+      "loss": 3.8221,
+      "step": 878
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.942624663432348e-06,
+      "loss": 3.7547,
+      "step": 879
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.942322075994795e-06,
+      "loss": 3.8021,
+      "step": 880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.94201869738653e-06,
+      "loss": 3.7656,
+      "step": 881
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.941714527656118e-06,
+      "loss": 3.7685,
+      "step": 882
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.941409566852255e-06,
+      "loss": 3.7945,
+      "step": 883
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.941103815023754e-06,
+      "loss": 3.7402,
+      "step": 884
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.940797272219562e-06,
+      "loss": 3.7632,
+      "step": 885
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.940489938488752e-06,
+      "loss": 3.9032,
+      "step": 886
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.940181813880521e-06,
+      "loss": 3.7354,
+      "step": 887
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.939872898444193e-06,
+      "loss": 3.7677,
+      "step": 888
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.93956319222922e-06,
+      "loss": 3.6806,
+      "step": 889
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.93925269528518e-06,
+      "loss": 3.6772,
+      "step": 890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.93894140766178e-06,
+      "loss": 3.6974,
+      "step": 891
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.938629329408846e-06,
+      "loss": 3.6691,
+      "step": 892
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.93831646057634e-06,
+      "loss": 3.769,
+      "step": 893
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.938002801214346e-06,
+      "loss": 3.6891,
+      "step": 894
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.937688351373073e-06,
+      "loss": 3.7207,
+      "step": 895
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.937373111102859e-06,
+      "loss": 3.6895,
+      "step": 896
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.937057080454168e-06,
+      "loss": 3.725,
+      "step": 897
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.936740259477592e-06,
+      "loss": 3.9493,
+      "step": 898
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.936422648223846e-06,
+      "loss": 3.6549,
+      "step": 899
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.936104246743773e-06,
+      "loss": 3.7128,
+      "step": 900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.935785055088344e-06,
+      "loss": 3.7883,
+      "step": 901
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.935465073308656e-06,
+      "loss": 3.7539,
+      "step": 902
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.93514430145593e-06,
+      "loss": 3.6796,
+      "step": 903
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.934822739581518e-06,
+      "loss": 3.7565,
+      "step": 904
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.934500387736893e-06,
+      "loss": 3.7676,
+      "step": 905
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.934177245973661e-06,
+      "loss": 3.7699,
+      "step": 906
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.933853314343549e-06,
+      "loss": 3.6246,
+      "step": 907
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.93352859289841e-06,
+      "loss": 3.6676,
+      "step": 908
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.933203081690227e-06,
+      "loss": 3.7923,
+      "step": 909
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.932876780771112e-06,
+      "loss": 3.6501,
+      "step": 910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.932549690193294e-06,
+      "loss": 3.7206,
+      "step": 911
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.932221810009136e-06,
+      "loss": 3.7201,
+      "step": 912
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.931893140271126e-06,
+      "loss": 3.6732,
+      "step": 913
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.931563681031874e-06,
+      "loss": 3.7068,
+      "step": 914
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.931233432344127e-06,
+      "loss": 3.8301,
+      "step": 915
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.930902394260746e-06,
+      "loss": 3.7443,
+      "step": 916
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.930570566834725e-06,
+      "loss": 3.6879,
+      "step": 917
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.930237950119184e-06,
+      "loss": 3.7423,
+      "step": 918
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.929904544167368e-06,
+      "loss": 3.7539,
+      "step": 919
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.92957034903265e-06,
+      "loss": 3.71,
+      "step": 920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.929235364768527e-06,
+      "loss": 3.8136,
+      "step": 921
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.928899591428624e-06,
+      "loss": 3.7136,
+      "step": 922
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.928563029066692e-06,
+      "loss": 3.7514,
+      "step": 923
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.928225677736605e-06,
+      "loss": 3.7486,
+      "step": 924
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.927887537492371e-06,
+      "loss": 3.5975,
+      "step": 925
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.927548608388119e-06,
+      "loss": 3.756,
+      "step": 926
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.927208890478105e-06,
+      "loss": 3.6783,
+      "step": 927
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.926868383816707e-06,
+      "loss": 3.8441,
+      "step": 928
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.926527088458439e-06,
+      "loss": 3.7648,
+      "step": 929
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.926185004457933e-06,
+      "loss": 3.7204,
+      "step": 930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.925842131869952e-06,
+      "loss": 3.7523,
+      "step": 931
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.92549847074938e-06,
+      "loss": 3.6363,
+      "step": 932
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.925154021151235e-06,
+      "loss": 3.7002,
+      "step": 933
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.924808783130653e-06,
+      "loss": 3.8723,
+      "step": 934
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.924462756742902e-06,
+      "loss": 3.8458,
+      "step": 935
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.924115942043372e-06,
+      "loss": 3.7354,
+      "step": 936
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.923768339087584e-06,
+      "loss": 3.7973,
+      "step": 937
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.923419947931182e-06,
+      "loss": 3.6664,
+      "step": 938
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.923070768629937e-06,
+      "loss": 3.7967,
+      "step": 939
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.922720801239744e-06,
+      "loss": 3.7488,
+      "step": 940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.922370045816627e-06,
+      "loss": 3.8542,
+      "step": 941
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.922018502416736e-06,
+      "loss": 3.7973,
+      "step": 942
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.921666171096345e-06,
+      "loss": 3.6217,
+      "step": 943
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.921313051911856e-06,
+      "loss": 3.762,
+      "step": 944
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.920959144919797e-06,
+      "loss": 3.8204,
+      "step": 945
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.920604450176823e-06,
+      "loss": 3.7561,
+      "step": 946
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.92024896773971e-06,
+      "loss": 3.6859,
+      "step": 947
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.919892697665366e-06,
+      "loss": 3.8233,
+      "step": 948
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.919535640010825e-06,
+      "loss": 3.6024,
+      "step": 949
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.919177794833241e-06,
+      "loss": 3.6499,
+      "step": 950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.918819162189902e-06,
+      "loss": 3.7713,
+      "step": 951
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.918459742138216e-06,
+      "loss": 3.7678,
+      "step": 952
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.91809953473572e-06,
+      "loss": 3.9789,
+      "step": 953
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.917738540040074e-06,
+      "loss": 3.8024,
+      "step": 954
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.917376758109069e-06,
+      "loss": 3.7476,
+      "step": 955
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.917014189000622e-06,
+      "loss": 3.674,
+      "step": 956
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.916650832772767e-06,
+      "loss": 3.6222,
+      "step": 957
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.916286689483674e-06,
+      "loss": 3.7034,
+      "step": 958
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.915921759191634e-06,
+      "loss": 3.6922,
+      "step": 959
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.915556041955067e-06,
+      "loss": 3.724,
+      "step": 960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.915189537832516e-06,
+      "loss": 3.7525,
+      "step": 961
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.914822246882652e-06,
+      "loss": 3.8822,
+      "step": 962
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.914454169164269e-06,
+      "loss": 3.7524,
+      "step": 963
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.914085304736292e-06,
+      "loss": 3.84,
+      "step": 964
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.913715653657769e-06,
+      "loss": 3.6856,
+      "step": 965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.913345215987872e-06,
+      "loss": 3.7905,
+      "step": 966
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.912973991785903e-06,
+      "loss": 3.6484,
+      "step": 967
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.912601981111287e-06,
+      "loss": 3.8708,
+      "step": 968
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.912229184023573e-06,
+      "loss": 3.8142,
+      "step": 969
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.911855600582444e-06,
+      "loss": 3.7556,
+      "step": 970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.9114812308477e-06,
+      "loss": 3.8445,
+      "step": 971
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.911106074879272e-06,
+      "loss": 3.8096,
+      "step": 972
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.910730132737212e-06,
+      "loss": 3.7461,
+      "step": 973
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.910353404481706e-06,
+      "loss": 3.7005,
+      "step": 974
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.909975890173057e-06,
+      "loss": 3.7945,
+      "step": 975
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.9095975898717e-06,
+      "loss": 3.7747,
+      "step": 976
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.909218503638193e-06,
+      "loss": 3.7727,
+      "step": 977
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.90883863153322e-06,
+      "loss": 3.7846,
+      "step": 978
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.908457973617592e-06,
+      "loss": 3.8188,
+      "step": 979
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.908076529952244e-06,
+      "loss": 3.6517,
+      "step": 980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.907694300598237e-06,
+      "loss": 3.7422,
+      "step": 981
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.907311285616761e-06,
+      "loss": 3.7474,
+      "step": 982
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.906927485069129e-06,
+      "loss": 3.6379,
+      "step": 983
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.90654289901678e-06,
+      "loss": 3.7734,
+      "step": 984
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.906157527521277e-06,
+      "loss": 3.6955,
+      "step": 985
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.905771370644311e-06,
+      "loss": 3.7665,
+      "step": 986
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.9053844284477e-06,
+      "loss": 3.7995,
+      "step": 987
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.904996700993383e-06,
+      "loss": 3.7547,
+      "step": 988
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.904608188343433e-06,
+      "loss": 3.733,
+      "step": 989
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.904218890560038e-06,
+      "loss": 3.7364,
+      "step": 990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.903828807705517e-06,
+      "loss": 3.7707,
+      "step": 991
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.90343793984232e-06,
+      "loss": 3.7533,
+      "step": 992
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.903046287033015e-06,
+      "loss": 3.744,
+      "step": 993
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.902653849340296e-06,
+      "loss": 3.6338,
+      "step": 994
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.902260626826987e-06,
+      "loss": 3.7743,
+      "step": 995
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.901866619556033e-06,
+      "loss": 3.7613,
+      "step": 996
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.90147182759051e-06,
+      "loss": 3.6403,
+      "step": 997
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.901076250993615e-06,
+      "loss": 3.6376,
+      "step": 998
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.900679889828673e-06,
+      "loss": 3.7363,
+      "step": 999
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.90028274415913e-06,
+      "loss": 3.7236,
+      "step": 1000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.899884814048569e-06,
+      "loss": 3.8368,
+      "step": 1001
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.899486099560684e-06,
+      "loss": 3.6462,
+      "step": 1002
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.899086600759305e-06,
+      "loss": 3.7003,
+      "step": 1003
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.898686317708381e-06,
+      "loss": 3.8304,
+      "step": 1004
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.898285250471993e-06,
+      "loss": 3.6542,
+      "step": 1005
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.897883399114343e-06,
+      "loss": 3.7955,
+      "step": 1006
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.897480763699759e-06,
+      "loss": 3.7329,
+      "step": 1007
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.897077344292695e-06,
+      "loss": 3.8364,
+      "step": 1008
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.896673140957735e-06,
+      "loss": 3.7776,
+      "step": 1009
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.896268153759578e-06,
+      "loss": 3.729,
+      "step": 1010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.895862382763058e-06,
+      "loss": 3.7262,
+      "step": 1011
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.89545582803313e-06,
+      "loss": 3.6804,
+      "step": 1012
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.895048489634873e-06,
+      "loss": 3.7523,
+      "step": 1013
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.894640367633502e-06,
+      "loss": 3.696,
+      "step": 1014
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.894231462094341e-06,
+      "loss": 3.7073,
+      "step": 1015
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.893821773082852e-06,
+      "loss": 3.7869,
+      "step": 1016
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.89341130066462e-06,
+      "loss": 3.713,
+      "step": 1017
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.893000044905348e-06,
+      "loss": 3.6837,
+      "step": 1018
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.892588005870876e-06,
+      "loss": 3.6917,
+      "step": 1019
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.892175183627161e-06,
+      "loss": 3.7121,
+      "step": 1020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.891761578240288e-06,
+      "loss": 3.8913,
+      "step": 1021
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.891347189776466e-06,
+      "loss": 3.7696,
+      "step": 1022
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.890932018302034e-06,
+      "loss": 3.7067,
+      "step": 1023
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.890516063883449e-06,
+      "loss": 3.763,
+      "step": 1024
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.8900993265873e-06,
+      "loss": 3.6897,
+      "step": 1025
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.889681806480297e-06,
+      "loss": 3.7727,
+      "step": 1026
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.88926350362928e-06,
+      "loss": 3.7406,
+      "step": 1027
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.888844418101207e-06,
+      "loss": 3.7535,
+      "step": 1028
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.888424549963169e-06,
+      "loss": 3.8446,
+      "step": 1029
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.888003899282375e-06,
+      "loss": 3.7535,
+      "step": 1030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.887582466126167e-06,
+      "loss": 3.6411,
+      "step": 1031
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.887160250562007e-06,
+      "loss": 3.7396,
+      "step": 1032
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.886737252657483e-06,
+      "loss": 3.6899,
+      "step": 1033
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.886313472480309e-06,
+      "loss": 3.7003,
+      "step": 1034
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.885888910098325e-06,
+      "loss": 3.6126,
+      "step": 1035
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.885463565579494e-06,
+      "loss": 3.7539,
+      "step": 1036
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.885037438991905e-06,
+      "loss": 3.7748,
+      "step": 1037
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.884610530403776e-06,
+      "loss": 3.7029,
+      "step": 1038
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.884182839883442e-06,
+      "loss": 3.8887,
+      "step": 1039
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.883754367499372e-06,
+      "loss": 3.7958,
+      "step": 1040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.883325113320155e-06,
+      "loss": 3.8878,
+      "step": 1041
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.882895077414506e-06,
+      "loss": 3.7851,
+      "step": 1042
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.882464259851265e-06,
+      "loss": 3.7987,
+      "step": 1043
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.8820326606994e-06,
+      "loss": 3.6933,
+      "step": 1044
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.881600280027997e-06,
+      "loss": 3.7175,
+      "step": 1045
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.881167117906276e-06,
+      "loss": 3.7012,
+      "step": 1046
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.880733174403576e-06,
+      "loss": 3.744,
+      "step": 1047
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.880298449589365e-06,
+      "loss": 3.681,
+      "step": 1048
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.879862943533232e-06,
+      "loss": 3.7427,
+      "step": 1049
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.879426656304894e-06,
+      "loss": 3.7364,
+      "step": 1050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.878989587974191e-06,
+      "loss": 3.7093,
+      "step": 1051
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.878551738611092e-06,
+      "loss": 3.7367,
+      "step": 1052
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.878113108285685e-06,
+      "loss": 3.7314,
+      "step": 1053
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.87767369706819e-06,
+      "loss": 3.6713,
+      "step": 1054
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.877233505028944e-06,
+      "loss": 3.8605,
+      "step": 1055
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.876792532238417e-06,
+      "loss": 3.6206,
+      "step": 1056
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.876350778767198e-06,
+      "loss": 3.6738,
+      "step": 1057
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.875908244686004e-06,
+      "loss": 3.8736,
+      "step": 1058
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.875464930065677e-06,
+      "loss": 3.7654,
+      "step": 1059
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.87502083497718e-06,
+      "loss": 3.6915,
+      "step": 1060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.874575959491608e-06,
+      "loss": 3.8439,
+      "step": 1061
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.874130303680175e-06,
+      "loss": 3.7554,
+      "step": 1062
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.873683867614222e-06,
+      "loss": 3.8128,
+      "step": 1063
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.873236651365215e-06,
+      "loss": 3.7068,
+      "step": 1064
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.872788655004746e-06,
+      "loss": 3.714,
+      "step": 1065
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.872339878604529e-06,
+      "loss": 3.658,
+      "step": 1066
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.871890322236404e-06,
+      "loss": 3.6466,
+      "step": 1067
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.871439985972338e-06,
+      "loss": 3.924,
+      "step": 1068
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.870988869884419e-06,
+      "loss": 3.7269,
+      "step": 1069
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.870536974044863e-06,
+      "loss": 3.7896,
+      "step": 1070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.870084298526012e-06,
+      "loss": 3.8404,
+      "step": 1071
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.869630843400331e-06,
+      "loss": 3.6866,
+      "step": 1072
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.869176608740404e-06,
+      "loss": 3.6917,
+      "step": 1073
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.868721594618948e-06,
+      "loss": 3.7082,
+      "step": 1074
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.868265801108806e-06,
+      "loss": 3.6903,
+      "step": 1075
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.867809228282937e-06,
+      "loss": 3.7997,
+      "step": 1076
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.867351876214429e-06,
+      "loss": 3.7888,
+      "step": 1077
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.866893744976498e-06,
+      "loss": 3.6602,
+      "step": 1078
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.866434834642482e-06,
+      "loss": 3.6954,
+      "step": 1079
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.865975145285842e-06,
+      "loss": 3.8268,
+      "step": 1080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.865514676980167e-06,
+      "loss": 3.7485,
+      "step": 1081
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.865053429799167e-06,
+      "loss": 3.7449,
+      "step": 1082
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.864591403816682e-06,
+      "loss": 3.7376,
+      "step": 1083
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.864128599106672e-06,
+      "loss": 3.6937,
+      "step": 1084
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.863665015743221e-06,
+      "loss": 3.8024,
+      "step": 1085
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.863200653800541e-06,
+      "loss": 3.7683,
+      "step": 1086
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.862735513352971e-06,
+      "loss": 3.6985,
+      "step": 1087
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.862269594474967e-06,
+      "loss": 3.7467,
+      "step": 1088
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.861802897241113e-06,
+      "loss": 3.6895,
+      "step": 1089
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.861335421726121e-06,
+      "loss": 3.6723,
+      "step": 1090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.860867168004823e-06,
+      "loss": 3.8022,
+      "step": 1091
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.86039813615218e-06,
+      "loss": 3.8243,
+      "step": 1092
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.859928326243273e-06,
+      "loss": 3.6808,
+      "step": 1093
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.85945773835331e-06,
+      "loss": 3.7585,
+      "step": 1094
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.858986372557621e-06,
+      "loss": 3.8255,
+      "step": 1095
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.858514228931664e-06,
+      "loss": 3.7236,
+      "step": 1096
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.858041307551022e-06,
+      "loss": 3.7568,
+      "step": 1097
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.8575676084914e-06,
+      "loss": 3.7106,
+      "step": 1098
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.857093131828625e-06,
+      "loss": 3.722,
+      "step": 1099
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.856617877638656e-06,
+      "loss": 3.7766,
+      "step": 1100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.856141845997569e-06,
+      "loss": 3.6122,
+      "step": 1101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.855665036981567e-06,
+      "loss": 3.7737,
+      "step": 1102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.855187450666981e-06,
+      "loss": 3.6897,
+      "step": 1103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.854709087130261e-06,
+      "loss": 3.6946,
+      "step": 1104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.854229946447983e-06,
+      "loss": 3.7795,
+      "step": 1105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.853750028696853e-06,
+      "loss": 3.7092,
+      "step": 1106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.853269333953693e-06,
+      "loss": 3.8062,
+      "step": 1107
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.85278786229545e-06,
+      "loss": 3.7503,
+      "step": 1108
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.852305613799204e-06,
+      "loss": 3.7587,
+      "step": 1109
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.851822588542152e-06,
+      "loss": 3.7014,
+      "step": 1110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.851338786601616e-06,
+      "loss": 3.7862,
+      "step": 1111
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.850854208055043e-06,
+      "loss": 3.7737,
+      "step": 1112
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.850368852980005e-06,
+      "loss": 3.8673,
+      "step": 1113
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.8498827214542e-06,
+      "loss": 3.6881,
+      "step": 1114
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.849395813555446e-06,
+      "loss": 3.7958,
+      "step": 1115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.848908129361689e-06,
+      "loss": 3.8067,
+      "step": 1116
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.848419668950996e-06,
+      "loss": 3.7766,
+      "step": 1117
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.847930432401563e-06,
+      "loss": 3.7911,
+      "step": 1118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.847440419791705e-06,
+      "loss": 3.5337,
+      "step": 1119
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.846949631199865e-06,
+      "loss": 3.6838,
+      "step": 1120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.846458066704607e-06,
+      "loss": 3.6371,
+      "step": 1121
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.845965726384623e-06,
+      "loss": 3.8433,
+      "step": 1122
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.845472610318727e-06,
+      "loss": 3.6758,
+      "step": 1123
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.844978718585855e-06,
+      "loss": 3.7567,
+      "step": 1124
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.844484051265073e-06,
+      "loss": 3.8342,
+      "step": 1125
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.843988608435567e-06,
+      "loss": 3.7918,
+      "step": 1126
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.843492390176645e-06,
+      "loss": 3.6752,
+      "step": 1127
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.842995396567744e-06,
+      "loss": 3.7299,
+      "step": 1128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.842497627688424e-06,
+      "loss": 3.8505,
+      "step": 1129
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.841999083618368e-06,
+      "loss": 3.6891,
+      "step": 1130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.841499764437382e-06,
+      "loss": 3.777,
+      "step": 1131
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.840999670225396e-06,
+      "loss": 3.7287,
+      "step": 1132
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.84049880106247e-06,
+      "loss": 3.7452,
+      "step": 1133
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.83999715702878e-06,
+      "loss": 3.6332,
+      "step": 1134
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.839494738204631e-06,
+      "loss": 3.7054,
+      "step": 1135
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.83899154467045e-06,
+      "loss": 3.713,
+      "step": 1136
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.838487576506787e-06,
+      "loss": 3.7594,
+      "step": 1137
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.837982833794322e-06,
+      "loss": 3.6983,
+      "step": 1138
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.837477316613852e-06,
+      "loss": 3.7444,
+      "step": 1139
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.8369710250463e-06,
+      "loss": 3.5749,
+      "step": 1140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.836463959172713e-06,
+      "loss": 3.7605,
+      "step": 1141
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.835956119074265e-06,
+      "loss": 3.7913,
+      "step": 1142
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.835447504832249e-06,
+      "loss": 3.781,
+      "step": 1143
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.834938116528086e-06,
+      "loss": 3.662,
+      "step": 1144
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.83442795424332e-06,
+      "loss": 3.766,
+      "step": 1145
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.833917018059614e-06,
+      "loss": 3.7511,
+      "step": 1146
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.833405308058763e-06,
+      "loss": 3.7205,
+      "step": 1147
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.832892824322681e-06,
+      "loss": 3.8372,
+      "step": 1148
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.832379566933407e-06,
+      "loss": 3.8206,
+      "step": 1149
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.831865535973103e-06,
+      "loss": 3.6257,
+      "step": 1150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.831350731524056e-06,
+      "loss": 3.7235,
+      "step": 1151
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.830835153668675e-06,
+      "loss": 3.7934,
+      "step": 1152
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.830318802489496e-06,
+      "loss": 3.7811,
+      "step": 1153
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.829801678069174e-06,
+      "loss": 3.704,
+      "step": 1154
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.829283780490496e-06,
+      "loss": 3.6958,
+      "step": 1155
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.828765109836362e-06,
+      "loss": 3.7175,
+      "step": 1156
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.828245666189805e-06,
+      "loss": 3.7423,
+      "step": 1157
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.827725449633976e-06,
+      "loss": 3.7485,
+      "step": 1158
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.827204460252151e-06,
+      "loss": 3.7292,
+      "step": 1159
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.826682698127731e-06,
+      "loss": 3.7631,
+      "step": 1160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.82616016334424e-06,
+      "loss": 3.8412,
+      "step": 1161
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.825636855985328e-06,
+      "loss": 3.7885,
+      "step": 1162
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.825112776134763e-06,
+      "loss": 3.7901,
+      "step": 1163
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.824587923876442e-06,
+      "loss": 3.7722,
+      "step": 1164
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.824062299294383e-06,
+      "loss": 3.8109,
+      "step": 1165
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.82353590247273e-06,
+      "loss": 3.8629,
+      "step": 1166
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.823008733495747e-06,
+      "loss": 3.6409,
+      "step": 1167
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.822480792447826e-06,
+      "loss": 3.7113,
+      "step": 1168
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.821952079413476e-06,
+      "loss": 3.6828,
+      "step": 1169
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.821422594477339e-06,
+      "loss": 3.7622,
+      "step": 1170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.820892337724172e-06,
+      "loss": 3.7348,
+      "step": 1171
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.82036130923886e-06,
+      "loss": 3.791,
+      "step": 1172
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.819829509106408e-06,
+      "loss": 3.6859,
+      "step": 1173
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.819296937411952e-06,
+      "loss": 3.6287,
+      "step": 1174
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.818763594240742e-06,
+      "loss": 3.7468,
+      "step": 1175
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.81822947967816e-06,
+      "loss": 3.7245,
+      "step": 1176
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.817694593809701e-06,
+      "loss": 3.7977,
+      "step": 1177
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.817158936720997e-06,
+      "loss": 3.7335,
+      "step": 1178
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.816622508497792e-06,
+      "loss": 3.6807,
+      "step": 1179
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.81608530922596e-06,
+      "loss": 3.7322,
+      "step": 1180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.815547338991496e-06,
+      "loss": 3.748,
+      "step": 1181
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.815008597880517e-06,
+      "loss": 3.7912,
+      "step": 1182
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.814469085979266e-06,
+      "loss": 3.7751,
+      "step": 1183
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.81392880337411e-06,
+      "loss": 3.5173,
+      "step": 1184
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.813387750151536e-06,
+      "loss": 3.6448,
+      "step": 1185
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.812845926398157e-06,
+      "loss": 3.7407,
+      "step": 1186
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.812303332200709e-06,
+      "loss": 3.6606,
+      "step": 1187
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.81175996764605e-06,
+      "loss": 3.7937,
+      "step": 1188
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.811215832821162e-06,
+      "loss": 3.6235,
+      "step": 1189
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.810670927813152e-06,
+      "loss": 3.7753,
+      "step": 1190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.810125252709247e-06,
+      "loss": 3.7013,
+      "step": 1191
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.8095788075968e-06,
+      "loss": 3.8353,
+      "step": 1192
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.809031592563288e-06,
+      "loss": 3.6643,
+      "step": 1193
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.808483607696305e-06,
+      "loss": 3.7363,
+      "step": 1194
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.807934853083579e-06,
+      "loss": 3.7601,
+      "step": 1195
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.807385328812948e-06,
+      "loss": 3.7949,
+      "step": 1196
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.806835034972386e-06,
+      "loss": 3.7553,
+      "step": 1197
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.806283971649984e-06,
+      "loss": 3.856,
+      "step": 1198
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.805732138933953e-06,
+      "loss": 3.8239,
+      "step": 1199
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.805179536912635e-06,
+      "loss": 3.6929,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.804626165674486e-06,
+      "loss": 3.7101,
+      "step": 1201
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.804072025308096e-06,
+      "loss": 3.9157,
+      "step": 1202
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.803517115902168e-06,
+      "loss": 3.7147,
+      "step": 1203
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.802961437545532e-06,
+      "loss": 3.7905,
+      "step": 1204
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.802404990327146e-06,
+      "loss": 3.7241,
+      "step": 1205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.801847774336082e-06,
+      "loss": 3.7344,
+      "step": 1206
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.801289789661541e-06,
+      "loss": 3.8376,
+      "step": 1207
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.800731036392846e-06,
+      "loss": 3.7064,
+      "step": 1208
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.800171514619442e-06,
+      "loss": 3.7541,
+      "step": 1209
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.799611224430897e-06,
+      "loss": 3.7483,
+      "step": 1210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.799050165916906e-06,
+      "loss": 3.6982,
+      "step": 1211
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.79848833916728e-06,
+      "loss": 3.7908,
+      "step": 1212
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.797925744271957e-06,
+      "loss": 3.6975,
+      "step": 1213
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.797362381321e-06,
+      "loss": 3.8141,
+      "step": 1214
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.796798250404591e-06,
+      "loss": 3.7277,
+      "step": 1215
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.796233351613036e-06,
+      "loss": 3.6538,
+      "step": 1216
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.795667685036764e-06,
+      "loss": 3.8003,
+      "step": 1217
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.795101250766332e-06,
+      "loss": 3.7727,
+      "step": 1218
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.794534048892411e-06,
+      "loss": 3.597,
+      "step": 1219
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.793966079505797e-06,
+      "loss": 3.6649,
+      "step": 1220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.793397342697416e-06,
+      "loss": 3.7803,
+      "step": 1221
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.792827838558309e-06,
+      "loss": 3.6963,
+      "step": 1222
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.792257567179643e-06,
+      "loss": 3.7542,
+      "step": 1223
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.791686528652708e-06,
+      "loss": 3.7524,
+      "step": 1224
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.791114723068917e-06,
+      "loss": 3.6435,
+      "step": 1225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.790542150519803e-06,
+      "loss": 3.7896,
+      "step": 1226
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.789968811097026e-06,
+      "loss": 3.689,
+      "step": 1227
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.789394704892364e-06,
+      "loss": 3.885,
+      "step": 1228
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.788819831997723e-06,
+      "loss": 3.7268,
+      "step": 1229
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.788244192505128e-06,
+      "loss": 3.7951,
+      "step": 1230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.787667786506728e-06,
+      "loss": 3.6553,
+      "step": 1231
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.787090614094795e-06,
+      "loss": 3.7302,
+      "step": 1232
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.786512675361722e-06,
+      "loss": 3.6612,
+      "step": 1233
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.785933970400026e-06,
+      "loss": 3.6822,
+      "step": 1234
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.785354499302346e-06,
+      "loss": 3.8469,
+      "step": 1235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.784774262161448e-06,
+      "loss": 3.6907,
+      "step": 1236
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.784193259070212e-06,
+      "loss": 3.6472,
+      "step": 1237
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.783611490121646e-06,
+      "loss": 3.7807,
+      "step": 1238
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.783028955408884e-06,
+      "loss": 3.7491,
+      "step": 1239
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.782445655025175e-06,
+      "loss": 3.777,
+      "step": 1240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.781861589063895e-06,
+      "loss": 3.7968,
+      "step": 1241
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.781276757618543e-06,
+      "loss": 3.6958,
+      "step": 1242
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.780691160782737e-06,
+      "loss": 3.7969,
+      "step": 1243
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.780104798650221e-06,
+      "loss": 3.6403,
+      "step": 1244
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.779517671314863e-06,
+      "loss": 3.7722,
+      "step": 1245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.778929778870646e-06,
+      "loss": 3.8111,
+      "step": 1246
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.778341121411681e-06,
+      "loss": 3.6883,
+      "step": 1247
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.777751699032203e-06,
+      "loss": 3.7051,
+      "step": 1248
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.777161511826567e-06,
+      "loss": 3.8817,
+      "step": 1249
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.77657055988925e-06,
+      "loss": 3.824,
+      "step": 1250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.77597884331485e-06,
+      "loss": 3.7396,
+      "step": 1251
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.775386362198093e-06,
+      "loss": 3.7839,
+      "step": 1252
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.77479311663382e-06,
+      "loss": 3.67,
+      "step": 1253
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.774199106717004e-06,
+      "loss": 3.7758,
+      "step": 1254
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.77360433254273e-06,
+      "loss": 3.6503,
+      "step": 1255
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.773008794206209e-06,
+      "loss": 3.7167,
+      "step": 1256
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.772412491802778e-06,
+      "loss": 3.7578,
+      "step": 1257
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.771815425427893e-06,
+      "loss": 3.6301,
+      "step": 1258
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.771217595177133e-06,
+      "loss": 3.7514,
+      "step": 1259
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.770619001146198e-06,
+      "loss": 3.8135,
+      "step": 1260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.770019643430913e-06,
+      "loss": 3.7755,
+      "step": 1261
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.769419522127221e-06,
+      "loss": 3.8474,
+      "step": 1262
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.768818637331193e-06,
+      "loss": 3.7036,
+      "step": 1263
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.768216989139017e-06,
+      "loss": 3.6177,
+      "step": 1264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.767614577647005e-06,
+      "loss": 3.7257,
+      "step": 1265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.767011402951594e-06,
+      "loss": 3.74,
+      "step": 1266
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.766407465149337e-06,
+      "loss": 3.7177,
+      "step": 1267
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.765802764336917e-06,
+      "loss": 3.8101,
+      "step": 1268
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.765197300611132e-06,
+      "loss": 3.7661,
+      "step": 1269
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.764591074068906e-06,
+      "loss": 3.7377,
+      "step": 1270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.763984084807285e-06,
+      "loss": 3.773,
+      "step": 1271
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.763376332923435e-06,
+      "loss": 3.6617,
+      "step": 1272
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.762767818514646e-06,
+      "loss": 3.7083,
+      "step": 1273
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.76215854167833e-06,
+      "loss": 3.7652,
+      "step": 1274
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.761548502512017e-06,
+      "loss": 3.7851,
+      "step": 1275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.760937701113368e-06,
+      "loss": 3.8012,
+      "step": 1276
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.760326137580158e-06,
+      "loss": 3.7721,
+      "step": 1277
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.759713812010286e-06,
+      "loss": 3.5796,
+      "step": 1278
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.759100724501774e-06,
+      "loss": 3.8198,
+      "step": 1279
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.758486875152766e-06,
+      "loss": 3.8425,
+      "step": 1280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.757872264061527e-06,
+      "loss": 3.6878,
+      "step": 1281
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.757256891326445e-06,
+      "loss": 3.8372,
+      "step": 1282
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.756640757046027e-06,
+      "loss": 3.686,
+      "step": 1283
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.756023861318907e-06,
+      "loss": 3.7048,
+      "step": 1284
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.755406204243838e-06,
+      "loss": 3.8412,
+      "step": 1285
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.754787785919693e-06,
+      "loss": 3.6152,
+      "step": 1286
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.754168606445471e-06,
+      "loss": 3.6524,
+      "step": 1287
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.75354866592029e-06,
+      "loss": 3.6692,
+      "step": 1288
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.75292796444339e-06,
+      "loss": 3.6518,
+      "step": 1289
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.752306502114132e-06,
+      "loss": 3.8174,
+      "step": 1290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.751684279032004e-06,
+      "loss": 3.764,
+      "step": 1291
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.75106129529661e-06,
+      "loss": 3.687,
+      "step": 1292
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.750437551007677e-06,
+      "loss": 3.6516,
+      "step": 1293
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.749813046265055e-06,
+      "loss": 3.8743,
+      "step": 1294
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.749187781168715e-06,
+      "loss": 3.7185,
+      "step": 1295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.748561755818751e-06,
+      "loss": 3.7566,
+      "step": 1296
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.747934970315376e-06,
+      "loss": 3.6914,
+      "step": 1297
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.747307424758927e-06,
+      "loss": 3.639,
+      "step": 1298
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.746679119249862e-06,
+      "loss": 3.8508,
+      "step": 1299
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.746050053888761e-06,
+      "loss": 3.7435,
+      "step": 1300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.745420228776326e-06,
+      "loss": 3.6739,
+      "step": 1301
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.744789644013377e-06,
+      "loss": 3.6921,
+      "step": 1302
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.74415829970086e-06,
+      "loss": 3.7446,
+      "step": 1303
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.743526195939844e-06,
+      "loss": 3.7084,
+      "step": 1304
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.742893332831514e-06,
+      "loss": 3.6619,
+      "step": 1305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.742259710477178e-06,
+      "loss": 3.7778,
+      "step": 1306
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.741625328978268e-06,
+      "loss": 3.7568,
+      "step": 1307
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.740990188436336e-06,
+      "loss": 3.6651,
+      "step": 1308
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.740354288953055e-06,
+      "loss": 3.7875,
+      "step": 1309
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.739717630630223e-06,
+      "loss": 3.7709,
+      "step": 1310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.739080213569756e-06,
+      "loss": 3.7657,
+      "step": 1311
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.73844203787369e-06,
+      "loss": 3.7167,
+      "step": 1312
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.737803103644185e-06,
+      "loss": 3.6245,
+      "step": 1313
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.737163410983526e-06,
+      "loss": 3.7694,
+      "step": 1314
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.73652295999411e-06,
+      "loss": 3.7101,
+      "step": 1315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.735881750778463e-06,
+      "loss": 3.8499,
+      "step": 1316
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.735239783439234e-06,
+      "loss": 3.7635,
+      "step": 1317
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.734597058079184e-06,
+      "loss": 3.8191,
+      "step": 1318
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.733953574801203e-06,
+      "loss": 3.6704,
+      "step": 1319
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.733309333708303e-06,
+      "loss": 3.7831,
+      "step": 1320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.73266433490361e-06,
+      "loss": 3.7716,
+      "step": 1321
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.732018578490379e-06,
+      "loss": 3.6752,
+      "step": 1322
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.731372064571983e-06,
+      "loss": 3.8451,
+      "step": 1323
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.730724793251914e-06,
+      "loss": 3.6987,
+      "step": 1324
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.730076764633788e-06,
+      "loss": 3.747,
+      "step": 1325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.729427978821346e-06,
+      "loss": 3.8127,
+      "step": 1326
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.728778435918444e-06,
+      "loss": 3.6474,
+      "step": 1327
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.72812813602906e-06,
+      "loss": 3.7913,
+      "step": 1328
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.727477079257295e-06,
+      "loss": 3.6697,
+      "step": 1329
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.726825265707371e-06,
+      "loss": 3.5998,
+      "step": 1330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.72617269548363e-06,
+      "loss": 3.741,
+      "step": 1331
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.725519368690539e-06,
+      "loss": 3.8358,
+      "step": 1332
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.724865285432678e-06,
+      "loss": 3.7942,
+      "step": 1333
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.72421044581476e-06,
+      "loss": 3.8034,
+      "step": 1334
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.723554849941605e-06,
+      "loss": 3.6752,
+      "step": 1335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.722898497918165e-06,
+      "loss": 3.7504,
+      "step": 1336
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.722241389849508e-06,
+      "loss": 3.7316,
+      "step": 1337
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.721583525840827e-06,
+      "loss": 3.7567,
+      "step": 1338
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.72092490599743e-06,
+      "loss": 3.7276,
+      "step": 1339
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.720265530424752e-06,
+      "loss": 3.7936,
+      "step": 1340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.719605399228344e-06,
+      "loss": 3.7718,
+      "step": 1341
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.718944512513883e-06,
+      "loss": 3.6587,
+      "step": 1342
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.71828287038716e-06,
+      "loss": 3.7223,
+      "step": 1343
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.717620472954098e-06,
+      "loss": 3.769,
+      "step": 1344
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.716957320320728e-06,
+      "loss": 3.7485,
+      "step": 1345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.716293412593211e-06,
+      "loss": 3.8574,
+      "step": 1346
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.715628749877823e-06,
+      "loss": 3.8123,
+      "step": 1347
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.714963332280967e-06,
+      "loss": 3.7038,
+      "step": 1348
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.714297159909163e-06,
+      "loss": 3.7108,
+      "step": 1349
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.71363023286905e-06,
+      "loss": 3.782,
+      "step": 1350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.712962551267393e-06,
+      "loss": 3.6717,
+      "step": 1351
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.712294115211074e-06,
+      "loss": 3.6399,
+      "step": 1352
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.711624924807095e-06,
+      "loss": 3.7288,
+      "step": 1353
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.710954980162584e-06,
+      "loss": 3.7785,
+      "step": 1354
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.710284281384783e-06,
+      "loss": 3.7301,
+      "step": 1355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.709612828581058e-06,
+      "loss": 3.6453,
+      "step": 1356
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.708940621858898e-06,
+      "loss": 3.7093,
+      "step": 1357
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.70826766132591e-06,
+      "loss": 3.8248,
+      "step": 1358
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.707593947089821e-06,
+      "loss": 3.8534,
+      "step": 1359
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.70691947925848e-06,
+      "loss": 3.8321,
+      "step": 1360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.706244257939856e-06,
+      "loss": 3.7174,
+      "step": 1361
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.70556828324204e-06,
+      "loss": 3.7796,
+      "step": 1362
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.704891555273241e-06,
+      "loss": 3.7415,
+      "step": 1363
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.704214074141793e-06,
+      "loss": 3.8143,
+      "step": 1364
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.703535839956145e-06,
+      "loss": 3.6984,
+      "step": 1365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.702856852824868e-06,
+      "loss": 3.8652,
+      "step": 1366
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.702177112856658e-06,
+      "loss": 3.71,
+      "step": 1367
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.701496620160327e-06,
+      "loss": 3.7929,
+      "step": 1368
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.700815374844808e-06,
+      "loss": 3.7427,
+      "step": 1369
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.700133377019158e-06,
+      "loss": 3.8849,
+      "step": 1370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.699450626792549e-06,
+      "loss": 3.7638,
+      "step": 1371
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.698767124274276e-06,
+      "loss": 3.6362,
+      "step": 1372
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.698082869573758e-06,
+      "loss": 3.7658,
+      "step": 1373
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.697397862800526e-06,
+      "loss": 3.7744,
+      "step": 1374
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.69671210406424e-06,
+      "loss": 3.5646,
+      "step": 1375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.696025593474676e-06,
+      "loss": 3.8221,
+      "step": 1376
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.695338331141731e-06,
+      "loss": 3.7146,
+      "step": 1377
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.694650317175421e-06,
+      "loss": 3.551,
+      "step": 1378
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.693961551685886e-06,
+      "loss": 3.7977,
+      "step": 1379
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.693272034783385e-06,
+      "loss": 3.6171,
+      "step": 1380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.692581766578293e-06,
+      "loss": 3.8481,
+      "step": 1381
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.691890747181112e-06,
+      "loss": 3.8458,
+      "step": 1382
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.691198976702459e-06,
+      "loss": 3.7432,
+      "step": 1383
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.690506455253073e-06,
+      "loss": 3.6571,
+      "step": 1384
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.689813182943813e-06,
+      "loss": 3.7839,
+      "step": 1385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.68911915988566e-06,
+      "loss": 3.7228,
+      "step": 1386
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.688424386189713e-06,
+      "loss": 3.7451,
+      "step": 1387
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.687728861967192e-06,
+      "loss": 3.6572,
+      "step": 1388
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.687032587329439e-06,
+      "loss": 3.6848,
+      "step": 1389
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.68633556238791e-06,
+      "loss": 3.7882,
+      "step": 1390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.685637787254187e-06,
+      "loss": 3.7207,
+      "step": 1391
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.684939262039974e-06,
+      "loss": 3.8191,
+      "step": 1392
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.684239986857085e-06,
+      "loss": 3.7208,
+      "step": 1393
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.683539961817468e-06,
+      "loss": 3.7827,
+      "step": 1394
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.682839187033176e-06,
+      "loss": 3.7526,
+      "step": 1395
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.682137662616393e-06,
+      "loss": 3.773,
+      "step": 1396
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.68143538867942e-06,
+      "loss": 3.7228,
+      "step": 1397
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.68073236533468e-06,
+      "loss": 3.8429,
+      "step": 1398
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.680028592694708e-06,
+      "loss": 3.7494,
+      "step": 1399
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.679324070872168e-06,
+      "loss": 3.6211,
+      "step": 1400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.67861879997984e-06,
+      "loss": 3.6693,
+      "step": 1401
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.677912780130623e-06,
+      "loss": 3.7348,
+      "step": 1402
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.677206011437536e-06,
+      "loss": 3.7846,
+      "step": 1403
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.676498494013724e-06,
+      "loss": 3.7822,
+      "step": 1404
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.675790227972444e-06,
+      "loss": 3.7637,
+      "step": 1405
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.675081213427076e-06,
+      "loss": 3.7263,
+      "step": 1406
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.674371450491118e-06,
+      "loss": 3.8042,
+      "step": 1407
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.67366093927819e-06,
+      "loss": 3.7158,
+      "step": 1408
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.672949679902035e-06,
+      "loss": 3.6733,
+      "step": 1409
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.672237672476506e-06,
+      "loss": 3.7646,
+      "step": 1410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.671524917115585e-06,
+      "loss": 3.7216,
+      "step": 1411
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.670811413933372e-06,
+      "loss": 3.7157,
+      "step": 1412
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.67009716304408e-06,
+      "loss": 3.8071,
+      "step": 1413
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.669382164562054e-06,
+      "loss": 3.7275,
+      "step": 1414
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.668666418601746e-06,
+      "loss": 3.683,
+      "step": 1415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.667949925277733e-06,
+      "loss": 3.7298,
+      "step": 1416
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.667232684704714e-06,
+      "loss": 3.7509,
+      "step": 1417
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.666514696997506e-06,
+      "loss": 3.7309,
+      "step": 1418
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.665795962271044e-06,
+      "loss": 3.6549,
+      "step": 1419
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.665076480640383e-06,
+      "loss": 3.8278,
+      "step": 1420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.664356252220699e-06,
+      "loss": 3.7513,
+      "step": 1421
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.663635277127285e-06,
+      "loss": 3.7599,
+      "step": 1422
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.66291355547556e-06,
+      "loss": 3.7344,
+      "step": 1423
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.662191087381051e-06,
+      "loss": 3.8295,
+      "step": 1424
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.661467872959417e-06,
+      "loss": 3.701,
+      "step": 1425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.660743912326428e-06,
+      "loss": 3.7022,
+      "step": 1426
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.660019205597974e-06,
+      "loss": 3.6515,
+      "step": 1427
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.659293752890072e-06,
+      "loss": 3.7338,
+      "step": 1428
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.65856755431885e-06,
+      "loss": 3.7646,
+      "step": 1429
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.657840610000557e-06,
+      "loss": 3.7752,
+      "step": 1430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.657112920051565e-06,
+      "loss": 3.6252,
+      "step": 1431
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.656384484588363e-06,
+      "loss": 3.7869,
+      "step": 1432
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.65565530372756e-06,
+      "loss": 3.7332,
+      "step": 1433
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.654925377585881e-06,
+      "loss": 3.8005,
+      "step": 1434
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.654194706280176e-06,
+      "loss": 3.7358,
+      "step": 1435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.65346328992741e-06,
+      "loss": 3.6107,
+      "step": 1436
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.65273112864467e-06,
+      "loss": 3.7596,
+      "step": 1437
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.65199822254916e-06,
+      "loss": 3.685,
+      "step": 1438
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.651264571758203e-06,
+      "loss": 3.6881,
+      "step": 1439
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.650530176389245e-06,
+      "loss": 3.786,
+      "step": 1440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.649795036559848e-06,
+      "loss": 3.771,
+      "step": 1441
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.649059152387692e-06,
+      "loss": 3.7835,
+      "step": 1442
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.648322523990579e-06,
+      "loss": 3.7184,
+      "step": 1443
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.64758515148643e-06,
+      "loss": 3.722,
+      "step": 1444
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.646847034993281e-06,
+      "loss": 3.6813,
+      "step": 1445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.646108174629294e-06,
+      "loss": 3.6439,
+      "step": 1446
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.645368570512743e-06,
+      "loss": 3.7044,
+      "step": 1447
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.644628222762027e-06,
+      "loss": 3.7525,
+      "step": 1448
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.643887131495662e-06,
+      "loss": 3.7546,
+      "step": 1449
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.643145296832279e-06,
+      "loss": 3.6286,
+      "step": 1450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.642402718890633e-06,
+      "loss": 3.6178,
+      "step": 1451
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.641659397789599e-06,
+      "loss": 3.869,
+      "step": 1452
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.640915333648164e-06,
+      "loss": 3.7843,
+      "step": 1453
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.640170526585443e-06,
+      "loss": 3.6238,
+      "step": 1454
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.639424976720661e-06,
+      "loss": 3.7019,
+      "step": 1455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.63867868417317e-06,
+      "loss": 3.7847,
+      "step": 1456
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.637931649062434e-06,
+      "loss": 3.8104,
+      "step": 1457
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.63718387150804e-06,
+      "loss": 3.7932,
+      "step": 1458
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.636435351629694e-06,
+      "loss": 3.6793,
+      "step": 1459
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.635686089547219e-06,
+      "loss": 3.7649,
+      "step": 1460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.634936085380555e-06,
+      "loss": 3.8376,
+      "step": 1461
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.634185339249766e-06,
+      "loss": 3.8554,
+      "step": 1462
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.63343385127503e-06,
+      "loss": 3.7305,
+      "step": 1463
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.632681621576651e-06,
+      "loss": 3.6892,
+      "step": 1464
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.63192865027504e-06,
+      "loss": 3.7374,
+      "step": 1465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.631174937490736e-06,
+      "loss": 3.5448,
+      "step": 1466
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.630420483344394e-06,
+      "loss": 3.7672,
+      "step": 1467
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.629665287956788e-06,
+      "loss": 3.7525,
+      "step": 1468
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.628909351448808e-06,
+      "loss": 3.6539,
+      "step": 1469
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.628152673941467e-06,
+      "loss": 3.7902,
+      "step": 1470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.627395255555896e-06,
+      "loss": 3.7799,
+      "step": 1471
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.626637096413339e-06,
+      "loss": 3.7509,
+      "step": 1472
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.625878196635165e-06,
+      "loss": 3.7144,
+      "step": 1473
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.62511855634286e-06,
+      "loss": 3.6131,
+      "step": 1474
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.624358175658027e-06,
+      "loss": 3.6784,
+      "step": 1475
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.623597054702388e-06,
+      "loss": 3.7553,
+      "step": 1476
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.622835193597782e-06,
+      "loss": 3.6574,
+      "step": 1477
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.622072592466172e-06,
+      "loss": 3.7974,
+      "step": 1478
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.621309251429635e-06,
+      "loss": 3.8368,
+      "step": 1479
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.620545170610364e-06,
+      "loss": 3.7829,
+      "step": 1480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.619780350130678e-06,
+      "loss": 3.7651,
+      "step": 1481
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.619014790113008e-06,
+      "loss": 3.6341,
+      "step": 1482
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.618248490679905e-06,
+      "loss": 3.7688,
+      "step": 1483
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.61748145195404e-06,
+      "loss": 3.7879,
+      "step": 1484
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.616713674058199e-06,
+      "loss": 3.7036,
+      "step": 1485
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.615945157115291e-06,
+      "loss": 3.7377,
+      "step": 1486
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.615175901248339e-06,
+      "loss": 3.6505,
+      "step": 1487
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.614405906580486e-06,
+      "loss": 3.84,
+      "step": 1488
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.613635173234997e-06,
+      "loss": 3.6821,
+      "step": 1489
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.612863701335245e-06,
+      "loss": 3.8578,
+      "step": 1490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.612091491004733e-06,
+      "loss": 3.6883,
+      "step": 1491
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.611318542367076e-06,
+      "loss": 3.7829,
+      "step": 1492
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.610544855546007e-06,
+      "loss": 3.7336,
+      "step": 1493
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.609770430665381e-06,
+      "loss": 3.7601,
+      "step": 1494
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.608995267849165e-06,
+      "loss": 3.7233,
+      "step": 1495
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.60821936722145e-06,
+      "loss": 3.6477,
+      "step": 1496
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.607442728906441e-06,
+      "loss": 3.783,
+      "step": 1497
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.606665353028464e-06,
+      "loss": 3.7049,
+      "step": 1498
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.605887239711962e-06,
+      "loss": 3.6973,
+      "step": 1499
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.605108389081497e-06,
+      "loss": 3.7757,
+      "step": 1500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.604328801261747e-06,
+      "loss": 3.7316,
+      "step": 1501
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.603548476377506e-06,
+      "loss": 3.7589,
+      "step": 1502
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.602767414553695e-06,
+      "loss": 3.7559,
+      "step": 1503
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.601985615915342e-06,
+      "loss": 3.7291,
+      "step": 1504
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.6012030805876e-06,
+      "loss": 3.868,
+      "step": 1505
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.600419808695737e-06,
+      "loss": 3.7447,
+      "step": 1506
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.599635800365141e-06,
+      "loss": 3.7313,
+      "step": 1507
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.598851055721317e-06,
+      "loss": 3.7226,
+      "step": 1508
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.598065574889886e-06,
+      "loss": 3.6454,
+      "step": 1509
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.59727935799659e-06,
+      "loss": 3.6704,
+      "step": 1510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.596492405167284e-06,
+      "loss": 3.6596,
+      "step": 1511
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.595704716527946e-06,
+      "loss": 3.6892,
+      "step": 1512
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.594916292204672e-06,
+      "loss": 3.8372,
+      "step": 1513
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.594127132323669e-06,
+      "loss": 3.7406,
+      "step": 1514
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.59333723701127e-06,
+      "loss": 3.7564,
+      "step": 1515
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.592546606393922e-06,
+      "loss": 3.7846,
+      "step": 1516
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.591755240598188e-06,
+      "loss": 3.7826,
+      "step": 1517
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.59096313975075e-06,
+      "loss": 3.7472,
+      "step": 1518
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.590170303978412e-06,
+      "loss": 3.6229,
+      "step": 1519
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.589376733408086e-06,
+      "loss": 3.7215,
+      "step": 1520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.58858242816681e-06,
+      "loss": 3.7868,
+      "step": 1521
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.58778738838174e-06,
+      "loss": 3.6951,
+      "step": 1522
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.586991614180143e-06,
+      "loss": 3.5673,
+      "step": 1523
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.586195105689406e-06,
+      "loss": 3.7705,
+      "step": 1524
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.58539786303704e-06,
+      "loss": 3.7361,
+      "step": 1525
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.584599886350663e-06,
+      "loss": 3.6574,
+      "step": 1526
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.58380117575802e-06,
+      "loss": 3.7735,
+      "step": 1527
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.583001731386962e-06,
+      "loss": 3.8213,
+      "step": 1528
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.582201553365473e-06,
+      "loss": 3.7624,
+      "step": 1529
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.581400641821643e-06,
+      "loss": 3.7295,
+      "step": 1530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.580598996883681e-06,
+      "loss": 3.7284,
+      "step": 1531
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.579796618679917e-06,
+      "loss": 3.7202,
+      "step": 1532
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.578993507338795e-06,
+      "loss": 3.7395,
+      "step": 1533
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.578189662988877e-06,
+      "loss": 3.7311,
+      "step": 1534
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.577385085758845e-06,
+      "loss": 3.7303,
+      "step": 1535
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.576579775777497e-06,
+      "loss": 3.7352,
+      "step": 1536
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.575773733173744e-06,
+      "loss": 3.7536,
+      "step": 1537
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.574966958076621e-06,
+      "loss": 3.7297,
+      "step": 1538
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.574159450615275e-06,
+      "loss": 3.7187,
+      "step": 1539
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.573351210918976e-06,
+      "loss": 3.6826,
+      "step": 1540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.572542239117105e-06,
+      "loss": 3.7121,
+      "step": 1541
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.57173253533916e-06,
+      "loss": 3.8864,
+      "step": 1542
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.570922099714765e-06,
+      "loss": 3.7991,
+      "step": 1543
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.570110932373652e-06,
+      "loss": 3.6652,
+      "step": 1544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.569299033445673e-06,
+      "loss": 3.799,
+      "step": 1545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.568486403060797e-06,
+      "loss": 3.6705,
+      "step": 1546
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.567673041349112e-06,
+      "loss": 3.7416,
+      "step": 1547
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.566858948440823e-06,
+      "loss": 3.7769,
+      "step": 1548
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.566044124466247e-06,
+      "loss": 3.6388,
+      "step": 1549
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.565228569555824e-06,
+      "loss": 3.7045,
+      "step": 1550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.564412283840106e-06,
+      "loss": 3.8105,
+      "step": 1551
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.56359526744977e-06,
+      "loss": 3.7438,
+      "step": 1552
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.562777520515598e-06,
+      "loss": 3.6917,
+      "step": 1553
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.561959043168501e-06,
+      "loss": 3.6485,
+      "step": 1554
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.561139835539497e-06,
+      "loss": 3.7375,
+      "step": 1555
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.560319897759729e-06,
+      "loss": 3.7383,
+      "step": 1556
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.55949922996045e-06,
+      "loss": 3.7161,
+      "step": 1557
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.558677832273038e-06,
+      "loss": 3.7644,
+      "step": 1558
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.557855704828977e-06,
+      "loss": 3.7258,
+      "step": 1559
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.557032847759879e-06,
+      "loss": 3.6844,
+      "step": 1560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.556209261197466e-06,
+      "loss": 3.7272,
+      "step": 1561
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.555384945273576e-06,
+      "loss": 3.7377,
+      "step": 1562
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.554559900120167e-06,
+      "loss": 3.8123,
+      "step": 1563
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.553734125869316e-06,
+      "loss": 3.8111,
+      "step": 1564
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.552907622653212e-06,
+      "loss": 3.7337,
+      "step": 1565
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.55208039060416e-06,
+      "loss": 3.7221,
+      "step": 1566
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.551252429854586e-06,
+      "loss": 3.6675,
+      "step": 1567
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.550423740537028e-06,
+      "loss": 3.7467,
+      "step": 1568
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.549594322784149e-06,
+      "loss": 3.777,
+      "step": 1569
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.548764176728716e-06,
+      "loss": 3.6784,
+      "step": 1570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.547933302503623e-06,
+      "loss": 3.6346,
+      "step": 1571
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.547101700241878e-06,
+      "loss": 3.6938,
+      "step": 1572
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.546269370076601e-06,
+      "loss": 3.7718,
+      "step": 1573
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.545436312141037e-06,
+      "loss": 3.7205,
+      "step": 1574
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.544602526568539e-06,
+      "loss": 3.7554,
+      "step": 1575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.543768013492578e-06,
+      "loss": 3.7277,
+      "step": 1576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.54293277304675e-06,
+      "loss": 3.5312,
+      "step": 1577
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.542096805364754e-06,
+      "loss": 3.688,
+      "step": 1578
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.541260110580417e-06,
+      "loss": 3.7715,
+      "step": 1579
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.540422688827677e-06,
+      "loss": 3.7595,
+      "step": 1580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.539584540240587e-06,
+      "loss": 3.7301,
+      "step": 1581
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.538745664953319e-06,
+      "loss": 3.7511,
+      "step": 1582
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.537906063100164e-06,
+      "loss": 3.7312,
+      "step": 1583
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.537065734815523e-06,
+      "loss": 3.7473,
+      "step": 1584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.536224680233916e-06,
+      "loss": 3.6966,
+      "step": 1585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.535382899489982e-06,
+      "loss": 3.8881,
+      "step": 1586
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.534540392718472e-06,
+      "loss": 3.7434,
+      "step": 1587
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.533697160054256e-06,
+      "loss": 3.8035,
+      "step": 1588
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.53285320163232e-06,
+      "loss": 3.6275,
+      "step": 1589
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.532008517587764e-06,
+      "loss": 3.7536,
+      "step": 1590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.531163108055807e-06,
+      "loss": 3.7036,
+      "step": 1591
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.53031697317178e-06,
+      "loss": 3.8148,
+      "step": 1592
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.529470113071138e-06,
+      "loss": 3.7922,
+      "step": 1593
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.528622527889441e-06,
+      "loss": 3.7329,
+      "step": 1594
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.527774217762378e-06,
+      "loss": 3.7492,
+      "step": 1595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.52692518282574e-06,
+      "loss": 3.734,
+      "step": 1596
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.526075423215447e-06,
+      "loss": 3.8389,
+      "step": 1597
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.525224939067524e-06,
+      "loss": 3.7643,
+      "step": 1598
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.524373730518122e-06,
+      "loss": 3.7488,
+      "step": 1599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.5235217977035e-06,
+      "loss": 3.6807,
+      "step": 1600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.522669140760036e-06,
+      "loss": 3.7921,
+      "step": 1601
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.521815759824226e-06,
+      "loss": 3.7588,
+      "step": 1602
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.52096165503268e-06,
+      "loss": 3.7179,
+      "step": 1603
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.520106826522123e-06,
+      "loss": 3.7385,
+      "step": 1604
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.519251274429393e-06,
+      "loss": 3.7531,
+      "step": 1605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.518394998891452e-06,
+      "loss": 3.6577,
+      "step": 1606
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.517538000045373e-06,
+      "loss": 3.6597,
+      "step": 1607
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.516680278028343e-06,
+      "loss": 3.6869,
+      "step": 1608
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.515821832977667e-06,
+      "loss": 3.7416,
+      "step": 1609
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.514962665030767e-06,
+      "loss": 3.7134,
+      "step": 1610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.514102774325179e-06,
+      "loss": 3.7992,
+      "step": 1611
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.513242160998555e-06,
+      "loss": 3.8211,
+      "step": 1612
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.512380825188661e-06,
+      "loss": 3.768,
+      "step": 1613
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.51151876703338e-06,
+      "loss": 3.8016,
+      "step": 1614
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.510655986670715e-06,
+      "loss": 3.7618,
+      "step": 1615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.509792484238776e-06,
+      "loss": 3.7631,
+      "step": 1616
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.508928259875796e-06,
+      "loss": 3.7357,
+      "step": 1617
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.50806331372012e-06,
+      "loss": 3.6823,
+      "step": 1618
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.507197645910208e-06,
+      "loss": 3.6786,
+      "step": 1619
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.506331256584637e-06,
+      "loss": 3.7337,
+      "step": 1620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.505464145882103e-06,
+      "loss": 3.7472,
+      "step": 1621
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.50459631394141e-06,
+      "loss": 3.7285,
+      "step": 1622
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.50372776090148e-06,
+      "loss": 3.7048,
+      "step": 1623
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.502858486901354e-06,
+      "loss": 3.6933,
+      "step": 1624
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.501988492080186e-06,
+      "loss": 3.6837,
+      "step": 1625
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.501117776577246e-06,
+      "loss": 3.7109,
+      "step": 1626
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.500246340531917e-06,
+      "loss": 3.7114,
+      "step": 1627
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.4993741840837e-06,
+      "loss": 3.7822,
+      "step": 1628
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.49850130737221e-06,
+      "loss": 3.7786,
+      "step": 1629
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.49762771053718e-06,
+      "loss": 3.7021,
+      "step": 1630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.496753393718453e-06,
+      "loss": 3.8262,
+      "step": 1631
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.495878357055992e-06,
+      "loss": 3.7864,
+      "step": 1632
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.495002600689873e-06,
+      "loss": 3.7267,
+      "step": 1633
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.494126124760288e-06,
+      "loss": 3.7856,
+      "step": 1634
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.493248929407544e-06,
+      "loss": 3.7336,
+      "step": 1635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.492371014772063e-06,
+      "loss": 3.7486,
+      "step": 1636
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.491492380994383e-06,
+      "loss": 3.6903,
+      "step": 1637
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.490613028215154e-06,
+      "loss": 3.8166,
+      "step": 1638
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.489732956575147e-06,
+      "loss": 3.7182,
+      "step": 1639
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.48885216621524e-06,
+      "loss": 3.7646,
+      "step": 1640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.487970657276435e-06,
+      "loss": 3.8637,
+      "step": 1641
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.487088429899843e-06,
+      "loss": 3.7612,
+      "step": 1642
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.486205484226689e-06,
+      "loss": 3.7163,
+      "step": 1643
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.485321820398321e-06,
+      "loss": 3.7774,
+      "step": 1644
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.484437438556194e-06,
+      "loss": 3.7345,
+      "step": 1645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.483552338841877e-06,
+      "loss": 3.8107,
+      "step": 1646
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.482666521397063e-06,
+      "loss": 3.7647,
+      "step": 1647
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.481779986363552e-06,
+      "loss": 3.6324,
+      "step": 1648
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.48089273388326e-06,
+      "loss": 3.7613,
+      "step": 1649
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.48000476409822e-06,
+      "loss": 3.7572,
+      "step": 1650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.47911607715058e-06,
+      "loss": 3.9054,
+      "step": 1651
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.478226673182602e-06,
+      "loss": 3.761,
+      "step": 1652
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.47733655233666e-06,
+      "loss": 3.7444,
+      "step": 1653
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.476445714755247e-06,
+      "loss": 3.8312,
+      "step": 1654
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.475554160580967e-06,
+      "loss": 3.5968,
+      "step": 1655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.474661889956544e-06,
+      "loss": 3.7326,
+      "step": 1656
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.47376890302481e-06,
+      "loss": 3.6439,
+      "step": 1657
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.472875199928717e-06,
+      "loss": 3.737,
+      "step": 1658
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.471980780811329e-06,
+      "loss": 3.6304,
+      "step": 1659
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.471085645815828e-06,
+      "loss": 3.7799,
+      "step": 1660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.470189795085502e-06,
+      "loss": 3.7653,
+      "step": 1661
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.469293228763765e-06,
+      "loss": 3.7202,
+      "step": 1662
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.468395946994138e-06,
+      "loss": 3.7489,
+      "step": 1663
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.467497949920256e-06,
+      "loss": 3.6906,
+      "step": 1664
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.466599237685875e-06,
+      "loss": 3.7199,
+      "step": 1665
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.465699810434861e-06,
+      "loss": 3.8426,
+      "step": 1666
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.464799668311194e-06,
+      "loss": 3.7023,
+      "step": 1667
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.463898811458968e-06,
+      "loss": 3.6764,
+      "step": 1668
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.462997240022397e-06,
+      "loss": 3.6245,
+      "step": 1669
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.4620949541458e-06,
+      "loss": 3.6821,
+      "step": 1670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.46119195397362e-06,
+      "loss": 3.6373,
+      "step": 1671
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.460288239650409e-06,
+      "loss": 3.7607,
+      "step": 1672
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.45938381132083e-06,
+      "loss": 3.7493,
+      "step": 1673
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.458478669129671e-06,
+      "loss": 3.7074,
+      "step": 1674
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.457572813221824e-06,
+      "loss": 3.6872,
+      "step": 1675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.4566662437423e-06,
+      "loss": 3.7645,
+      "step": 1676
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.455758960836223e-06,
+      "loss": 3.6791,
+      "step": 1677
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.454850964648832e-06,
+      "loss": 3.6942,
+      "step": 1678
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.45394225532548e-06,
+      "loss": 3.713,
+      "step": 1679
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.453032833011633e-06,
+      "loss": 3.7924,
+      "step": 1680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.452122697852873e-06,
+      "loss": 3.783,
+      "step": 1681
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.451211849994895e-06,
+      "loss": 3.7484,
+      "step": 1682
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.450300289583507e-06,
+      "loss": 3.6953,
+      "step": 1683
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.449388016764634e-06,
+      "loss": 3.7517,
+      "step": 1684
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.448475031684314e-06,
+      "loss": 3.7641,
+      "step": 1685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.447561334488694e-06,
+      "loss": 3.7989,
+      "step": 1686
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.446646925324043e-06,
+      "loss": 3.8407,
+      "step": 1687
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.44573180433674e-06,
+      "loss": 3.8188,
+      "step": 1688
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.444815971673279e-06,
+      "loss": 3.6545,
+      "step": 1689
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.443899427480264e-06,
+      "loss": 3.7341,
+      "step": 1690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.44298217190442e-06,
+      "loss": 3.7154,
+      "step": 1691
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.442064205092578e-06,
+      "loss": 3.7187,
+      "step": 1692
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.44114552719169e-06,
+      "loss": 3.8689,
+      "step": 1693
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.440226138348818e-06,
+      "loss": 3.7331,
+      "step": 1694
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.439306038711138e-06,
+      "loss": 3.6363,
+      "step": 1695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.43838522842594e-06,
+      "loss": 3.7075,
+      "step": 1696
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.437463707640628e-06,
+      "loss": 3.6272,
+      "step": 1697
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.43654147650272e-06,
+      "loss": 3.7368,
+      "step": 1698
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.435618535159849e-06,
+      "loss": 3.8396,
+      "step": 1699
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.434694883759757e-06,
+      "loss": 3.6488,
+      "step": 1700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.433770522450305e-06,
+      "loss": 3.7341,
+      "step": 1701
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.432845451379465e-06,
+      "loss": 3.6629,
+      "step": 1702
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.431919670695325e-06,
+      "loss": 3.8952,
+      "step": 1703
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.43099318054608e-06,
+      "loss": 3.7802,
+      "step": 1704
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.430065981080047e-06,
+      "loss": 3.7765,
+      "step": 1705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.429138072445652e-06,
+      "loss": 3.7001,
+      "step": 1706
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.428209454791434e-06,
+      "loss": 3.6274,
+      "step": 1707
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.427280128266049e-06,
+      "loss": 3.6632,
+      "step": 1708
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.426350093018263e-06,
+      "loss": 3.7349,
+      "step": 1709
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.425419349196959e-06,
+      "loss": 3.7733,
+      "step": 1710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.424487896951128e-06,
+      "loss": 3.746,
+      "step": 1711
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.42355573642988e-06,
+      "loss": 3.7079,
+      "step": 1712
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.422622867782431e-06,
+      "loss": 3.7804,
+      "step": 1713
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.421689291158123e-06,
+      "loss": 3.7348,
+      "step": 1714
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.420755006706397e-06,
+      "loss": 3.6909,
+      "step": 1715
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.419820014576818e-06,
+      "loss": 3.7756,
+      "step": 1716
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.418884314919059e-06,
+      "loss": 3.6439,
+      "step": 1717
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.417947907882907e-06,
+      "loss": 3.7575,
+      "step": 1718
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.417010793618264e-06,
+      "loss": 3.6987,
+      "step": 1719
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.416072972275144e-06,
+      "loss": 3.8099,
+      "step": 1720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.415134444003673e-06,
+      "loss": 3.7692,
+      "step": 1721
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.41419520895409e-06,
+      "loss": 3.6018,
+      "step": 1722
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.413255267276753e-06,
+      "loss": 3.7576,
+      "step": 1723
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.412314619122124e-06,
+      "loss": 3.7339,
+      "step": 1724
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.411373264640784e-06,
+      "loss": 3.6358,
+      "step": 1725
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.410431203983428e-06,
+      "loss": 3.6402,
+      "step": 1726
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.409488437300858e-06,
+      "loss": 3.7186,
+      "step": 1727
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.408544964743996e-06,
+      "loss": 3.766,
+      "step": 1728
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.407600786463874e-06,
+      "loss": 3.6671,
+      "step": 1729
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.406655902611632e-06,
+      "loss": 3.7782,
+      "step": 1730
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.405710313338534e-06,
+      "loss": 3.7571,
+      "step": 1731
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.404764018795947e-06,
+      "loss": 3.7824,
+      "step": 1732
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.403817019135355e-06,
+      "loss": 3.8983,
+      "step": 1733
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.402869314508354e-06,
+      "loss": 3.6769,
+      "step": 1734
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.401920905066658e-06,
+      "loss": 3.7354,
+      "step": 1735
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.400971790962081e-06,
+      "loss": 3.7019,
+      "step": 1736
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.400021972346563e-06,
+      "loss": 3.7186,
+      "step": 1737
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.399071449372153e-06,
+      "loss": 3.7433,
+      "step": 1738
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.398120222191007e-06,
+      "loss": 3.7615,
+      "step": 1739
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.3971682909554e-06,
+      "loss": 3.8167,
+      "step": 1740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.39621565581772e-06,
+      "loss": 3.6364,
+      "step": 1741
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.395262316930463e-06,
+      "loss": 3.6873,
+      "step": 1742
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.394308274446242e-06,
+      "loss": 3.6932,
+      "step": 1743
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.393353528517778e-06,
+      "loss": 3.6416,
+      "step": 1744
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.392398079297911e-06,
+      "loss": 3.7267,
+      "step": 1745
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.391441926939588e-06,
+      "loss": 3.6981,
+      "step": 1746
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.390485071595871e-06,
+      "loss": 3.8166,
+      "step": 1747
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.389527513419935e-06,
+      "loss": 3.6599,
+      "step": 1748
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.388569252565066e-06,
+      "loss": 3.7574,
+      "step": 1749
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.387610289184662e-06,
+      "loss": 3.7517,
+      "step": 1750
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.386650623432237e-06,
+      "loss": 3.651,
+      "step": 1751
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.38569025546141e-06,
+      "loss": 3.786,
+      "step": 1752
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.384729185425925e-06,
+      "loss": 3.8421,
+      "step": 1753
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.383767413479626e-06,
+      "loss": 3.6921,
+      "step": 1754
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.382804939776474e-06,
+      "loss": 3.6826,
+      "step": 1755
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.381841764470543e-06,
+      "loss": 3.7025,
+      "step": 1756
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.380877887716022e-06,
+      "loss": 3.7781,
+      "step": 1757
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.379913309667204e-06,
+      "loss": 3.7427,
+      "step": 1758
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.378948030478503e-06,
+      "loss": 3.6482,
+      "step": 1759
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.377982050304441e-06,
+      "loss": 3.6892,
+      "step": 1760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.377015369299651e-06,
+      "loss": 3.6268,
+      "step": 1761
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.376047987618883e-06,
+      "loss": 3.7339,
+      "step": 1762
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.375079905416996e-06,
+      "loss": 3.7688,
+      "step": 1763
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.374111122848957e-06,
+      "loss": 3.633,
+      "step": 1764
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.373141640069854e-06,
+      "loss": 3.6792,
+      "step": 1765
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.37217145723488e-06,
+      "loss": 3.7471,
+      "step": 1766
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.371200574499345e-06,
+      "loss": 3.7659,
+      "step": 1767
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.370228992018666e-06,
+      "loss": 3.7013,
+      "step": 1768
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.369256709948378e-06,
+      "loss": 3.7679,
+      "step": 1769
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.368283728444121e-06,
+      "loss": 3.8199,
+      "step": 1770
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.367310047661653e-06,
+      "loss": 3.7482,
+      "step": 1771
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.36633566775684e-06,
+      "loss": 3.6685,
+      "step": 1772
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.365360588885663e-06,
+      "loss": 3.7412,
+      "step": 1773
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.364384811204212e-06,
+      "loss": 3.7659,
+      "step": 1774
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.363408334868692e-06,
+      "loss": 3.6462,
+      "step": 1775
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.362431160035415e-06,
+      "loss": 3.7379,
+      "step": 1776
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.361453286860811e-06,
+      "loss": 3.7207,
+      "step": 1777
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.360474715501418e-06,
+      "loss": 3.7776,
+      "step": 1778
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.359495446113887e-06,
+      "loss": 3.7656,
+      "step": 1779
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.358515478854977e-06,
+      "loss": 3.726,
+      "step": 1780
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.357534813881566e-06,
+      "loss": 3.7712,
+      "step": 1781
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.356553451350636e-06,
+      "loss": 3.7782,
+      "step": 1782
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.355571391419286e-06,
+      "loss": 3.6721,
+      "step": 1783
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.354588634244728e-06,
+      "loss": 3.7548,
+      "step": 1784
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.353605179984276e-06,
+      "loss": 3.7208,
+      "step": 1785
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.352621028795368e-06,
+      "loss": 3.669,
+      "step": 1786
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.351636180835545e-06,
+      "loss": 3.7625,
+      "step": 1787
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.35065063626246e-06,
+      "loss": 3.7326,
+      "step": 1788
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.349664395233884e-06,
+      "loss": 3.7695,
+      "step": 1789
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.348677457907693e-06,
+      "loss": 3.7681,
+      "step": 1790
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.347689824441877e-06,
+      "loss": 3.7412,
+      "step": 1791
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.346701494994537e-06,
+      "loss": 3.6781,
+      "step": 1792
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.345712469723887e-06,
+      "loss": 3.858,
+      "step": 1793
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.34472274878825e-06,
+      "loss": 3.6685,
+      "step": 1794
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.343732332346058e-06,
+      "loss": 3.6769,
+      "step": 1795
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.342741220555862e-06,
+      "loss": 3.6689,
+      "step": 1796
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.34174941357632e-06,
+      "loss": 3.7585,
+      "step": 1797
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.340756911566197e-06,
+      "loss": 3.6814,
+      "step": 1798
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.339763714684378e-06,
+      "loss": 3.6659,
+      "step": 1799
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.338769823089853e-06,
+      "loss": 3.8019,
+      "step": 1800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.337775236941726e-06,
+      "loss": 3.6984,
+      "step": 1801
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.336779956399208e-06,
+      "loss": 3.654,
+      "step": 1802
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.335783981621627e-06,
+      "loss": 3.8028,
+      "step": 1803
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.334787312768419e-06,
+      "loss": 3.6643,
+      "step": 1804
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.33378994999913e-06,
+      "loss": 3.657,
+      "step": 1805
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.332791893473421e-06,
+      "loss": 3.7561,
+      "step": 1806
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.33179314335106e-06,
+      "loss": 3.6635,
+      "step": 1807
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.330793699791929e-06,
+      "loss": 3.817,
+      "step": 1808
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.329793562956017e-06,
+      "loss": 3.6226,
+      "step": 1809
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.32879273300343e-06,
+      "loss": 3.7534,
+      "step": 1810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.327791210094379e-06,
+      "loss": 3.7771,
+      "step": 1811
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.326788994389191e-06,
+      "loss": 3.7964,
+      "step": 1812
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.3257860860483e-06,
+      "loss": 3.6808,
+      "step": 1813
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.324782485232251e-06,
+      "loss": 3.6997,
+      "step": 1814
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.323778192101704e-06,
+      "loss": 3.7621,
+      "step": 1815
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.322773206817425e-06,
+      "loss": 3.7152,
+      "step": 1816
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.321767529540293e-06,
+      "loss": 3.7378,
+      "step": 1817
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.3207611604313e-06,
+      "loss": 3.7477,
+      "step": 1818
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.319754099651542e-06,
+      "loss": 3.8777,
+      "step": 1819
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.318746347362234e-06,
+      "loss": 3.798,
+      "step": 1820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.317737903724695e-06,
+      "loss": 3.5889,
+      "step": 1821
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.31672876890036e-06,
+      "loss": 3.8026,
+      "step": 1822
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.31571894305077e-06,
+      "loss": 3.7209,
+      "step": 1823
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.31470842633758e-06,
+      "loss": 3.7477,
+      "step": 1824
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.313697218922552e-06,
+      "loss": 3.6187,
+      "step": 1825
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.312685320967566e-06,
+      "loss": 3.7665,
+      "step": 1826
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.311672732634601e-06,
+      "loss": 3.7884,
+      "step": 1827
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.310659454085757e-06,
+      "loss": 3.7634,
+      "step": 1828
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.309645485483239e-06,
+      "loss": 3.7234,
+      "step": 1829
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.308630826989364e-06,
+      "loss": 3.6987,
+      "step": 1830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.30761547876656e-06,
+      "loss": 3.7568,
+      "step": 1831
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.306599440977364e-06,
+      "loss": 3.6649,
+      "step": 1832
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.305582713784424e-06,
+      "loss": 3.7242,
+      "step": 1833
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.304565297350501e-06,
+      "loss": 3.6225,
+      "step": 1834
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.30354719183846e-06,
+      "loss": 3.705,
+      "step": 1835
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.302528397411282e-06,
+      "loss": 3.6948,
+      "step": 1836
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.301508914232053e-06,
+      "loss": 3.6889,
+      "step": 1837
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.30048874246398e-06,
+      "loss": 3.6418,
+      "step": 1838
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.299467882270366e-06,
+      "loss": 3.7632,
+      "step": 1839
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.298446333814634e-06,
+      "loss": 3.6214,
+      "step": 1840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.297424097260317e-06,
+      "loss": 3.7497,
+      "step": 1841
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.29640117277105e-06,
+      "loss": 3.7729,
+      "step": 1842
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.295377560510586e-06,
+      "loss": 3.7587,
+      "step": 1843
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.294353260642785e-06,
+      "loss": 3.8354,
+      "step": 1844
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.29332827333162e-06,
+      "loss": 3.7291,
+      "step": 1845
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.292302598741168e-06,
+      "loss": 3.876,
+      "step": 1846
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.291276237035622e-06,
+      "loss": 3.6486,
+      "step": 1847
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.290249188379284e-06,
+      "loss": 3.6991,
+      "step": 1848
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.289221452936564e-06,
+      "loss": 3.7016,
+      "step": 1849
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.288193030871979e-06,
+      "loss": 3.626,
+      "step": 1850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.287163922350163e-06,
+      "loss": 3.7073,
+      "step": 1851
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.286134127535859e-06,
+      "loss": 3.7929,
+      "step": 1852
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.285103646593913e-06,
+      "loss": 3.7701,
+      "step": 1853
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.284072479689286e-06,
+      "loss": 3.8326,
+      "step": 1854
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.28304062698705e-06,
+      "loss": 3.7094,
+      "step": 1855
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.282008088652381e-06,
+      "loss": 3.854,
+      "step": 1856
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.280974864850572e-06,
+      "loss": 3.7909,
+      "step": 1857
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.279940955747022e-06,
+      "loss": 3.7038,
+      "step": 1858
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.278906361507238e-06,
+      "loss": 3.6147,
+      "step": 1859
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.27787108229684e-06,
+      "loss": 3.7363,
+      "step": 1860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.276835118281558e-06,
+      "loss": 3.6466,
+      "step": 1861
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.275798469627224e-06,
+      "loss": 3.6776,
+      "step": 1862
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.274761136499793e-06,
+      "loss": 3.7286,
+      "step": 1863
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.273723119065316e-06,
+      "loss": 3.6591,
+      "step": 1864
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.272684417489963e-06,
+      "loss": 3.7503,
+      "step": 1865
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.27164503194001e-06,
+      "loss": 3.7208,
+      "step": 1866
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.270604962581841e-06,
+      "loss": 3.623,
+      "step": 1867
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.269564209581953e-06,
+      "loss": 3.7662,
+      "step": 1868
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.268522773106949e-06,
+      "loss": 3.7303,
+      "step": 1869
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.267480653323545e-06,
+      "loss": 3.6893,
+      "step": 1870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.266437850398563e-06,
+      "loss": 3.6982,
+      "step": 1871
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.265394364498933e-06,
+      "loss": 3.7907,
+      "step": 1872
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.2643501957917e-06,
+      "loss": 3.8095,
+      "step": 1873
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.263305344444015e-06,
+      "loss": 3.6551,
+      "step": 1874
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.262259810623142e-06,
+      "loss": 3.8045,
+      "step": 1875
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.261213594496443e-06,
+      "loss": 3.7072,
+      "step": 1876
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.260166696231403e-06,
+      "loss": 3.7756,
+      "step": 1877
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.25911911599561e-06,
+      "loss": 3.7292,
+      "step": 1878
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.258070853956759e-06,
+      "loss": 3.7462,
+      "step": 1879
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.257021910282658e-06,
+      "loss": 3.6283,
+      "step": 1880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.255972285141222e-06,
+      "loss": 3.8393,
+      "step": 1881
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.254921978700478e-06,
+      "loss": 3.7192,
+      "step": 1882
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.253870991128555e-06,
+      "loss": 3.6568,
+      "step": 1883
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.252819322593702e-06,
+      "loss": 3.7554,
+      "step": 1884
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.251766973264267e-06,
+      "loss": 3.7298,
+      "step": 1885
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.250713943308711e-06,
+      "loss": 3.7566,
+      "step": 1886
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.249660232895605e-06,
+      "loss": 3.7582,
+      "step": 1887
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.248605842193628e-06,
+      "loss": 3.6904,
+      "step": 1888
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.247550771371568e-06,
+      "loss": 3.7142,
+      "step": 1889
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.246495020598318e-06,
+      "loss": 3.6907,
+      "step": 1890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.245438590042887e-06,
+      "loss": 3.7433,
+      "step": 1891
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.24438147987439e-06,
+      "loss": 3.7001,
+      "step": 1892
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.243323690262048e-06,
+      "loss": 3.7418,
+      "step": 1893
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.242265221375194e-06,
+      "loss": 3.6605,
+      "step": 1894
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.241206073383264e-06,
+      "loss": 3.7684,
+      "step": 1895
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.240146246455814e-06,
+      "loss": 3.6444,
+      "step": 1896
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.239085740762496e-06,
+      "loss": 3.6887,
+      "step": 1897
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.238024556473081e-06,
+      "loss": 3.6547,
+      "step": 1898
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.236962693757444e-06,
+      "loss": 3.6776,
+      "step": 1899
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.235900152785564e-06,
+      "loss": 3.7541,
+      "step": 1900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.234836933727537e-06,
+      "loss": 3.6958,
+      "step": 1901
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.233773036753567e-06,
+      "loss": 3.7661,
+      "step": 1902
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.232708462033958e-06,
+      "loss": 3.7662,
+      "step": 1903
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.231643209739128e-06,
+      "loss": 3.6872,
+      "step": 1904
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.230577280039607e-06,
+      "loss": 3.7967,
+      "step": 1905
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.229510673106029e-06,
+      "loss": 3.8289,
+      "step": 1906
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.228443389109135e-06,
+      "loss": 3.6615,
+      "step": 1907
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.227375428219777e-06,
+      "loss": 3.6492,
+      "step": 1908
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.226306790608918e-06,
+      "loss": 3.7598,
+      "step": 1909
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.225237476447626e-06,
+      "loss": 3.8443,
+      "step": 1910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.224167485907074e-06,
+      "loss": 3.7825,
+      "step": 1911
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.22309681915855e-06,
+      "loss": 3.6608,
+      "step": 1912
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.222025476373445e-06,
+      "loss": 3.6889,
+      "step": 1913
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.220953457723262e-06,
+      "loss": 3.6256,
+      "step": 1914
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.219880763379609e-06,
+      "loss": 3.8102,
+      "step": 1915
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.218807393514207e-06,
+      "loss": 3.7828,
+      "step": 1916
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.217733348298879e-06,
+      "loss": 3.678,
+      "step": 1917
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.216658627905559e-06,
+      "loss": 3.7873,
+      "step": 1918
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.215583232506289e-06,
+      "loss": 3.7336,
+      "step": 1919
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.21450716227322e-06,
+      "loss": 3.6853,
+      "step": 1920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.21343041737861e-06,
+      "loss": 3.6254,
+      "step": 1921
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.212352997994827e-06,
+      "loss": 3.6876,
+      "step": 1922
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.211274904294341e-06,
+      "loss": 3.6378,
+      "step": 1923
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.210196136449738e-06,
+      "loss": 3.5938,
+      "step": 1924
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.209116694633707e-06,
+      "loss": 3.7069,
+      "step": 1925
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.208036579019042e-06,
+      "loss": 3.8643,
+      "step": 1926
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.206955789778653e-06,
+      "loss": 3.7548,
+      "step": 1927
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.205874327085554e-06,
+      "loss": 3.7413,
+      "step": 1928
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.204792191112864e-06,
+      "loss": 3.7095,
+      "step": 1929
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.203709382033814e-06,
+      "loss": 3.6617,
+      "step": 1930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.202625900021738e-06,
+      "loss": 3.7007,
+      "step": 1931
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.201541745250084e-06,
+      "loss": 3.7809,
+      "step": 1932
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.200456917892402e-06,
+      "loss": 3.6925,
+      "step": 1933
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.199371418122354e-06,
+      "loss": 3.701,
+      "step": 1934
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.198285246113705e-06,
+      "loss": 3.6733,
+      "step": 1935
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.197198402040335e-06,
+      "loss": 3.6291,
+      "step": 1936
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.19611088607622e-06,
+      "loss": 3.7462,
+      "step": 1937
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.195022698395454e-06,
+      "loss": 3.6407,
+      "step": 1938
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.193933839172238e-06,
+      "loss": 3.6139,
+      "step": 1939
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.192844308580872e-06,
+      "loss": 3.7687,
+      "step": 1940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.191754106795772e-06,
+      "loss": 3.7016,
+      "step": 1941
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.190663233991456e-06,
+      "loss": 3.6996,
+      "step": 1942
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.189571690342555e-06,
+      "loss": 3.7618,
+      "step": 1943
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.188479476023803e-06,
+      "loss": 3.7172,
+      "step": 1944
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.187386591210042e-06,
+      "loss": 3.7301,
+      "step": 1945
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.186293036076219e-06,
+      "loss": 3.7952,
+      "step": 1946
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.185198810797397e-06,
+      "loss": 3.8819,
+      "step": 1947
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.184103915548737e-06,
+      "loss": 3.6184,
+      "step": 1948
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.183008350505508e-06,
+      "loss": 3.6999,
+      "step": 1949
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.181912115843094e-06,
+      "loss": 3.6908,
+      "step": 1950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.18081521173698e-06,
+      "loss": 3.8461,
+      "step": 1951
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.179717638362756e-06,
+      "loss": 3.7305,
+      "step": 1952
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.17861939589613e-06,
+      "loss": 3.7973,
+      "step": 1953
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.177520484512898e-06,
+      "loss": 3.7363,
+      "step": 1954
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.176420904388982e-06,
+      "loss": 3.686,
+      "step": 1955
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.175320655700407e-06,
+      "loss": 3.7558,
+      "step": 1956
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.17421973862329e-06,
+      "loss": 3.7288,
+      "step": 1957
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.17311815333388e-06,
+      "loss": 3.678,
+      "step": 1958
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.17201590000851e-06,
+      "loss": 3.6769,
+      "step": 1959
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.170912978823634e-06,
+      "loss": 3.7677,
+      "step": 1960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.169809389955806e-06,
+      "loss": 3.6633,
+      "step": 1961
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.16870513358169e-06,
+      "loss": 3.7721,
+      "step": 1962
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.16760020987806e-06,
+      "loss": 3.7031,
+      "step": 1963
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.166494619021788e-06,
+      "loss": 3.6681,
+      "step": 1964
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.165388361189859e-06,
+      "loss": 3.6538,
+      "step": 1965
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.164281436559366e-06,
+      "loss": 3.6552,
+      "step": 1966
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.163173845307504e-06,
+      "loss": 3.6171,
+      "step": 1967
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.162065587611578e-06,
+      "loss": 3.7076,
+      "step": 1968
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.160956663648998e-06,
+      "loss": 3.7024,
+      "step": 1969
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.159847073597282e-06,
+      "loss": 3.6759,
+      "step": 1970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.158736817634054e-06,
+      "loss": 3.697,
+      "step": 1971
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.157625895937045e-06,
+      "loss": 3.7496,
+      "step": 1972
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.156514308684091e-06,
+      "loss": 3.6882,
+      "step": 1973
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.155402056053138e-06,
+      "loss": 3.7847,
+      "step": 1974
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.154289138222233e-06,
+      "loss": 3.6934,
+      "step": 1975
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.153175555369536e-06,
+      "loss": 3.7197,
+      "step": 1976
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.152061307673307e-06,
+      "loss": 3.7217,
+      "step": 1977
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.150946395311916e-06,
+      "loss": 3.5905,
+      "step": 1978
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.149830818463844e-06,
+      "loss": 3.8101,
+      "step": 1979
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.14871457730767e-06,
+      "loss": 3.821,
+      "step": 1980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.14759767202208e-06,
+      "loss": 3.7286,
+      "step": 1981
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.146480102785871e-06,
+      "loss": 3.764,
+      "step": 1982
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.145361869777947e-06,
+      "loss": 3.7484,
+      "step": 1983
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.144242973177311e-06,
+      "loss": 3.7615,
+      "step": 1984
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.143123413163082e-06,
+      "loss": 3.7084,
+      "step": 1985
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.142003189914477e-06,
+      "loss": 3.8491,
+      "step": 1986
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.14088230361082e-06,
+      "loss": 3.7393,
+      "step": 1987
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.139760754431549e-06,
+      "loss": 3.7442,
+      "step": 1988
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.138638542556197e-06,
+      "loss": 3.76,
+      "step": 1989
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.137515668164411e-06,
+      "loss": 3.6998,
+      "step": 1990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.136392131435943e-06,
+      "loss": 3.644,
+      "step": 1991
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.135267932550647e-06,
+      "loss": 3.5994,
+      "step": 1992
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.134143071688488e-06,
+      "loss": 3.7568,
+      "step": 1993
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.133017549029533e-06,
+      "loss": 3.6748,
+      "step": 1994
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.131891364753956e-06,
+      "loss": 3.7461,
+      "step": 1995
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.130764519042036e-06,
+      "loss": 3.8036,
+      "step": 1996
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.129637012074164e-06,
+      "loss": 3.7783,
+      "step": 1997
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.12850884403083e-06,
+      "loss": 3.6905,
+      "step": 1998
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.127380015092633e-06,
+      "loss": 3.623,
+      "step": 1999
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.126250525440273e-06,
+      "loss": 3.6493,
+      "step": 2000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.125120375254564e-06,
+      "loss": 3.7395,
+      "step": 2001
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.123989564716418e-06,
+      "loss": 3.6524,
+      "step": 2002
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.122858094006858e-06,
+      "loss": 3.6769,
+      "step": 2003
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.12172596330701e-06,
+      "loss": 3.8078,
+      "step": 2004
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.120593172798107e-06,
+      "loss": 3.8189,
+      "step": 2005
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.119459722661486e-06,
+      "loss": 3.7041,
+      "step": 2006
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.118325613078592e-06,
+      "loss": 3.7593,
+      "step": 2007
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.117190844230971e-06,
+      "loss": 3.7876,
+      "step": 2008
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.116055416300282e-06,
+      "loss": 3.6832,
+      "step": 2009
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.114919329468283e-06,
+      "loss": 3.6791,
+      "step": 2010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.11378258391684e-06,
+      "loss": 3.8093,
+      "step": 2011
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.112645179827922e-06,
+      "loss": 3.7029,
+      "step": 2012
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.111507117383608e-06,
+      "loss": 3.7005,
+      "step": 2013
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.110368396766078e-06,
+      "loss": 3.7011,
+      "step": 2014
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.109229018157623e-06,
+      "loss": 3.6995,
+      "step": 2015
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.108088981740633e-06,
+      "loss": 3.7897,
+      "step": 2016
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.106948287697604e-06,
+      "loss": 3.7233,
+      "step": 2017
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.105806936211144e-06,
+      "loss": 3.7197,
+      "step": 2018
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.104664927463956e-06,
+      "loss": 3.6497,
+      "step": 2019
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.103522261638857e-06,
+      "loss": 3.7594,
+      "step": 2020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.102378938918764e-06,
+      "loss": 3.7106,
+      "step": 2021
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.101234959486703e-06,
+      "loss": 3.7429,
+      "step": 2022
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.100090323525803e-06,
+      "loss": 3.6771,
+      "step": 2023
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.098945031219297e-06,
+      "loss": 3.6325,
+      "step": 2024
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.097799082750523e-06,
+      "loss": 3.6554,
+      "step": 2025
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.096652478302928e-06,
+      "loss": 3.7743,
+      "step": 2026
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.095505218060062e-06,
+      "loss": 3.6432,
+      "step": 2027
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.094357302205575e-06,
+      "loss": 3.7823,
+      "step": 2028
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.09320873092323e-06,
+      "loss": 3.8117,
+      "step": 2029
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.092059504396889e-06,
+      "loss": 3.7324,
+      "step": 2030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.090909622810522e-06,
+      "loss": 3.7116,
+      "step": 2031
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.089759086348204e-06,
+      "loss": 3.7849,
+      "step": 2032
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.088607895194113e-06,
+      "loss": 3.6624,
+      "step": 2033
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.08745604953253e-06,
+      "loss": 3.6292,
+      "step": 2034
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.086303549547847e-06,
+      "loss": 3.6292,
+      "step": 2035
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.085150395424557e-06,
+      "loss": 3.8024,
+      "step": 2036
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.083996587347254e-06,
+      "loss": 3.7783,
+      "step": 2037
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.082842125500646e-06,
+      "loss": 3.6839,
+      "step": 2038
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.081687010069535e-06,
+      "loss": 3.7388,
+      "step": 2039
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.080531241238836e-06,
+      "loss": 3.7268,
+      "step": 2040
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.079374819193563e-06,
+      "loss": 3.6926,
+      "step": 2041
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.078217744118841e-06,
+      "loss": 3.7129,
+      "step": 2042
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.077060016199888e-06,
+      "loss": 3.7134,
+      "step": 2043
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.075901635622041e-06,
+      "loss": 3.7596,
+      "step": 2044
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.074742602570732e-06,
+      "loss": 3.7155,
+      "step": 2045
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.0735829172315e-06,
+      "loss": 3.7621,
+      "step": 2046
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.07242257978999e-06,
+      "loss": 3.7288,
+      "step": 2047
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.071261590431945e-06,
+      "loss": 3.668,
+      "step": 2048
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.070099949343223e-06,
+      "loss": 3.7881,
+      "step": 2049
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.068937656709775e-06,
+      "loss": 3.6837,
+      "step": 2050
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.067774712717666e-06,
+      "loss": 3.6986,
+      "step": 2051
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.066611117553058e-06,
+      "loss": 3.6987,
+      "step": 2052
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.065446871402221e-06,
+      "loss": 3.6874,
+      "step": 2053
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.06428197445153e-06,
+      "loss": 3.6927,
+      "step": 2054
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.06311642688746e-06,
+      "loss": 3.6277,
+      "step": 2055
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.061950228896594e-06,
+      "loss": 3.6867,
+      "step": 2056
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.060783380665619e-06,
+      "loss": 3.583,
+      "step": 2057
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.059615882381324e-06,
+      "loss": 3.845,
+      "step": 2058
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.058447734230601e-06,
+      "loss": 3.6763,
+      "step": 2059
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.057278936400453e-06,
+      "loss": 3.764,
+      "step": 2060
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.056109489077978e-06,
+      "loss": 3.8378,
+      "step": 2061
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.054939392450382e-06,
+      "loss": 3.7365,
+      "step": 2062
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.053768646704977e-06,
+      "loss": 3.6604,
+      "step": 2063
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.052597252029175e-06,
+      "loss": 3.7554,
+      "step": 2064
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.051425208610496e-06,
+      "loss": 3.7262,
+      "step": 2065
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.05025251663656e-06,
+      "loss": 3.6897,
+      "step": 2066
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.049079176295092e-06,
+      "loss": 3.8272,
+      "step": 2067
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.047905187773922e-06,
+      "loss": 3.7238,
+      "step": 2068
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.046730551260983e-06,
+      "loss": 3.8441,
+      "step": 2069
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.045555266944313e-06,
+      "loss": 3.6056,
+      "step": 2070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.044379335012048e-06,
+      "loss": 3.6974,
+      "step": 2071
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.043202755652436e-06,
+      "loss": 3.6629,
+      "step": 2072
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.042025529053825e-06,
+      "loss": 3.62,
+      "step": 2073
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.040847655404665e-06,
+      "loss": 3.7626,
+      "step": 2074
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.039669134893511e-06,
+      "loss": 3.5699,
+      "step": 2075
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.038489967709023e-06,
+      "loss": 3.6891,
+      "step": 2076
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.03731015403996e-06,
+      "loss": 3.6812,
+      "step": 2077
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.03612969407519e-06,
+      "loss": 3.8056,
+      "step": 2078
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.03494858800368e-06,
+      "loss": 3.7448,
+      "step": 2079
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.033766836014504e-06,
+      "loss": 3.6471,
+      "step": 2080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.032584438296836e-06,
+      "loss": 3.7905,
+      "step": 2081
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.031401395039958e-06,
+      "loss": 3.7379,
+      "step": 2082
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.030217706433249e-06,
+      "loss": 3.8158,
+      "step": 2083
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.029033372666199e-06,
+      "loss": 3.6376,
+      "step": 2084
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.027848393928391e-06,
+      "loss": 3.808,
+      "step": 2085
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.026662770409524e-06,
+      "loss": 3.6794,
+      "step": 2086
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.025476502299389e-06,
+      "loss": 3.5895,
+      "step": 2087
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.024289589787885e-06,
+      "loss": 3.6978,
+      "step": 2088
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.023102033065016e-06,
+      "loss": 3.6021,
+      "step": 2089
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.021913832320888e-06,
+      "loss": 3.7762,
+      "step": 2090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.020724987745703e-06,
+      "loss": 3.6342,
+      "step": 2091
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.019535499529781e-06,
+      "loss": 3.7455,
+      "step": 2092
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.018345367863529e-06,
+      "loss": 3.6126,
+      "step": 2093
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.017154592937464e-06,
+      "loss": 3.7417,
+      "step": 2094
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.015963174942213e-06,
+      "loss": 3.6816,
+      "step": 2095
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.014771114068492e-06,
+      "loss": 3.869,
+      "step": 2096
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.013578410507131e-06,
+      "loss": 3.6833,
+      "step": 2097
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.012385064449059e-06,
+      "loss": 3.8512,
+      "step": 2098
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.011191076085303e-06,
+      "loss": 3.8227,
+      "step": 2099
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.009996445607004e-06,
+      "loss": 3.7381,
+      "step": 2100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.008801173205397e-06,
+      "loss": 3.7914,
+      "step": 2101
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.00760525907182e-06,
+      "loss": 3.8843,
+      "step": 2102
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.006408703397717e-06,
+      "loss": 3.7864,
+      "step": 2103
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.005211506374636e-06,
+      "loss": 3.8137,
+      "step": 2104
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.004013668194222e-06,
+      "loss": 3.81,
+      "step": 2105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.002815189048227e-06,
+      "loss": 3.7702,
+      "step": 2106
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.001616069128504e-06,
+      "loss": 3.932,
+      "step": 2107
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.00041630862701e-06,
+      "loss": 3.7096,
+      "step": 2108
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.999215907735803e-06,
+      "loss": 3.7525,
+      "step": 2109
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.998014866647045e-06,
+      "loss": 3.7364,
+      "step": 2110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.996813185552998e-06,
+      "loss": 3.7199,
+      "step": 2111
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.99561086464603e-06,
+      "loss": 3.7705,
+      "step": 2112
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.994407904118608e-06,
+      "loss": 3.8011,
+      "step": 2113
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.9932043041633e-06,
+      "loss": 3.7447,
+      "step": 2114
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.992000064972785e-06,
+      "loss": 3.6787,
+      "step": 2115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.990795186739838e-06,
+      "loss": 3.8243,
+      "step": 2116
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.989589669657332e-06,
+      "loss": 3.6482,
+      "step": 2117
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.98838351391825e-06,
+      "loss": 3.759,
+      "step": 2118
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.987176719715674e-06,
+      "loss": 3.6936,
+      "step": 2119
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.985969287242791e-06,
+      "loss": 3.7305,
+      "step": 2120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.984761216692884e-06,
+      "loss": 3.6524,
+      "step": 2121
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.983552508259342e-06,
+      "loss": 3.6615,
+      "step": 2122
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.982343162135658e-06,
+      "loss": 3.7571,
+      "step": 2123
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.981133178515424e-06,
+      "loss": 3.7217,
+      "step": 2124
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.979922557592335e-06,
+      "loss": 3.7112,
+      "step": 2125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.97871129956019e-06,
+      "loss": 3.7468,
+      "step": 2126
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.977499404612886e-06,
+      "loss": 3.6761,
+      "step": 2127
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.976286872944425e-06,
+      "loss": 3.7475,
+      "step": 2128
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.97507370474891e-06,
+      "loss": 3.8043,
+      "step": 2129
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.973859900220545e-06,
+      "loss": 3.6632,
+      "step": 2130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.972645459553637e-06,
+      "loss": 3.7454,
+      "step": 2131
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.971430382942597e-06,
+      "loss": 3.7049,
+      "step": 2132
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.970214670581931e-06,
+      "loss": 3.7073,
+      "step": 2133
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.968998322666254e-06,
+      "loss": 3.6031,
+      "step": 2134
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.96778133939028e-06,
+      "loss": 3.8178,
+      "step": 2135
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.966563720948824e-06,
+      "loss": 3.7061,
+      "step": 2136
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.965345467536803e-06,
+      "loss": 3.694,
+      "step": 2137
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.964126579349237e-06,
+      "loss": 3.7673,
+      "step": 2138
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.962907056581245e-06,
+      "loss": 3.6638,
+      "step": 2139
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.96168689942805e-06,
+      "loss": 3.6649,
+      "step": 2140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.960466108084977e-06,
+      "loss": 3.6545,
+      "step": 2141
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.959244682747449e-06,
+      "loss": 3.7523,
+      "step": 2142
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.958022623610994e-06,
+      "loss": 3.7097,
+      "step": 2143
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.956799930871238e-06,
+      "loss": 3.7083,
+      "step": 2144
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.955576604723915e-06,
+      "loss": 3.7665,
+      "step": 2145
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.954352645364852e-06,
+      "loss": 3.7231,
+      "step": 2146
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.953128052989984e-06,
+      "loss": 3.7953,
+      "step": 2147
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.95190282779534e-06,
+      "loss": 3.6788,
+      "step": 2148
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.95067696997706e-06,
+      "loss": 3.7905,
+      "step": 2149
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.94945047973138e-06,
+      "loss": 3.6931,
+      "step": 2150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.948223357254636e-06,
+      "loss": 3.6041,
+      "step": 2151
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.946995602743267e-06,
+      "loss": 3.7181,
+      "step": 2152
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.94576721639381e-06,
+      "loss": 3.8625,
+      "step": 2153
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.944538198402912e-06,
+      "loss": 3.7183,
+      "step": 2154
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.943308548967311e-06,
+      "loss": 3.6367,
+      "step": 2155
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.94207826828385e-06,
+      "loss": 3.7924,
+      "step": 2156
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.940847356549476e-06,
+      "loss": 3.772,
+      "step": 2157
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.939615813961233e-06,
+      "loss": 3.7281,
+      "step": 2158
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.938383640716264e-06,
+      "loss": 3.7645,
+      "step": 2159
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.937150837011822e-06,
+      "loss": 3.6491,
+      "step": 2160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.935917403045251e-06,
+      "loss": 3.7438,
+      "step": 2161
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.934683339014003e-06,
+      "loss": 3.7432,
+      "step": 2162
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.933448645115625e-06,
+      "loss": 3.802,
+      "step": 2163
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.932213321547769e-06,
+      "loss": 3.7027,
+      "step": 2164
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.930977368508187e-06,
+      "loss": 3.7377,
+      "step": 2165
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.929740786194731e-06,
+      "loss": 3.7176,
+      "step": 2166
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.928503574805354e-06,
+      "loss": 3.7784,
+      "step": 2167
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.927265734538111e-06,
+      "loss": 3.6641,
+      "step": 2168
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.926027265591154e-06,
+      "loss": 3.7483,
+      "step": 2169
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.924788168162739e-06,
+      "loss": 3.8929,
+      "step": 2170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.923548442451222e-06,
+      "loss": 3.7391,
+      "step": 2171
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.92230808865506e-06,
+      "loss": 3.6635,
+      "step": 2172
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.921067106972809e-06,
+      "loss": 3.6936,
+      "step": 2173
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.919825497603126e-06,
+      "loss": 3.7192,
+      "step": 2174
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.918583260744768e-06,
+      "loss": 3.7755,
+      "step": 2175
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.917340396596594e-06,
+      "loss": 3.7361,
+      "step": 2176
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.916096905357562e-06,
+      "loss": 3.7771,
+      "step": 2177
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.914852787226736e-06,
+      "loss": 3.7821,
+      "step": 2178
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.913608042403268e-06,
+      "loss": 3.6697,
+      "step": 2179
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.912362671086424e-06,
+      "loss": 3.6901,
+      "step": 2180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.911116673475556e-06,
+      "loss": 3.7969,
+      "step": 2181
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.909870049770133e-06,
+      "loss": 3.7587,
+      "step": 2182
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.908622800169711e-06,
+      "loss": 3.7493,
+      "step": 2183
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.907374924873953e-06,
+      "loss": 3.6347,
+      "step": 2184
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.906126424082616e-06,
+      "loss": 3.6965,
+      "step": 2185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.904877297995565e-06,
+      "loss": 3.7603,
+      "step": 2186
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.903627546812757e-06,
+      "loss": 3.7729,
+      "step": 2187
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.902377170734258e-06,
+      "loss": 3.7832,
+      "step": 2188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.901126169960223e-06,
+      "loss": 3.7511,
+      "step": 2189
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.899874544690921e-06,
+      "loss": 3.7582,
+      "step": 2190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.898622295126706e-06,
+      "loss": 3.7843,
+      "step": 2191
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.897369421468042e-06,
+      "loss": 3.7276,
+      "step": 2192
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.89611592391549e-06,
+      "loss": 3.7911,
+      "step": 2193
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.894861802669714e-06,
+      "loss": 3.7379,
+      "step": 2194
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.893607057931468e-06,
+      "loss": 3.7692,
+      "step": 2195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.892351689901618e-06,
+      "loss": 3.6911,
+      "step": 2196
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.89109569878112e-06,
+      "loss": 3.6518,
+      "step": 2197
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.889839084771036e-06,
+      "loss": 3.7549,
+      "step": 2198
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.888581848072529e-06,
+      "loss": 3.8268,
+      "step": 2199
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.887323988886853e-06,
+      "loss": 3.6837,
+      "step": 2200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.88606550741537e-06,
+      "loss": 3.6489,
+      "step": 2201
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.884806403859539e-06,
+      "loss": 3.7204,
+      "step": 2202
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.883546678420918e-06,
+      "loss": 3.7022,
+      "step": 2203
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.882286331301162e-06,
+      "loss": 3.7481,
+      "step": 2204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.88102536270203e-06,
+      "loss": 3.6946,
+      "step": 2205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.879763772825383e-06,
+      "loss": 3.7925,
+      "step": 2206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.878501561873172e-06,
+      "loss": 3.8007,
+      "step": 2207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.877238730047454e-06,
+      "loss": 3.7074,
+      "step": 2208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.875975277550385e-06,
+      "loss": 3.6856,
+      "step": 2209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.874711204584218e-06,
+      "loss": 3.8018,
+      "step": 2210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.87344651135131e-06,
+      "loss": 3.6518,
+      "step": 2211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.872181198054109e-06,
+      "loss": 3.7657,
+      "step": 2212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.87091526489517e-06,
+      "loss": 3.8154,
+      "step": 2213
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.869648712077146e-06,
+      "loss": 3.8347,
+      "step": 2214
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.868381539802785e-06,
+      "loss": 3.6985,
+      "step": 2215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.86711374827494e-06,
+      "loss": 3.6342,
+      "step": 2216
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.865845337696556e-06,
+      "loss": 3.687,
+      "step": 2217
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.864576308270688e-06,
+      "loss": 3.6683,
+      "step": 2218
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.863306660200474e-06,
+      "loss": 3.7592,
+      "step": 2219
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.862036393689167e-06,
+      "loss": 3.7758,
+      "step": 2220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.86076550894011e-06,
+      "loss": 3.63,
+      "step": 2221
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.859494006156746e-06,
+      "loss": 3.7577,
+      "step": 2222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.858221885542621e-06,
+      "loss": 3.662,
+      "step": 2223
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.856949147301375e-06,
+      "loss": 3.7557,
+      "step": 2224
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.855675791636749e-06,
+      "loss": 3.741,
+      "step": 2225
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.854401818752584e-06,
+      "loss": 3.7077,
+      "step": 2226
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.853127228852817e-06,
+      "loss": 3.7404,
+      "step": 2227
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.851852022141486e-06,
+      "loss": 3.7555,
+      "step": 2228
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.85057619882273e-06,
+      "loss": 3.8241,
+      "step": 2229
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.849299759100777e-06,
+      "loss": 3.8002,
+      "step": 2230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.848022703179965e-06,
+      "loss": 3.8494,
+      "step": 2231
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.846745031264727e-06,
+      "loss": 3.6421,
+      "step": 2232
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.845466743559593e-06,
+      "loss": 3.7479,
+      "step": 2233
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.844187840269188e-06,
+      "loss": 3.6913,
+      "step": 2234
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.842908321598245e-06,
+      "loss": 3.6217,
+      "step": 2235
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.84162818775159e-06,
+      "loss": 3.6939,
+      "step": 2236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.840347438934145e-06,
+      "loss": 3.6623,
+      "step": 2237
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.839066075350933e-06,
+      "loss": 3.6737,
+      "step": 2238
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.83778409720708e-06,
+      "loss": 3.7466,
+      "step": 2239
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.836501504707802e-06,
+      "loss": 3.6757,
+      "step": 2240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.83521829805842e-06,
+      "loss": 3.7817,
+      "step": 2241
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.833934477464348e-06,
+      "loss": 3.7319,
+      "step": 2242
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.832650043131103e-06,
+      "loss": 3.6938,
+      "step": 2243
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.831364995264297e-06,
+      "loss": 3.6703,
+      "step": 2244
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.830079334069643e-06,
+      "loss": 3.7414,
+      "step": 2245
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.82879305975295e-06,
+      "loss": 3.7767,
+      "step": 2246
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.827506172520126e-06,
+      "loss": 3.8063,
+      "step": 2247
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.826218672577175e-06,
+      "loss": 3.7114,
+      "step": 2248
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.824930560130202e-06,
+      "loss": 3.8127,
+      "step": 2249
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.823641835385413e-06,
+      "loss": 3.6661,
+      "step": 2250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.8223524985491e-06,
+      "loss": 3.6949,
+      "step": 2251
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.821062549827669e-06,
+      "loss": 3.6942,
+      "step": 2252
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.81977198942761e-06,
+      "loss": 3.7133,
+      "step": 2253
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.81848081755552e-06,
+      "loss": 3.6796,
+      "step": 2254
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.81718903441809e-06,
+      "loss": 3.7164,
+      "step": 2255
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.815896640222112e-06,
+      "loss": 3.8242,
+      "step": 2256
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.81460363517447e-06,
+      "loss": 3.7508,
+      "step": 2257
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.813310019482149e-06,
+      "loss": 3.75,
+      "step": 2258
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.812015793352235e-06,
+      "loss": 3.6156,
+      "step": 2259
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.810720956991906e-06,
+      "loss": 3.7161,
+      "step": 2260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.809425510608444e-06,
+      "loss": 3.7435,
+      "step": 2261
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.80812945440922e-06,
+      "loss": 3.7469,
+      "step": 2262
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.806832788601712e-06,
+      "loss": 3.7548,
+      "step": 2263
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.805535513393488e-06,
+      "loss": 3.6808,
+      "step": 2264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.80423762899222e-06,
+      "loss": 3.6705,
+      "step": 2265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.802939135605672e-06,
+      "loss": 3.6142,
+      "step": 2266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.80164003344171e-06,
+      "loss": 3.756,
+      "step": 2267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.800340322708291e-06,
+      "loss": 3.8734,
+      "step": 2268
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.79904000361348e-06,
+      "loss": 3.753,
+      "step": 2269
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.797739076365426e-06,
+      "loss": 3.7161,
+      "step": 2270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.79643754117239e-06,
+      "loss": 3.7605,
+      "step": 2271
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.795135398242715e-06,
+      "loss": 3.6005,
+      "step": 2272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.793832647784855e-06,
+      "loss": 3.7025,
+      "step": 2273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.792529290007353e-06,
+      "loss": 3.546,
+      "step": 2274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.79122532511885e-06,
+      "loss": 3.8437,
+      "step": 2275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.78992075332809e-06,
+      "loss": 3.7008,
+      "step": 2276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.788615574843906e-06,
+      "loss": 3.8218,
+      "step": 2277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.787309789875231e-06,
+      "loss": 3.73,
+      "step": 2278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.7860033986311e-06,
+      "loss": 3.6698,
+      "step": 2279
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.78469640132064e-06,
+      "loss": 3.6774,
+      "step": 2280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.783388798153075e-06,
+      "loss": 3.7908,
+      "step": 2281
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.782080589337727e-06,
+      "loss": 3.653,
+      "step": 2282
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.780771775084017e-06,
+      "loss": 3.697,
+      "step": 2283
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.779462355601457e-06,
+      "loss": 3.7061,
+      "step": 2284
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.778152331099665e-06,
+      "loss": 3.755,
+      "step": 2285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.776841701788348e-06,
+      "loss": 3.7455,
+      "step": 2286
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.775530467877312e-06,
+      "loss": 3.6508,
+      "step": 2287
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.774218629576461e-06,
+      "loss": 3.8041,
+      "step": 2288
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.772906187095795e-06,
+      "loss": 3.7364,
+      "step": 2289
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.771593140645411e-06,
+      "loss": 3.726,
+      "step": 2290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.770279490435504e-06,
+      "loss": 3.788,
+      "step": 2291
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.768965236676362e-06,
+      "loss": 3.7218,
+      "step": 2292
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.767650379578371e-06,
+      "loss": 3.7965,
+      "step": 2293
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.766334919352018e-06,
+      "loss": 3.7318,
+      "step": 2294
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.765018856207879e-06,
+      "loss": 3.7368,
+      "step": 2295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.763702190356632e-06,
+      "loss": 3.6258,
+      "step": 2296
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.76238492200905e-06,
+      "loss": 3.6664,
+      "step": 2297
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.761067051376002e-06,
+      "loss": 3.8398,
+      "step": 2298
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.759748578668454e-06,
+      "loss": 3.7535,
+      "step": 2299
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.758429504097469e-06,
+      "loss": 3.7933,
+      "step": 2300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.757109827874205e-06,
+      "loss": 3.7693,
+      "step": 2301
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.755789550209914e-06,
+      "loss": 3.7462,
+      "step": 2302
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.75446867131595e-06,
+      "loss": 3.7959,
+      "step": 2303
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.753147191403762e-06,
+      "loss": 3.7116,
+      "step": 2304
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.751825110684888e-06,
+      "loss": 3.8059,
+      "step": 2305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.750502429370973e-06,
+      "loss": 3.7676,
+      "step": 2306
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.74917914767375e-06,
+      "loss": 3.6656,
+      "step": 2307
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.747855265805053e-06,
+      "loss": 3.5976,
+      "step": 2308
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.746530783976805e-06,
+      "loss": 3.6633,
+      "step": 2309
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.745205702401036e-06,
+      "loss": 3.8414,
+      "step": 2310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.743880021289866e-06,
+      "loss": 3.7451,
+      "step": 2311
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.742553740855507e-06,
+      "loss": 3.6046,
+      "step": 2312
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.741226861310272e-06,
+      "loss": 3.7915,
+      "step": 2313
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.739899382866571e-06,
+      "loss": 3.7377,
+      "step": 2314
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.738571305736905e-06,
+      "loss": 3.7588,
+      "step": 2315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.737242630133878e-06,
+      "loss": 3.7243,
+      "step": 2316
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.73591335627018e-06,
+      "loss": 3.7245,
+      "step": 2317
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.734583484358606e-06,
+      "loss": 3.7482,
+      "step": 2318
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.733253014612041e-06,
+      "loss": 3.7968,
+      "step": 2319
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.73192194724347e-06,
+      "loss": 3.6779,
+      "step": 2320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.730590282465967e-06,
+      "loss": 3.6947,
+      "step": 2321
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.72925802049271e-06,
+      "loss": 3.7103,
+      "step": 2322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.727925161536967e-06,
+      "loss": 3.7955,
+      "step": 2323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.726591705812102e-06,
+      "loss": 3.6121,
+      "step": 2324
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.725257653531577e-06,
+      "loss": 3.6985,
+      "step": 2325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.72392300490895e-06,
+      "loss": 3.6645,
+      "step": 2326
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.722587760157868e-06,
+      "loss": 3.8738,
+      "step": 2327
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.721251919492083e-06,
+      "loss": 3.7394,
+      "step": 2328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.719915483125432e-06,
+      "loss": 3.7392,
+      "step": 2329
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.718578451271856e-06,
+      "loss": 3.6468,
+      "step": 2330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.717240824145388e-06,
+      "loss": 3.6972,
+      "step": 2331
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.715902601960157e-06,
+      "loss": 3.6694,
+      "step": 2332
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.714563784930384e-06,
+      "loss": 3.8071,
+      "step": 2333
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.71322437327039e-06,
+      "loss": 3.8316,
+      "step": 2334
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.71188436719459e-06,
+      "loss": 3.8673,
+      "step": 2335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.71054376691749e-06,
+      "loss": 3.7886,
+      "step": 2336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.709202572653697e-06,
+      "loss": 3.6749,
+      "step": 2337
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.707860784617909e-06,
+      "loss": 3.7973,
+      "step": 2338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.706518403024922e-06,
+      "loss": 3.7954,
+      "step": 2339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.705175428089622e-06,
+      "loss": 3.7416,
+      "step": 2340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.703831860027e-06,
+      "loss": 3.7035,
+      "step": 2341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.702487699052127e-06,
+      "loss": 3.6989,
+      "step": 2342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.701142945380183e-06,
+      "loss": 3.8259,
+      "step": 2343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.699797599226438e-06,
+      "loss": 3.7867,
+      "step": 2344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.698451660806251e-06,
+      "loss": 3.7798,
+      "step": 2345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.697105130335084e-06,
+      "loss": 3.8059,
+      "step": 2346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.695758008028492e-06,
+      "loss": 3.7368,
+      "step": 2347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.694410294102121e-06,
+      "loss": 3.7858,
+      "step": 2348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.693061988771715e-06,
+      "loss": 3.7253,
+      "step": 2349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.691713092253112e-06,
+      "loss": 3.7105,
+      "step": 2350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.690363604762242e-06,
+      "loss": 3.7226,
+      "step": 2351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.689013526515136e-06,
+      "loss": 3.7164,
+      "step": 2352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.687662857727911e-06,
+      "loss": 3.64,
+      "step": 2353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.686311598616788e-06,
+      "loss": 3.5749,
+      "step": 2354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.684959749398074e-06,
+      "loss": 3.6545,
+      "step": 2355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.683607310288176e-06,
+      "loss": 3.673,
+      "step": 2356
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.682254281503593e-06,
+      "loss": 3.7455,
+      "step": 2357
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.680900663260918e-06,
+      "loss": 3.6163,
+      "step": 2358
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.67954645577684e-06,
+      "loss": 3.5932,
+      "step": 2359
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.67819165926814e-06,
+      "loss": 3.7165,
+      "step": 2360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.676836273951699e-06,
+      "loss": 3.8445,
+      "step": 2361
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.675480300044483e-06,
+      "loss": 3.8563,
+      "step": 2362
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.674123737763561e-06,
+      "loss": 3.652,
+      "step": 2363
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.672766587326091e-06,
+      "loss": 3.6455,
+      "step": 2364
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.671408848949327e-06,
+      "loss": 3.7976,
+      "step": 2365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.670050522850618e-06,
+      "loss": 3.7899,
+      "step": 2366
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.668691609247403e-06,
+      "loss": 3.7187,
+      "step": 2367
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.66733210835722e-06,
+      "loss": 3.678,
+      "step": 2368
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.6659720203977e-06,
+      "loss": 3.7915,
+      "step": 2369
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.664611345586563e-06,
+      "loss": 3.7994,
+      "step": 2370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.66325008414163e-06,
+      "loss": 3.7402,
+      "step": 2371
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.661888236280813e-06,
+      "loss": 3.7192,
+      "step": 2372
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.660525802222115e-06,
+      "loss": 3.5899,
+      "step": 2373
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.65916278218364e-06,
+      "loss": 3.6593,
+      "step": 2374
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.657799176383577e-06,
+      "loss": 3.6849,
+      "step": 2375
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.656434985040215e-06,
+      "loss": 3.5566,
+      "step": 2376
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.655070208371934e-06,
+      "loss": 3.6874,
+      "step": 2377
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.653704846597209e-06,
+      "loss": 3.6258,
+      "step": 2378
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.65233889993461e-06,
+      "loss": 3.721,
+      "step": 2379
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.650972368602793e-06,
+      "loss": 3.896,
+      "step": 2380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.649605252820522e-06,
+      "loss": 3.7085,
+      "step": 2381
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.648237552806638e-06,
+      "loss": 3.6903,
+      "step": 2382
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.646869268780085e-06,
+      "loss": 3.6532,
+      "step": 2383
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.645500400959904e-06,
+      "loss": 3.7109,
+      "step": 2384
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.64413094956522e-06,
+      "loss": 3.8335,
+      "step": 2385
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.642760914815256e-06,
+      "loss": 3.7738,
+      "step": 2386
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.64139029692933e-06,
+      "loss": 3.676,
+      "step": 2387
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.640019096126851e-06,
+      "loss": 3.6244,
+      "step": 2388
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.63864731262732e-06,
+      "loss": 3.7699,
+      "step": 2389
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.637274946650336e-06,
+      "loss": 3.6622,
+      "step": 2390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.635901998415586e-06,
+      "loss": 3.7839,
+      "step": 2391
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.634528468142855e-06,
+      "loss": 3.7046,
+      "step": 2392
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.633154356052017e-06,
+      "loss": 3.6972,
+      "step": 2393
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.631779662363041e-06,
+      "loss": 3.8332,
+      "step": 2394
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.63040438729599e-06,
+      "loss": 3.7146,
+      "step": 2395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.629028531071018e-06,
+      "loss": 3.6485,
+      "step": 2396
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.627652093908374e-06,
+      "loss": 3.795,
+      "step": 2397
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.626275076028397e-06,
+      "loss": 3.7062,
+      "step": 2398
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.624897477651524e-06,
+      "loss": 3.7736,
+      "step": 2399
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.623519298998282e-06,
+      "loss": 3.7463,
+      "step": 2400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.622140540289289e-06,
+      "loss": 3.5855,
+      "step": 2401
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.620761201745258e-06,
+      "loss": 3.7105,
+      "step": 2402
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.619381283586994e-06,
+      "loss": 3.6905,
+      "step": 2403
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.618000786035399e-06,
+      "loss": 3.677,
+      "step": 2404
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.616619709311458e-06,
+      "loss": 3.8142,
+      "step": 2405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.615238053636261e-06,
+      "loss": 3.7791,
+      "step": 2406
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.613855819230982e-06,
+      "loss": 3.7746,
+      "step": 2407
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.612473006316888e-06,
+      "loss": 3.7038,
+      "step": 2408
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.611089615115344e-06,
+      "loss": 3.6517,
+      "step": 2409
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.609705645847803e-06,
+      "loss": 3.7509,
+      "step": 2410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.608321098735813e-06,
+      "loss": 3.7842,
+      "step": 2411
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.606935974001009e-06,
+      "loss": 3.7432,
+      "step": 2412
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.605550271865129e-06,
+      "loss": 3.7753,
+      "step": 2413
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.604163992549992e-06,
+      "loss": 3.6343,
+      "step": 2414
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.602777136277519e-06,
+      "loss": 3.6716,
+      "step": 2415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.601389703269716e-06,
+      "loss": 3.6838,
+      "step": 2416
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.600001693748688e-06,
+      "loss": 3.6882,
+      "step": 2417
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.598613107936622e-06,
+      "loss": 3.6829,
+      "step": 2418
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.597223946055811e-06,
+      "loss": 3.7046,
+      "step": 2419
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.595834208328629e-06,
+      "loss": 3.7661,
+      "step": 2420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.594443894977546e-06,
+      "loss": 3.7652,
+      "step": 2421
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.593053006225127e-06,
+      "loss": 3.6361,
+      "step": 2422
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.591661542294024e-06,
+      "loss": 3.7405,
+      "step": 2423
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.590269503406986e-06,
+      "loss": 3.6403,
+      "step": 2424
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.58887688978685e-06,
+      "loss": 3.7605,
+      "step": 2425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.587483701656548e-06,
+      "loss": 3.7393,
+      "step": 2426
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.5860899392391e-06,
+      "loss": 3.6259,
+      "step": 2427
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.584695602757624e-06,
+      "loss": 3.74,
+      "step": 2428
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.583300692435324e-06,
+      "loss": 3.8758,
+      "step": 2429
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.5819052084955e-06,
+      "loss": 3.7696,
+      "step": 2430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.58050915116154e-06,
+      "loss": 3.6525,
+      "step": 2431
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.579112520656928e-06,
+      "loss": 3.675,
+      "step": 2432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.577715317205237e-06,
+      "loss": 3.811,
+      "step": 2433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.576317541030131e-06,
+      "loss": 3.6772,
+      "step": 2434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.574919192355368e-06,
+      "loss": 3.6502,
+      "step": 2435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.5735202714048e-06,
+      "loss": 3.8672,
+      "step": 2436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.572120778402362e-06,
+      "loss": 3.7722,
+      "step": 2437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.570720713572089e-06,
+      "loss": 3.7392,
+      "step": 2438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.569320077138104e-06,
+      "loss": 3.6237,
+      "step": 2439
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.56791886932462e-06,
+      "loss": 3.5999,
+      "step": 2440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.566517090355949e-06,
+      "loss": 3.7325,
+      "step": 2441
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.565114740456482e-06,
+      "loss": 3.773,
+      "step": 2442
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.56371181985071e-06,
+      "loss": 3.6803,
+      "step": 2443
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.562308328763217e-06,
+      "loss": 3.718,
+      "step": 2444
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.560904267418673e-06,
+      "loss": 3.7023,
+      "step": 2445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.559499636041842e-06,
+      "loss": 3.7587,
+      "step": 2446
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.558094434857576e-06,
+      "loss": 3.6266,
+      "step": 2447
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.556688664090822e-06,
+      "loss": 3.68,
+      "step": 2448
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.555282323966616e-06,
+      "loss": 3.6901,
+      "step": 2449
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.553875414710088e-06,
+      "loss": 3.6931,
+      "step": 2450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.552467936546459e-06,
+      "loss": 3.6984,
+      "step": 2451
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.551059889701032e-06,
+      "loss": 3.7519,
+      "step": 2452
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.549651274399215e-06,
+      "loss": 3.6417,
+      "step": 2453
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.548242090866497e-06,
+      "loss": 3.8187,
+      "step": 2454
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.546832339328461e-06,
+      "loss": 3.6904,
+      "step": 2455
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.545422020010782e-06,
+      "loss": 3.6578,
+      "step": 2456
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.544011133139227e-06,
+      "loss": 3.716,
+      "step": 2457
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.54259967893965e-06,
+      "loss": 3.717,
+      "step": 2458
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.541187657637996e-06,
+      "loss": 3.7779,
+      "step": 2459
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.539775069460306e-06,
+      "loss": 3.7805,
+      "step": 2460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.538361914632703e-06,
+      "loss": 3.7956,
+      "step": 2461
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.53694819338141e-06,
+      "loss": 3.704,
+      "step": 2462
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.535533905932739e-06,
+      "loss": 3.6991,
+      "step": 2463
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.534119052513084e-06,
+      "loss": 3.8252,
+      "step": 2464
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.53270363334894e-06,
+      "loss": 3.7578,
+      "step": 2465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.531287648666886e-06,
+      "loss": 3.8081,
+      "step": 2466
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.529871098693596e-06,
+      "loss": 3.7648,
+      "step": 2467
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.52845398365583e-06,
+      "loss": 3.6864,
+      "step": 2468
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.527036303780445e-06,
+      "loss": 3.6907,
+      "step": 2469
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.52561805929438e-06,
+      "loss": 3.6791,
+      "step": 2470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.52419925042467e-06,
+      "loss": 3.8055,
+      "step": 2471
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.52277987739844e-06,
+      "loss": 3.7405,
+      "step": 2472
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.521359940442903e-06,
+      "loss": 3.7616,
+      "step": 2473
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.519939439785364e-06,
+      "loss": 3.6595,
+      "step": 2474
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.51851837565322e-06,
+      "loss": 3.7116,
+      "step": 2475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.517096748273951e-06,
+      "loss": 3.673,
+      "step": 2476
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.515674557875138e-06,
+      "loss": 3.6625,
+      "step": 2477
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.51425180468444e-06,
+      "loss": 3.7636,
+      "step": 2478
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.51282848892962e-06,
+      "loss": 3.626,
+      "step": 2479
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.511404610838519e-06,
+      "loss": 3.816,
+      "step": 2480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.509980170639072e-06,
+      "loss": 3.6582,
+      "step": 2481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.508555168559304e-06,
+      "loss": 3.6911,
+      "step": 2482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.507129604827335e-06,
+      "loss": 3.697,
+      "step": 2483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.505703479671365e-06,
+      "loss": 3.6399,
+      "step": 2484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.504276793319693e-06,
+      "loss": 3.6801,
+      "step": 2485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.502849546000702e-06,
+      "loss": 3.7057,
+      "step": 2486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.501421737942868e-06,
+      "loss": 3.6885,
+      "step": 2487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.499993369374754e-06,
+      "loss": 3.6924,
+      "step": 2488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.498564440525014e-06,
+      "loss": 3.7171,
+      "step": 2489
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.497134951622397e-06,
+      "loss": 3.7135,
+      "step": 2490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.495704902895729e-06,
+      "loss": 3.7879,
+      "step": 2491
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.49427429457394e-06,
+      "loss": 3.6987,
+      "step": 2492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.492843126886039e-06,
+      "loss": 3.7451,
+      "step": 2493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.491411400061132e-06,
+      "loss": 3.79,
+      "step": 2494
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.489979114328407e-06,
+      "loss": 3.6451,
+      "step": 2495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.488546269917145e-06,
+      "loss": 3.6952,
+      "step": 2496
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.48711286705672e-06,
+      "loss": 3.8029,
+      "step": 2497
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.485678905976592e-06,
+      "loss": 3.8007,
+      "step": 2498
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.484244386906308e-06,
+      "loss": 3.7222,
+      "step": 2499
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.48280931007551e-06,
+      "loss": 3.7333,
+      "step": 2500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.481373675713924e-06,
+      "loss": 3.7766,
+      "step": 2501
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.479937484051368e-06,
+      "loss": 3.7788,
+      "step": 2502
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.47850073531775e-06,
+      "loss": 3.6279,
+      "step": 2503
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.477063429743061e-06,
+      "loss": 3.7086,
+      "step": 2504
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.475625567557395e-06,
+      "loss": 3.6229,
+      "step": 2505
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.474187148990917e-06,
+      "loss": 3.7583,
+      "step": 2506
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.472748174273896e-06,
+      "loss": 3.7697,
+      "step": 2507
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.471308643636682e-06,
+      "loss": 3.6239,
+      "step": 2508
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.469868557309714e-06,
+      "loss": 3.7598,
+      "step": 2509
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.468427915523527e-06,
+      "loss": 3.7494,
+      "step": 2510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.466986718508735e-06,
+      "loss": 3.8246,
+      "step": 2511
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.465544966496049e-06,
+      "loss": 3.6316,
+      "step": 2512
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.464102659716268e-06,
+      "loss": 3.6228,
+      "step": 2513
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.46265979840027e-06,
+      "loss": 3.7696,
+      "step": 2514
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.461216382779036e-06,
+      "loss": 3.7169,
+      "step": 2515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.459772413083628e-06,
+      "loss": 3.7761,
+      "step": 2516
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.458327889545196e-06,
+      "loss": 3.7661,
+      "step": 2517
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.456882812394981e-06,
+      "loss": 3.7419,
+      "step": 2518
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.455437181864313e-06,
+      "loss": 3.6951,
+      "step": 2519
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.45399099818461e-06,
+      "loss": 3.6964,
+      "step": 2520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.452544261587374e-06,
+      "loss": 3.6559,
+      "step": 2521
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.451096972304206e-06,
+      "loss": 3.7635,
+      "step": 2522
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.449649130566785e-06,
+      "loss": 3.6891,
+      "step": 2523
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.448200736606883e-06,
+      "loss": 3.7128,
+      "step": 2524
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.446751790656362e-06,
+      "loss": 3.63,
+      "step": 2525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.445302292947169e-06,
+      "loss": 3.7055,
+      "step": 2526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.44385224371134e-06,
+      "loss": 3.7147,
+      "step": 2527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.442401643181e-06,
+      "loss": 3.8214,
+      "step": 2528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.440950491588364e-06,
+      "loss": 3.6951,
+      "step": 2529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.439498789165734e-06,
+      "loss": 3.6512,
+      "step": 2530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.438046536145496e-06,
+      "loss": 3.894,
+      "step": 2531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.43659373276013e-06,
+      "loss": 3.7019,
+      "step": 2532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.435140379242203e-06,
+      "loss": 3.7117,
+      "step": 2533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.433686475824367e-06,
+      "loss": 3.7654,
+      "step": 2534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.432232022739366e-06,
+      "loss": 3.8045,
+      "step": 2535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.430777020220026e-06,
+      "loss": 3.7409,
+      "step": 2536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.42932146849927e-06,
+      "loss": 3.6243,
+      "step": 2537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.4278653678101e-06,
+      "loss": 3.6981,
+      "step": 2538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.426408718385612e-06,
+      "loss": 3.7912,
+      "step": 2539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.424951520458987e-06,
+      "loss": 3.7342,
+      "step": 2540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.423493774263494e-06,
+      "loss": 3.7246,
+      "step": 2541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.42203548003249e-06,
+      "loss": 3.8264,
+      "step": 2542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.42057663799942e-06,
+      "loss": 3.7353,
+      "step": 2543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.419117248397816e-06,
+      "loss": 3.74,
+      "step": 2544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.4176573114613e-06,
+      "loss": 3.6632,
+      "step": 2545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.416196827423577e-06,
+      "loss": 3.6996,
+      "step": 2546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.414735796518445e-06,
+      "loss": 3.7748,
+      "step": 2547
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.413274218979787e-06,
+      "loss": 3.8412,
+      "step": 2548
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.411812095041572e-06,
+      "loss": 3.7009,
+      "step": 2549
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.410349424937856e-06,
+      "loss": 3.6758,
+      "step": 2550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.40888620890279e-06,
+      "loss": 3.7035,
+      "step": 2551
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.407422447170603e-06,
+      "loss": 3.6925,
+      "step": 2552
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.405958139975614e-06,
+      "loss": 3.737,
+      "step": 2553
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.404493287552232e-06,
+      "loss": 3.697,
+      "step": 2554
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.403027890134952e-06,
+      "loss": 3.7287,
+      "step": 2555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.401561947958357e-06,
+      "loss": 3.7464,
+      "step": 2556
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.400095461257113e-06,
+      "loss": 3.7614,
+      "step": 2557
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.39862843026598e-06,
+      "loss": 3.6746,
+      "step": 2558
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.397160855219799e-06,
+      "loss": 3.6809,
+      "step": 2559
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.3956927363535e-06,
+      "loss": 3.6531,
+      "step": 2560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.394224073902102e-06,
+      "loss": 3.797,
+      "step": 2561
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.39275486810071e-06,
+      "loss": 3.7478,
+      "step": 2562
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.391285119184516e-06,
+      "loss": 3.727,
+      "step": 2563
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.389814827388795e-06,
+      "loss": 3.6404,
+      "step": 2564
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.388343992948916e-06,
+      "loss": 3.7479,
+      "step": 2565
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.38687261610033e-06,
+      "loss": 3.7141,
+      "step": 2566
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.385400697078576e-06,
+      "loss": 3.7661,
+      "step": 2567
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.383928236119282e-06,
+      "loss": 3.7318,
+      "step": 2568
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.382455233458157e-06,
+      "loss": 3.7592,
+      "step": 2569
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.380981689331003e-06,
+      "loss": 3.725,
+      "step": 2570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.379507603973705e-06,
+      "loss": 3.6682,
+      "step": 2571
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.378032977622235e-06,
+      "loss": 3.7693,
+      "step": 2572
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.376557810512655e-06,
+      "loss": 3.6936,
+      "step": 2573
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.37508210288111e-06,
+      "loss": 3.6912,
+      "step": 2574
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.373605854963831e-06,
+      "loss": 3.7285,
+      "step": 2575
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.372129066997137e-06,
+      "loss": 3.667,
+      "step": 2576
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.370651739217434e-06,
+      "loss": 3.7147,
+      "step": 2577
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.369173871861213e-06,
+      "loss": 3.7943,
+      "step": 2578
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.367695465165056e-06,
+      "loss": 3.6945,
+      "step": 2579
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.366216519365623e-06,
+      "loss": 3.7309,
+      "step": 2580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.364737034699664e-06,
+      "loss": 3.6255,
+      "step": 2581
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.36325701140402e-06,
+      "loss": 3.7991,
+      "step": 2582
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.361776449715612e-06,
+      "loss": 3.6682,
+      "step": 2583
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.36029534987145e-06,
+      "loss": 3.7275,
+      "step": 2584
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.35881371210863e-06,
+      "loss": 3.739,
+      "step": 2585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.357331536664333e-06,
+      "loss": 3.6472,
+      "step": 2586
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.355848823775827e-06,
+      "loss": 3.6651,
+      "step": 2587
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.354365573680465e-06,
+      "loss": 3.8179,
+      "step": 2588
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.352881786615687e-06,
+      "loss": 3.7082,
+      "step": 2589
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.35139746281902e-06,
+      "loss": 3.7631,
+      "step": 2590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.349912602528077e-06,
+      "loss": 3.6947,
+      "step": 2591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.348427205980552e-06,
+      "loss": 3.6337,
+      "step": 2592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.346941273414231e-06,
+      "loss": 3.793,
+      "step": 2593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.345454805066982e-06,
+      "loss": 3.6416,
+      "step": 2594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.343967801176758e-06,
+      "loss": 3.6136,
+      "step": 2595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.342480261981604e-06,
+      "loss": 3.7397,
+      "step": 2596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.340992187719645e-06,
+      "loss": 3.7002,
+      "step": 2597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.339503578629093e-06,
+      "loss": 3.6046,
+      "step": 2598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.338014434948242e-06,
+      "loss": 3.6988,
+      "step": 2599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.336524756915482e-06,
+      "loss": 3.7381,
+      "step": 2600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.335034544769276e-06,
+      "loss": 3.7784,
+      "step": 2601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.333543798748181e-06,
+      "loss": 3.6646,
+      "step": 2602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.332052519090838e-06,
+      "loss": 3.7491,
+      "step": 2603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.330560706035968e-06,
+      "loss": 3.6768,
+      "step": 2604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.329068359822385e-06,
+      "loss": 3.8068,
+      "step": 2605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.327575480688985e-06,
+      "loss": 3.7404,
+      "step": 2606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.326082068874747e-06,
+      "loss": 3.7116,
+      "step": 2607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.32458812461874e-06,
+      "loss": 3.7462,
+      "step": 2608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.323093648160114e-06,
+      "loss": 3.7038,
+      "step": 2609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.321598639738106e-06,
+      "loss": 3.6861,
+      "step": 2610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.320103099592038e-06,
+      "loss": 3.8275,
+      "step": 2611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.318607027961318e-06,
+      "loss": 3.7989,
+      "step": 2612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.317110425085439e-06,
+      "loss": 3.8933,
+      "step": 2613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.315613291203977e-06,
+      "loss": 3.7418,
+      "step": 2614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.314115626556595e-06,
+      "loss": 3.6943,
+      "step": 2615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.312617431383039e-06,
+      "loss": 3.7819,
+      "step": 2616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.311118705923142e-06,
+      "loss": 3.8544,
+      "step": 2617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.309619450416823e-06,
+      "loss": 3.7679,
+      "step": 2618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.308119665104081e-06,
+      "loss": 3.7395,
+      "step": 2619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.306619350225006e-06,
+      "loss": 3.5412,
+      "step": 2620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.305118506019767e-06,
+      "loss": 3.7574,
+      "step": 2621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.30361713272862e-06,
+      "loss": 3.736,
+      "step": 2622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.302115230591907e-06,
+      "loss": 3.6998,
+      "step": 2623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.300612799850055e-06,
+      "loss": 3.7294,
+      "step": 2624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.299109840743573e-06,
+      "loss": 3.6968,
+      "step": 2625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.297606353513054e-06,
+      "loss": 3.7578,
+      "step": 2626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.296102338399182e-06,
+      "loss": 3.7161,
+      "step": 2627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.294597795642715e-06,
+      "loss": 3.7564,
+      "step": 2628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.293092725484505e-06,
+      "loss": 3.7214,
+      "step": 2629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.291587128165485e-06,
+      "loss": 3.808,
+      "step": 2630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.29008100392667e-06,
+      "loss": 3.6351,
+      "step": 2631
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.288574353009164e-06,
+      "loss": 3.82,
+      "step": 2632
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.287067175654152e-06,
+      "loss": 3.7073,
+      "step": 2633
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.285559472102904e-06,
+      "loss": 3.7592,
+      "step": 2634
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.284051242596774e-06,
+      "loss": 3.6329,
+      "step": 2635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.282542487377199e-06,
+      "loss": 3.7337,
+      "step": 2636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.281033206685705e-06,
+      "loss": 3.7382,
+      "step": 2637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.279523400763895e-06,
+      "loss": 3.6052,
+      "step": 2638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.278013069853465e-06,
+      "loss": 3.7862,
+      "step": 2639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.276502214196187e-06,
+      "loss": 3.7702,
+      "step": 2640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.274990834033918e-06,
+      "loss": 3.7562,
+      "step": 2641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.273478929608603e-06,
+      "loss": 3.9652,
+      "step": 2642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.27196650116227e-06,
+      "loss": 3.7435,
+      "step": 2643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.270453548937028e-06,
+      "loss": 3.7492,
+      "step": 2644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.268940073175072e-06,
+      "loss": 3.7578,
+      "step": 2645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.26742607411868e-06,
+      "loss": 3.7753,
+      "step": 2646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.265911552010215e-06,
+      "loss": 3.7506,
+      "step": 2647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.264396507092121e-06,
+      "loss": 3.6206,
+      "step": 2648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.26288093960693e-06,
+      "loss": 3.7741,
+      "step": 2649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.261364849797253e-06,
+      "loss": 3.7291,
+      "step": 2650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.259848237905788e-06,
+      "loss": 3.6126,
+      "step": 2651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.258331104175317e-06,
+      "loss": 3.5728,
+      "step": 2652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.256813448848699e-06,
+      "loss": 3.7153,
+      "step": 2653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.255295272168888e-06,
+      "loss": 3.7933,
+      "step": 2654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.25377657437891e-06,
+      "loss": 3.6115,
+      "step": 2655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.25225735572188e-06,
+      "loss": 3.7118,
+      "step": 2656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.250737616440997e-06,
+      "loss": 3.642,
+      "step": 2657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.249217356779544e-06,
+      "loss": 3.6439,
+      "step": 2658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.24769657698088e-06,
+      "loss": 3.7749,
+      "step": 2659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.246175277288457e-06,
+      "loss": 3.7925,
+      "step": 2660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.244653457945804e-06,
+      "loss": 3.7501,
+      "step": 2661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.243131119196535e-06,
+      "loss": 3.8436,
+      "step": 2662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.241608261284346e-06,
+      "loss": 3.687,
+      "step": 2663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.240084884453022e-06,
+      "loss": 3.6763,
+      "step": 2664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.23856098894642e-06,
+      "loss": 3.7113,
+      "step": 2665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.23703657500849e-06,
+      "loss": 3.7788,
+      "step": 2666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.235511642883261e-06,
+      "loss": 3.728,
+      "step": 2667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.233986192814846e-06,
+      "loss": 3.6439,
+      "step": 2668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.232460225047438e-06,
+      "loss": 3.6453,
+      "step": 2669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.230933739825318e-06,
+      "loss": 3.6866,
+      "step": 2670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.229406737392843e-06,
+      "loss": 3.7081,
+      "step": 2671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.227879217994461e-06,
+      "loss": 3.6689,
+      "step": 2672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.226351181874695e-06,
+      "loss": 3.674,
+      "step": 2673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.224822629278157e-06,
+      "loss": 3.736,
+      "step": 2674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.223293560449537e-06,
+      "loss": 3.7606,
+      "step": 2675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.221763975633611e-06,
+      "loss": 3.7747,
+      "step": 2676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.220233875075235e-06,
+      "loss": 3.7379,
+      "step": 2677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.218703259019348e-06,
+      "loss": 3.8427,
+      "step": 2678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.217172127710972e-06,
+      "loss": 3.6082,
+      "step": 2679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.215640481395215e-06,
+      "loss": 3.7319,
+      "step": 2680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.21410832031726e-06,
+      "loss": 3.7379,
+      "step": 2681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.21257564472238e-06,
+      "loss": 3.7498,
+      "step": 2682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.211042454855925e-06,
+      "loss": 3.7244,
+      "step": 2683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.20950875096333e-06,
+      "loss": 3.748,
+      "step": 2684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.20797453329011e-06,
+      "loss": 3.5991,
+      "step": 2685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.206439802081864e-06,
+      "loss": 3.6224,
+      "step": 2686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.204904557584275e-06,
+      "loss": 3.7004,
+      "step": 2687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.203368800043103e-06,
+      "loss": 3.5675,
+      "step": 2688
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.201832529704197e-06,
+      "loss": 3.6996,
+      "step": 2689
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.20029574681348e-06,
+      "loss": 3.7469,
+      "step": 2690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.198758451616966e-06,
+      "loss": 3.7982,
+      "step": 2691
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.197220644360745e-06,
+      "loss": 3.7606,
+      "step": 2692
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.195682325290985e-06,
+      "loss": 3.8601,
+      "step": 2693
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.19414349465395e-06,
+      "loss": 3.7015,
+      "step": 2694
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.192604152695972e-06,
+      "loss": 3.791,
+      "step": 2695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.19106429966347e-06,
+      "loss": 3.7099,
+      "step": 2696
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.189523935802948e-06,
+      "loss": 3.7458,
+      "step": 2697
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.187983061360985e-06,
+      "loss": 3.6537,
+      "step": 2698
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.186441676584245e-06,
+      "loss": 3.6921,
+      "step": 2699
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.18489978171948e-06,
+      "loss": 3.7198,
+      "step": 2700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.18335737701351e-06,
+      "loss": 3.5966,
+      "step": 2701
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.18181446271325e-06,
+      "loss": 3.6776,
+      "step": 2702
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.180271039065687e-06,
+      "loss": 3.6808,
+      "step": 2703
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.178727106317899e-06,
+      "loss": 3.6733,
+      "step": 2704
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.177182664717035e-06,
+      "loss": 3.6517,
+      "step": 2705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.17563771451033e-06,
+      "loss": 3.7779,
+      "step": 2706
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.174092255945101e-06,
+      "loss": 3.5975,
+      "step": 2707
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.17254628926875e-06,
+      "loss": 3.6958,
+      "step": 2708
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.170999814728755e-06,
+      "loss": 3.7518,
+      "step": 2709
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.169452832572676e-06,
+      "loss": 3.6216,
+      "step": 2710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.167905343048153e-06,
+      "loss": 3.7194,
+      "step": 2711
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.166357346402914e-06,
+      "loss": 3.8103,
+      "step": 2712
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.164808842884758e-06,
+      "loss": 3.7275,
+      "step": 2713
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.163259832741575e-06,
+      "loss": 3.7665,
+      "step": 2714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.16171031622133e-06,
+      "loss": 3.7885,
+      "step": 2715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.160160293572073e-06,
+      "loss": 3.887,
+      "step": 2716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.158609765041928e-06,
+      "loss": 3.7479,
+      "step": 2717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.15705873087911e-06,
+      "loss": 3.6287,
+      "step": 2718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.155507191331906e-06,
+      "loss": 3.7098,
+      "step": 2719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.15395514664869e-06,
+      "loss": 3.6649,
+      "step": 2720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.152402597077913e-06,
+      "loss": 3.6957,
+      "step": 2721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.150849542868109e-06,
+      "loss": 3.5488,
+      "step": 2722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.149295984267892e-06,
+      "loss": 3.7551,
+      "step": 2723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.147741921525956e-06,
+      "loss": 3.7125,
+      "step": 2724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.14618735489108e-06,
+      "loss": 3.897,
+      "step": 2725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.144632284612116e-06,
+      "loss": 3.8142,
+      "step": 2726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.143076710938003e-06,
+      "loss": 3.7119,
+      "step": 2727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.141520634117756e-06,
+      "loss": 3.7799,
+      "step": 2728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.139964054400476e-06,
+      "loss": 3.6947,
+      "step": 2729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.13840697203534e-06,
+      "loss": 3.7224,
+      "step": 2730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.136849387271609e-06,
+      "loss": 3.7466,
+      "step": 2731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.13529130035862e-06,
+      "loss": 3.7315,
+      "step": 2732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.133732711545793e-06,
+      "loss": 3.6828,
+      "step": 2733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.132173621082629e-06,
+      "loss": 3.6248,
+      "step": 2734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.130614029218706e-06,
+      "loss": 3.6837,
+      "step": 2735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.129053936203688e-06,
+      "loss": 3.6258,
+      "step": 2736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.127493342287316e-06,
+      "loss": 3.6934,
+      "step": 2737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.125932247719407e-06,
+      "loss": 3.5957,
+      "step": 2738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.124370652749867e-06,
+      "loss": 3.7832,
+      "step": 2739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.122808557628674e-06,
+      "loss": 3.7621,
+      "step": 2740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.12124596260589e-06,
+      "loss": 3.6592,
+      "step": 2741
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.119682867931656e-06,
+      "loss": 3.6373,
+      "step": 2742
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.118119273856195e-06,
+      "loss": 3.7646,
+      "step": 2743
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.116555180629808e-06,
+      "loss": 3.6131,
+      "step": 2744
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.114990588502876e-06,
+      "loss": 3.7789,
+      "step": 2745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.11342549772586e-06,
+      "loss": 3.6685,
+      "step": 2746
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.111859908549302e-06,
+      "loss": 3.6929,
+      "step": 2747
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.11029382122382e-06,
+      "loss": 3.7535,
+      "step": 2748
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.108727236000116e-06,
+      "loss": 3.7188,
+      "step": 2749
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.10716015312897e-06,
+      "loss": 3.6686,
+      "step": 2750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.105592572861242e-06,
+      "loss": 3.7768,
+      "step": 2751
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.104024495447873e-06,
+      "loss": 3.7929,
+      "step": 2752
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.10245592113988e-06,
+      "loss": 3.6724,
+      "step": 2753
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.100886850188363e-06,
+      "loss": 3.8311,
+      "step": 2754
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.099317282844496e-06,
+      "loss": 3.7558,
+      "step": 2755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.097747219359543e-06,
+      "loss": 3.7276,
+      "step": 2756
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.096176659984837e-06,
+      "loss": 3.6987,
+      "step": 2757
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.094605604971794e-06,
+      "loss": 3.6395,
+      "step": 2758
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.09303405457191e-06,
+      "loss": 3.6914,
+      "step": 2759
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.091462009036764e-06,
+      "loss": 3.6439,
+      "step": 2760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.089889468618003e-06,
+      "loss": 3.6632,
+      "step": 2761
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.088316433567369e-06,
+      "loss": 3.7264,
+      "step": 2762
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.086742904136668e-06,
+      "loss": 3.7157,
+      "step": 2763
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.085168880577792e-06,
+      "loss": 3.7796,
+      "step": 2764
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.083594363142717e-06,
+      "loss": 3.7902,
+      "step": 2765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.082019352083488e-06,
+      "loss": 3.6899,
+      "step": 2766
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.080443847652236e-06,
+      "loss": 3.7507,
+      "step": 2767
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.078867850101168e-06,
+      "loss": 3.7262,
+      "step": 2768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.077291359682573e-06,
+      "loss": 3.7094,
+      "step": 2769
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.075714376648815e-06,
+      "loss": 3.7426,
+      "step": 2770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.074136901252339e-06,
+      "loss": 3.7195,
+      "step": 2771
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.072558933745668e-06,
+      "loss": 3.6886,
+      "step": 2772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.070980474381405e-06,
+      "loss": 3.7479,
+      "step": 2773
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.069401523412233e-06,
+      "loss": 3.7243,
+      "step": 2774
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.067822081090907e-06,
+      "loss": 3.7608,
+      "step": 2775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.066242147670268e-06,
+      "loss": 3.685,
+      "step": 2776
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.064661723403233e-06,
+      "loss": 3.7383,
+      "step": 2777
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.063080808542799e-06,
+      "loss": 3.6104,
+      "step": 2778
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.061499403342036e-06,
+      "loss": 3.7299,
+      "step": 2779
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.059917508054101e-06,
+      "loss": 3.926,
+      "step": 2780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.05833512293222e-06,
+      "loss": 3.6908,
+      "step": 2781
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.056752248229708e-06,
+      "loss": 3.6963,
+      "step": 2782
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.05516888419995e-06,
+      "loss": 3.6685,
+      "step": 2783
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.05358503109641e-06,
+      "loss": 3.7696,
+      "step": 2784
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.052000689172637e-06,
+      "loss": 3.6862,
+      "step": 2785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.05041585868225e-06,
+      "loss": 3.7054,
+      "step": 2786
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.048830539878952e-06,
+      "loss": 3.6317,
+      "step": 2787
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.04724473301652e-06,
+      "loss": 3.6456,
+      "step": 2788
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.045658438348815e-06,
+      "loss": 3.6748,
+      "step": 2789
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.044071656129768e-06,
+      "loss": 3.6695,
+      "step": 2790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.042484386613395e-06,
+      "loss": 3.7189,
+      "step": 2791
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.040896630053786e-06,
+      "loss": 3.581,
+      "step": 2792
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.039308386705111e-06,
+      "loss": 3.6449,
+      "step": 2793
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.037719656821617e-06,
+      "loss": 3.6881,
+      "step": 2794
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.036130440657628e-06,
+      "loss": 3.6451,
+      "step": 2795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.034540738467549e-06,
+      "loss": 3.587,
+      "step": 2796
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.032950550505858e-06,
+      "loss": 3.642,
+      "step": 2797
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.031359877027116e-06,
+      "loss": 3.7468,
+      "step": 2798
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.02976871828596e-06,
+      "loss": 3.68,
+      "step": 2799
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.0281770745371e-06,
+      "loss": 3.7952,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.026584946035332e-06,
+      "loss": 3.7463,
+      "step": 2801
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.024992333035521e-06,
+      "loss": 3.5983,
+      "step": 2802
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.023399235792618e-06,
+      "loss": 3.7029,
+      "step": 2803
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.021805654561643e-06,
+      "loss": 3.6993,
+      "step": 2804
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.0202115895977e-06,
+      "loss": 3.668,
+      "step": 2805
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.01861704115597e-06,
+      "loss": 3.7577,
+      "step": 2806
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.017022009491705e-06,
+      "loss": 3.7079,
+      "step": 2807
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.015426494860242e-06,
+      "loss": 3.6594,
+      "step": 2808
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.013830497516992e-06,
+      "loss": 3.6392,
+      "step": 2809
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.012234017717441e-06,
+      "loss": 3.7026,
+      "step": 2810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.010637055717159e-06,
+      "loss": 3.6343,
+      "step": 2811
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.009039611771785e-06,
+      "loss": 3.6977,
+      "step": 2812
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.007441686137041e-06,
+      "loss": 3.8205,
+      "step": 2813
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.005843279068724e-06,
+      "loss": 3.7802,
+      "step": 2814
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.004244390822709e-06,
+      "loss": 3.8069,
+      "step": 2815
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.002645021654944e-06,
+      "loss": 3.6638,
+      "step": 2816
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.001045171821462e-06,
+      "loss": 3.7199,
+      "step": 2817
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.999444841578364e-06,
+      "loss": 3.6652,
+      "step": 2818
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.997844031181834e-06,
+      "loss": 3.651,
+      "step": 2819
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.996242740888131e-06,
+      "loss": 3.7255,
+      "step": 2820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.99464097095359e-06,
+      "loss": 3.8044,
+      "step": 2821
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.993038721634626e-06,
+      "loss": 3.6826,
+      "step": 2822
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.991435993187725e-06,
+      "loss": 3.6964,
+      "step": 2823
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.989832785869454e-06,
+      "loss": 3.713,
+      "step": 2824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.988229099936457e-06,
+      "loss": 3.6238,
+      "step": 2825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.986624935645453e-06,
+      "loss": 3.7052,
+      "step": 2826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.985020293253235e-06,
+      "loss": 3.7743,
+      "step": 2827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.98341517301668e-06,
+      "loss": 3.7696,
+      "step": 2828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.981809575192732e-06,
+      "loss": 3.7396,
+      "step": 2829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.98020350003842e-06,
+      "loss": 3.6522,
+      "step": 2830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.978596947810845e-06,
+      "loss": 3.686,
+      "step": 2831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.976989918767183e-06,
+      "loss": 3.7788,
+      "step": 2832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.97538241316469e-06,
+      "loss": 3.7089,
+      "step": 2833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.9737744312607e-06,
+      "loss": 3.6916,
+      "step": 2834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.972165973312613e-06,
+      "loss": 3.683,
+      "step": 2835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.970557039577918e-06,
+      "loss": 3.7108,
+      "step": 2836
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.968947630314172e-06,
+      "loss": 3.7885,
+      "step": 2837
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.967337745779012e-06,
+      "loss": 3.6735,
+      "step": 2838
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.965727386230145e-06,
+      "loss": 3.725,
+      "step": 2839
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.964116551925365e-06,
+      "loss": 3.756,
+      "step": 2840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.962505243122531e-06,
+      "loss": 3.5442,
+      "step": 2841
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.960893460079584e-06,
+      "loss": 3.7853,
+      "step": 2842
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.959281203054541e-06,
+      "loss": 3.6806,
+      "step": 2843
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.957668472305491e-06,
+      "loss": 3.6875,
+      "step": 2844
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.956055268090602e-06,
+      "loss": 3.598,
+      "step": 2845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.954441590668116e-06,
+      "loss": 3.7356,
+      "step": 2846
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.952827440296354e-06,
+      "loss": 3.6044,
+      "step": 2847
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.951212817233709e-06,
+      "loss": 3.732,
+      "step": 2848
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.94959772173865e-06,
+      "loss": 3.7121,
+      "step": 2849
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.947982154069726e-06,
+      "loss": 3.6902,
+      "step": 2850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.946366114485557e-06,
+      "loss": 3.6617,
+      "step": 2851
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.944749603244837e-06,
+      "loss": 3.7012,
+      "step": 2852
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.943132620606342e-06,
+      "loss": 3.7773,
+      "step": 2853
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.941515166828917e-06,
+      "loss": 3.6818,
+      "step": 2854
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.939897242171487e-06,
+      "loss": 3.7849,
+      "step": 2855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.938278846893051e-06,
+      "loss": 3.7875,
+      "step": 2856
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.936659981252681e-06,
+      "loss": 3.633,
+      "step": 2857
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.935040645509529e-06,
+      "loss": 3.6562,
+      "step": 2858
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.933420839922816e-06,
+      "loss": 3.5858,
+      "step": 2859
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.931800564751844e-06,
+      "loss": 3.6631,
+      "step": 2860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.930179820255988e-06,
+      "loss": 3.7064,
+      "step": 2861
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.928558606694698e-06,
+      "loss": 3.5648,
+      "step": 2862
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.926936924327498e-06,
+      "loss": 3.8206,
+      "step": 2863
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.925314773413988e-06,
+      "loss": 3.7175,
+      "step": 2864
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.923692154213847e-06,
+      "loss": 3.671,
+      "step": 2865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.92206906698682e-06,
+      "loss": 3.7181,
+      "step": 2866
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.920445511992734e-06,
+      "loss": 3.8102,
+      "step": 2867
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.918821489491489e-06,
+      "loss": 3.7547,
+      "step": 2868
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.91719699974306e-06,
+      "loss": 3.6252,
+      "step": 2869
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.915572043007497e-06,
+      "loss": 3.6848,
+      "step": 2870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.913946619544924e-06,
+      "loss": 3.7169,
+      "step": 2871
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.91232072961554e-06,
+      "loss": 3.6879,
+      "step": 2872
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.910694373479617e-06,
+      "loss": 3.7799,
+      "step": 2873
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.909067551397507e-06,
+      "loss": 3.6431,
+      "step": 2874
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.90744026362963e-06,
+      "loss": 3.7787,
+      "step": 2875
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.905812510436483e-06,
+      "loss": 3.7076,
+      "step": 2876
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.90418429207864e-06,
+      "loss": 3.8035,
+      "step": 2877
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.902555608816747e-06,
+      "loss": 3.6869,
+      "step": 2878
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.900926460911523e-06,
+      "loss": 3.6809,
+      "step": 2879
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.899296848623766e-06,
+      "loss": 3.7072,
+      "step": 2880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.897666772214345e-06,
+      "loss": 3.7289,
+      "step": 2881
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.896036231944201e-06,
+      "loss": 3.7919,
+      "step": 2882
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.894405228074359e-06,
+      "loss": 3.6769,
+      "step": 2883
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.892773760865901e-06,
+      "loss": 3.6876,
+      "step": 2884
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.891141830580003e-06,
+      "loss": 3.6897,
+      "step": 2885
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.8895094374779e-06,
+      "loss": 3.8846,
+      "step": 2886
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.88787658182091e-06,
+      "loss": 3.6804,
+      "step": 2887
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.886243263870421e-06,
+      "loss": 3.748,
+      "step": 2888
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.884609483887893e-06,
+      "loss": 3.8061,
+      "step": 2889
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.882975242134867e-06,
+      "loss": 3.6781,
+      "step": 2890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.88134053887295e-06,
+      "loss": 3.7176,
+      "step": 2891
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.879705374363831e-06,
+      "loss": 3.5058,
+      "step": 2892
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.878069748869262e-06,
+      "loss": 3.7485,
+      "step": 2893
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.876433662651081e-06,
+      "loss": 3.7845,
+      "step": 2894
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.87479711597119e-06,
+      "loss": 3.7849,
+      "step": 2895
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.873160109091574e-06,
+      "loss": 3.7061,
+      "step": 2896
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.871522642274278e-06,
+      "loss": 3.746,
+      "step": 2897
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.869884715781437e-06,
+      "loss": 3.8813,
+      "step": 2898
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.868246329875248e-06,
+      "loss": 3.6533,
+      "step": 2899
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.866607484817984e-06,
+      "loss": 3.8971,
+      "step": 2900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.864968180871994e-06,
+      "loss": 3.7219,
+      "step": 2901
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.863328418299699e-06,
+      "loss": 3.5737,
+      "step": 2902
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.861688197363593e-06,
+      "loss": 3.5745,
+      "step": 2903
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.860047518326243e-06,
+      "loss": 3.744,
+      "step": 2904
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.858406381450294e-06,
+      "loss": 3.575,
+      "step": 2905
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.856764786998456e-06,
+      "loss": 3.7584,
+      "step": 2906
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.85512273523352e-06,
+      "loss": 3.7361,
+      "step": 2907
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.853480226418345e-06,
+      "loss": 3.6749,
+      "step": 2908
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.851837260815864e-06,
+      "loss": 3.6573,
+      "step": 2909
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.850193838689086e-06,
+      "loss": 3.8047,
+      "step": 2910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.84854996030109e-06,
+      "loss": 3.725,
+      "step": 2911
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.846905625915032e-06,
+      "loss": 3.6567,
+      "step": 2912
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.845260835794136e-06,
+      "loss": 3.7814,
+      "step": 2913
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.8436155902017e-06,
+      "loss": 3.6934,
+      "step": 2914
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.8419698894011e-06,
+      "loss": 3.8128,
+      "step": 2915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.84032373365578e-06,
+      "loss": 3.7376,
+      "step": 2916
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.838677123229254e-06,
+      "loss": 3.6363,
+      "step": 2917
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.837030058385117e-06,
+      "loss": 3.6371,
+      "step": 2918
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.835382539387032e-06,
+      "loss": 3.662,
+      "step": 2919
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.833734566498732e-06,
+      "loss": 3.5863,
+      "step": 2920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.83208613998403e-06,
+      "loss": 3.7727,
+      "step": 2921
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.830437260106803e-06,
+      "loss": 3.7761,
+      "step": 2922
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.828787927131008e-06,
+      "loss": 3.7501,
+      "step": 2923
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.827138141320672e-06,
+      "loss": 3.6547,
+      "step": 2924
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.825487902939891e-06,
+      "loss": 3.8406,
+      "step": 2925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.823837212252839e-06,
+      "loss": 3.6127,
+      "step": 2926
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.822186069523756e-06,
+      "loss": 3.7171,
+      "step": 2927
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.820534475016965e-06,
+      "loss": 3.7049,
+      "step": 2928
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.818882428996848e-06,
+      "loss": 3.641,
+      "step": 2929
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.81722993172787e-06,
+      "loss": 3.7123,
+      "step": 2930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.815576983474562e-06,
+      "loss": 3.6933,
+      "step": 2931
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.813923584501529e-06,
+      "loss": 3.6012,
+      "step": 2932
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.81226973507345e-06,
+      "loss": 3.7683,
+      "step": 2933
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.810615435455073e-06,
+      "loss": 3.6522,
+      "step": 2934
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.808960685911219e-06,
+      "loss": 3.659,
+      "step": 2935
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.807305486706782e-06,
+      "loss": 3.7388,
+      "step": 2936
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.805649838106732e-06,
+      "loss": 3.7891,
+      "step": 2937
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.8039937403761e-06,
+      "loss": 3.6924,
+      "step": 2938
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.802337193779999e-06,
+      "loss": 3.692,
+      "step": 2939
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.800680198583608e-06,
+      "loss": 3.7761,
+      "step": 2940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.799022755052182e-06,
+      "loss": 3.803,
+      "step": 2941
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.797364863451045e-06,
+      "loss": 3.7109,
+      "step": 2942
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.795706524045595e-06,
+      "loss": 3.7399,
+      "step": 2943
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.794047737101298e-06,
+      "loss": 3.763,
+      "step": 2944
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.792388502883694e-06,
+      "loss": 3.6646,
+      "step": 2945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.790728821658398e-06,
+      "loss": 3.6684,
+      "step": 2946
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.789068693691089e-06,
+      "loss": 3.8352,
+      "step": 2947
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.787408119247522e-06,
+      "loss": 3.772,
+      "step": 2948
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.785747098593526e-06,
+      "loss": 3.7531,
+      "step": 2949
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.784085631994995e-06,
+      "loss": 3.7373,
+      "step": 2950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.7824237197179e-06,
+      "loss": 3.73,
+      "step": 2951
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.78076136202828e-06,
+      "loss": 3.7426,
+      "step": 2952
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.779098559192247e-06,
+      "loss": 3.6626,
+      "step": 2953
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.777435311475985e-06,
+      "loss": 3.7279,
+      "step": 2954
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.775771619145745e-06,
+      "loss": 3.782,
+      "step": 2955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.774107482467856e-06,
+      "loss": 3.6889,
+      "step": 2956
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.77244290170871e-06,
+      "loss": 3.7324,
+      "step": 2957
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.77077787713478e-06,
+      "loss": 3.6568,
+      "step": 2958
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.769112409012596e-06,
+      "loss": 3.6624,
+      "step": 2959
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.767446497608775e-06,
+      "loss": 3.7616,
+      "step": 2960
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.765780143189994e-06,
+      "loss": 3.7345,
+      "step": 2961
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.764113346023006e-06,
+      "loss": 3.7557,
+      "step": 2962
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.762446106374632e-06,
+      "loss": 3.7497,
+      "step": 2963
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.760778424511766e-06,
+      "loss": 3.6501,
+      "step": 2964
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.759110300701371e-06,
+      "loss": 3.6795,
+      "step": 2965
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.75744173521048e-06,
+      "loss": 3.61,
+      "step": 2966
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.7557727283062e-06,
+      "loss": 3.6369,
+      "step": 2967
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.754103280255707e-06,
+      "loss": 3.8043,
+      "step": 2968
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.752433391326248e-06,
+      "loss": 3.7059,
+      "step": 2969
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.750763061785139e-06,
+      "loss": 3.7436,
+      "step": 2970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.749092291899768e-06,
+      "loss": 3.7612,
+      "step": 2971
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.747421081937592e-06,
+      "loss": 3.7039,
+      "step": 2972
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.74574943216614e-06,
+      "loss": 3.692,
+      "step": 2973
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.744077342853014e-06,
+      "loss": 3.7607,
+      "step": 2974
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.74240481426588e-06,
+      "loss": 3.7485,
+      "step": 2975
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.740731846672479e-06,
+      "loss": 3.6701,
+      "step": 2976
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.739058440340622e-06,
+      "loss": 3.76,
+      "step": 2977
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.737384595538184e-06,
+      "loss": 3.6756,
+      "step": 2978
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.735710312533121e-06,
+      "loss": 3.7521,
+      "step": 2979
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.734035591593452e-06,
+      "loss": 3.7095,
+      "step": 2980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.732360432987267e-06,
+      "loss": 3.7697,
+      "step": 2981
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.730684836982727e-06,
+      "loss": 3.7097,
+      "step": 2982
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.729008803848062e-06,
+      "loss": 3.7805,
+      "step": 2983
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.727332333851574e-06,
+      "loss": 3.842,
+      "step": 2984
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.725655427261633e-06,
+      "loss": 3.6991,
+      "step": 2985
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.723978084346678e-06,
+      "loss": 3.7433,
+      "step": 2986
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.722300305375222e-06,
+      "loss": 3.7938,
+      "step": 2987
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.720622090615844e-06,
+      "loss": 3.6924,
+      "step": 2988
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.718943440337194e-06,
+      "loss": 3.6706,
+      "step": 2989
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.71726435480799e-06,
+      "loss": 3.6386,
+      "step": 2990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.715584834297023e-06,
+      "loss": 3.7648,
+      "step": 2991
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.713904879073153e-06,
+      "loss": 3.6741,
+      "step": 2992
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.712224489405304e-06,
+      "loss": 3.7159,
+      "step": 2993
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.71054366556248e-06,
+      "loss": 3.7613,
+      "step": 2994
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.708862407813744e-06,
+      "loss": 3.6137,
+      "step": 2995
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.707180716428237e-06,
+      "loss": 3.6325,
+      "step": 2996
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.705498591675162e-06,
+      "loss": 3.6924,
+      "step": 2997
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.703816033823797e-06,
+      "loss": 3.7375,
+      "step": 2998
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.702133043143485e-06,
+      "loss": 3.6364,
+      "step": 2999
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.700449619903643e-06,
+      "loss": 3.7703,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.69876576437375e-06,
+      "loss": 3.7314,
+      "step": 3001
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.697081476823367e-06,
+      "loss": 3.7032,
+      "step": 3002
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.695396757522109e-06,
+      "loss": 3.6893,
+      "step": 3003
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.693711606739667e-06,
+      "loss": 3.62,
+      "step": 3004
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.692026024745806e-06,
+      "loss": 3.6561,
+      "step": 3005
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.690340011810352e-06,
+      "loss": 3.7095,
+      "step": 3006
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.688653568203207e-06,
+      "loss": 3.7459,
+      "step": 3007
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.686966694194334e-06,
+      "loss": 3.6031,
+      "step": 3008
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.68527939005377e-06,
+      "loss": 3.6642,
+      "step": 3009
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.68359165605162e-06,
+      "loss": 3.6462,
+      "step": 3010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.681903492458059e-06,
+      "loss": 3.6247,
+      "step": 3011
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.680214899543328e-06,
+      "loss": 3.8062,
+      "step": 3012
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.678525877577737e-06,
+      "loss": 3.5639,
+      "step": 3013
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.676836426831672e-06,
+      "loss": 3.7289,
+      "step": 3014
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.675146547575577e-06,
+      "loss": 3.7183,
+      "step": 3015
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.673456240079966e-06,
+      "loss": 3.6418,
+      "step": 3016
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.671765504615429e-06,
+      "loss": 3.6525,
+      "step": 3017
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.670074341452622e-06,
+      "loss": 3.7398,
+      "step": 3018
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.668382750862265e-06,
+      "loss": 3.7667,
+      "step": 3019
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.666690733115147e-06,
+      "loss": 3.6824,
+      "step": 3020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.664998288482132e-06,
+      "loss": 3.6883,
+      "step": 3021
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.663305417234146e-06,
+      "loss": 3.5763,
+      "step": 3022
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.661612119642183e-06,
+      "loss": 3.6053,
+      "step": 3023
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.65991839597731e-06,
+      "loss": 3.6943,
+      "step": 3024
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.65822424651066e-06,
+      "loss": 3.6712,
+      "step": 3025
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.656529671513429e-06,
+      "loss": 3.6586,
+      "step": 3026
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.654834671256892e-06,
+      "loss": 3.7802,
+      "step": 3027
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.653139246012383e-06,
+      "loss": 3.6901,
+      "step": 3028
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.651443396051307e-06,
+      "loss": 3.7793,
+      "step": 3029
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.649747121645137e-06,
+      "loss": 3.6326,
+      "step": 3030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.648050423065415e-06,
+      "loss": 3.6576,
+      "step": 3031
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.64635330058375e-06,
+      "loss": 3.6335,
+      "step": 3032
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.644655754471816e-06,
+      "loss": 3.6996,
+      "step": 3033
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.642957785001359e-06,
+      "loss": 3.745,
+      "step": 3034
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.641259392444192e-06,
+      "loss": 3.7052,
+      "step": 3035
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.639560577072194e-06,
+      "loss": 3.6571,
+      "step": 3036
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.637861339157315e-06,
+      "loss": 3.7002,
+      "step": 3037
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.636161678971565e-06,
+      "loss": 3.6295,
+      "step": 3038
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.634461596787032e-06,
+      "loss": 3.6831,
+      "step": 3039
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.632761092875864e-06,
+      "loss": 3.7931,
+      "step": 3040
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.631060167510282e-06,
+      "loss": 3.6393,
+      "step": 3041
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.629358820962567e-06,
+      "loss": 3.7323,
+      "step": 3042
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.627657053505075e-06,
+      "loss": 3.605,
+      "step": 3043
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.625954865410224e-06,
+      "loss": 3.6539,
+      "step": 3044
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.6242522569505036e-06,
+      "loss": 3.7232,
+      "step": 3045
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.622549228398468e-06,
+      "loss": 3.7231,
+      "step": 3046
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.6208457800267396e-06,
+      "loss": 3.7302,
+      "step": 3047
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.619141912108008e-06,
+      "loss": 3.6897,
+      "step": 3048
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.617437624915027e-06,
+      "loss": 3.7073,
+      "step": 3049
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.615732918720622e-06,
+      "loss": 3.8155,
+      "step": 3050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.614027793797686e-06,
+      "loss": 3.6625,
+      "step": 3051
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.612322250419173e-06,
+      "loss": 3.6698,
+      "step": 3052
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.610616288858111e-06,
+      "loss": 3.5889,
+      "step": 3053
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.608909909387588e-06,
+      "loss": 3.7146,
+      "step": 3054
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.607203112280765e-06,
+      "loss": 3.6744,
+      "step": 3055
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.605495897810868e-06,
+      "loss": 3.6667,
+      "step": 3056
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.603788266251187e-06,
+      "loss": 3.6915,
+      "step": 3057
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.602080217875083e-06,
+      "loss": 3.7552,
+      "step": 3058
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.600371752955979e-06,
+      "loss": 3.7223,
+      "step": 3059
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.598662871767371e-06,
+      "loss": 3.6915,
+      "step": 3060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.596953574582815e-06,
+      "loss": 3.6656,
+      "step": 3061
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.595243861675937e-06,
+      "loss": 3.6173,
+      "step": 3062
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5935337333204305e-06,
+      "loss": 3.7299,
+      "step": 3063
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.591823189790054e-06,
+      "loss": 3.6769,
+      "step": 3064
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5901122313586325e-06,
+      "loss": 3.5956,
+      "step": 3065
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.588400858300056e-06,
+      "loss": 3.7361,
+      "step": 3066
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.586689070888284e-06,
+      "loss": 3.7543,
+      "step": 3067
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5849768693973415e-06,
+      "loss": 3.8078,
+      "step": 3068
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.583264254101317e-06,
+      "loss": 3.7907,
+      "step": 3069
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.581551225274368e-06,
+      "loss": 3.5672,
+      "step": 3070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.579837783190718e-06,
+      "loss": 3.7437,
+      "step": 3071
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.578123928124654e-06,
+      "loss": 3.622,
+      "step": 3072
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.576409660350535e-06,
+      "loss": 3.7824,
+      "step": 3073
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.57469498014278e-06,
+      "loss": 3.7616,
+      "step": 3074
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.572979887775875e-06,
+      "loss": 3.7291,
+      "step": 3075
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.571264383524377e-06,
+      "loss": 3.7347,
+      "step": 3076
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.569548467662901e-06,
+      "loss": 3.7953,
+      "step": 3077
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.567832140466137e-06,
+      "loss": 3.6777,
+      "step": 3078
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.566115402208831e-06,
+      "loss": 3.6472,
+      "step": 3079
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.564398253165804e-06,
+      "loss": 3.596,
+      "step": 3080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5626806936119345e-06,
+      "loss": 3.7295,
+      "step": 3081
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.560962723822174e-06,
+      "loss": 3.6551,
+      "step": 3082
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.559244344071535e-06,
+      "loss": 3.6626,
+      "step": 3083
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.557525554635097e-06,
+      "loss": 3.8419,
+      "step": 3084
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.555806355788007e-06,
+      "loss": 3.713,
+      "step": 3085
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.554086747805471e-06,
+      "loss": 3.7798,
+      "step": 3086
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.55236673096277e-06,
+      "loss": 3.7968,
+      "step": 3087
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.550646305535243e-06,
+      "loss": 3.7632,
+      "step": 3088
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5489254717982985e-06,
+      "loss": 3.7498,
+      "step": 3089
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.54720423002741e-06,
+      "loss": 3.7421,
+      "step": 3090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.545482580498111e-06,
+      "loss": 3.7686,
+      "step": 3091
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.543760523486008e-06,
+      "loss": 3.7178,
+      "step": 3092
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.542038059266767e-06,
+      "loss": 3.7142,
+      "step": 3093
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.540315188116125e-06,
+      "loss": 3.6917,
+      "step": 3094
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.538591910309876e-06,
+      "loss": 3.6775,
+      "step": 3095
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5368682261238865e-06,
+      "loss": 3.5973,
+      "step": 3096
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.535144135834085e-06,
+      "loss": 3.6976,
+      "step": 3097
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.533419639716462e-06,
+      "loss": 3.6672,
+      "step": 3098
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.531694738047082e-06,
+      "loss": 3.7184,
+      "step": 3099
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.529969431102063e-06,
+      "loss": 3.5999,
+      "step": 3100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.528243719157597e-06,
+      "loss": 3.7549,
+      "step": 3101
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.526517602489936e-06,
+      "loss": 3.7075,
+      "step": 3102
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.524791081375398e-06,
+      "loss": 3.6842,
+      "step": 3103
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.523064156090367e-06,
+      "loss": 3.8501,
+      "step": 3104
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5213368269112896e-06,
+      "loss": 3.7151,
+      "step": 3105
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.519609094114677e-06,
+      "loss": 3.7599,
+      "step": 3106
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.517880957977106e-06,
+      "loss": 3.6582,
+      "step": 3107
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.51615241877522e-06,
+      "loss": 3.68,
+      "step": 3108
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.514423476785723e-06,
+      "loss": 3.7683,
+      "step": 3109
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.512694132285386e-06,
+      "loss": 3.6833,
+      "step": 3110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.510964385551045e-06,
+      "loss": 3.6952,
+      "step": 3111
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.509234236859597e-06,
+      "loss": 3.7963,
+      "step": 3112
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.507503686488007e-06,
+      "loss": 3.779,
+      "step": 3113
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.505772734713304e-06,
+      "loss": 3.6168,
+      "step": 3114
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.504041381812577e-06,
+      "loss": 3.6509,
+      "step": 3115
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.502309628062984e-06,
+      "loss": 3.6391,
+      "step": 3116
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.500577473741746e-06,
+      "loss": 3.6828,
+      "step": 3117
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.4988449191261445e-06,
+      "loss": 3.7882,
+      "step": 3118
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.497111964493532e-06,
+      "loss": 3.7229,
+      "step": 3119
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.49537861012132e-06,
+      "loss": 3.6636,
+      "step": 3120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.4936448562869856e-06,
+      "loss": 3.7671,
+      "step": 3121
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.491910703268068e-06,
+      "loss": 3.587,
+      "step": 3122
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.490176151342174e-06,
+      "loss": 3.6663,
+      "step": 3123
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.488441200786969e-06,
+      "loss": 3.7633,
+      "step": 3124
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.4867058518801875e-06,
+      "loss": 3.7275,
+      "step": 3125
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.484970104899624e-06,
+      "loss": 3.6206,
+      "step": 3126
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.483233960123139e-06,
+      "loss": 3.8133,
+      "step": 3127
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.481497417828657e-06,
+      "loss": 3.7772,
+      "step": 3128
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.479760478294163e-06,
+      "loss": 3.5628,
+      "step": 3129
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.478023141797709e-06,
+      "loss": 3.8277,
+      "step": 3130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.476285408617408e-06,
+      "loss": 3.712,
+      "step": 3131
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.474547279031439e-06,
+      "loss": 3.6031,
+      "step": 3132
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.472808753318042e-06,
+      "loss": 3.7458,
+      "step": 3133
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.471069831755522e-06,
+      "loss": 3.7232,
+      "step": 3134
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.469330514622246e-06,
+      "loss": 3.7447,
+      "step": 3135
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.467590802196648e-06,
+      "loss": 3.7768,
+      "step": 3136
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4658506947572195e-06,
+      "loss": 3.6512,
+      "step": 3137
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.46411019258252e-06,
+      "loss": 3.6798,
+      "step": 3138
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.46236929595117e-06,
+      "loss": 3.6921,
+      "step": 3139
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.460628005141853e-06,
+      "loss": 3.7247,
+      "step": 3140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.458886320433315e-06,
+      "loss": 3.7331,
+      "step": 3141
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.457144242104369e-06,
+      "loss": 3.6509,
+      "step": 3142
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.455401770433887e-06,
+      "loss": 3.8216,
+      "step": 3143
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.453658905700804e-06,
+      "loss": 3.7582,
+      "step": 3144
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.451915648184122e-06,
+      "loss": 3.7132,
+      "step": 3145
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4501719981629e-06,
+      "loss": 3.6005,
+      "step": 3146
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.448427955916266e-06,
+      "loss": 3.7386,
+      "step": 3147
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4466835217234034e-06,
+      "loss": 3.7238,
+      "step": 3148
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.444938695863566e-06,
+      "loss": 3.7808,
+      "step": 3149
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.443193478616065e-06,
+      "loss": 3.7632,
+      "step": 3150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.441447870260279e-06,
+      "loss": 3.7489,
+      "step": 3151
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4397018710756415e-06,
+      "loss": 3.6414,
+      "step": 3152
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.437955481341659e-06,
+      "loss": 3.7859,
+      "step": 3153
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.436208701337889e-06,
+      "loss": 3.7048,
+      "step": 3154
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4344615313439615e-06,
+      "loss": 3.7768,
+      "step": 3155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.432713971639565e-06,
+      "loss": 3.6695,
+      "step": 3156
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4309660225044465e-06,
+      "loss": 3.7202,
+      "step": 3157
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.429217684218424e-06,
+      "loss": 3.6714,
+      "step": 3158
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.427468957061368e-06,
+      "loss": 3.6399,
+      "step": 3159
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.42571984131322e-06,
+      "loss": 3.6488,
+      "step": 3160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.423970337253978e-06,
+      "loss": 3.7729,
+      "step": 3161
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.422220445163704e-06,
+      "loss": 3.6295,
+      "step": 3162
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.420470165322521e-06,
+      "loss": 3.7644,
+      "step": 3163
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.418719498010618e-06,
+      "loss": 3.5867,
+      "step": 3164
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.41696844350824e-06,
+      "loss": 3.6109,
+      "step": 3165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4152170020957e-06,
+      "loss": 3.8147,
+      "step": 3166
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.413465174053367e-06,
+      "loss": 3.618,
+      "step": 3167
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.411712959661677e-06,
+      "loss": 3.6127,
+      "step": 3168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.409960359201127e-06,
+      "loss": 3.6578,
+      "step": 3169
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.408207372952272e-06,
+      "loss": 3.8292,
+      "step": 3170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.406454001195733e-06,
+      "loss": 3.7566,
+      "step": 3171
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.40470024421219e-06,
+      "loss": 3.7792,
+      "step": 3172
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.402946102282388e-06,
+      "loss": 3.6741,
+      "step": 3173
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4011915756871275e-06,
+      "loss": 3.8298,
+      "step": 3174
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.399436664707279e-06,
+      "loss": 3.56,
+      "step": 3175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.397681369623766e-06,
+      "loss": 3.6953,
+      "step": 3176
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3959256907175805e-06,
+      "loss": 3.643,
+      "step": 3177
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.394169628269771e-06,
+      "loss": 3.6829,
+      "step": 3178
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3924131825614485e-06,
+      "loss": 3.8505,
+      "step": 3179
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.39065635387379e-06,
+      "loss": 3.6337,
+      "step": 3180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.388899142488027e-06,
+      "loss": 3.7635,
+      "step": 3181
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.387141548685455e-06,
+      "loss": 3.7245,
+      "step": 3182
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3853835727474335e-06,
+      "loss": 3.6739,
+      "step": 3183
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.383625214955377e-06,
+      "loss": 3.7224,
+      "step": 3184
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.381866475590769e-06,
+      "loss": 3.7427,
+      "step": 3185
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.380107354935146e-06,
+      "loss": 3.6722,
+      "step": 3186
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.378347853270111e-06,
+      "loss": 3.7506,
+      "step": 3187
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.376587970877327e-06,
+      "loss": 3.7194,
+      "step": 3188
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.374827708038516e-06,
+      "loss": 3.7065,
+      "step": 3189
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.373067065035463e-06,
+      "loss": 3.6956,
+      "step": 3190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.371306042150013e-06,
+      "loss": 3.602,
+      "step": 3191
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.369544639664072e-06,
+      "loss": 3.7435,
+      "step": 3192
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.367782857859606e-06,
+      "loss": 3.7769,
+      "step": 3193
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.366020697018645e-06,
+      "loss": 3.7563,
+      "step": 3194
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.364258157423273e-06,
+      "loss": 3.723,
+      "step": 3195
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.362495239355642e-06,
+      "loss": 3.6688,
+      "step": 3196
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3607319430979585e-06,
+      "loss": 3.6592,
+      "step": 3197
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.358968268932495e-06,
+      "loss": 3.5686,
+      "step": 3198
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.357204217141581e-06,
+      "loss": 3.7407,
+      "step": 3199
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.355439788007604e-06,
+      "loss": 3.707,
+      "step": 3200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.35367498181302e-06,
+      "loss": 3.7564,
+      "step": 3201
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.351909798840339e-06,
+      "loss": 3.7529,
+      "step": 3202
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3501442393721314e-06,
+      "loss": 3.6693,
+      "step": 3203
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.34837830369103e-06,
+      "loss": 3.8248,
+      "step": 3204
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.346611992079728e-06,
+      "loss": 3.7394,
+      "step": 3205
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.344845304820979e-06,
+      "loss": 3.7055,
+      "step": 3206
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.343078242197592e-06,
+      "loss": 3.6302,
+      "step": 3207
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.341310804492443e-06,
+      "loss": 3.6027,
+      "step": 3208
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.339542991988463e-06,
+      "loss": 3.6779,
+      "step": 3209
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.337774804968645e-06,
+      "loss": 3.678,
+      "step": 3210
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.336006243716043e-06,
+      "loss": 3.6947,
+      "step": 3211
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.334237308513766e-06,
+      "loss": 3.7122,
+      "step": 3212
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.332467999644992e-06,
+      "loss": 3.6719,
+      "step": 3213
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.33069831739295e-06,
+      "loss": 3.6527,
+      "step": 3214
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.328928262040932e-06,
+      "loss": 3.7338,
+      "step": 3215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.327157833872291e-06,
+      "loss": 3.728,
+      "step": 3216
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.325387033170438e-06,
+      "loss": 3.6272,
+      "step": 3217
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.323615860218844e-06,
+      "loss": 3.7786,
+      "step": 3218
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.321844315301039e-06,
+      "loss": 3.7217,
+      "step": 3219
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.320072398700615e-06,
+      "loss": 3.6387,
+      "step": 3220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.318300110701221e-06,
+      "loss": 3.7722,
+      "step": 3221
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.316527451586566e-06,
+      "loss": 3.6424,
+      "step": 3222
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.3147544216404175e-06,
+      "loss": 3.6641,
+      "step": 3223
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.312981021146606e-06,
+      "loss": 3.6341,
+      "step": 3224
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.311207250389016e-06,
+      "loss": 3.7069,
+      "step": 3225
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.309433109651597e-06,
+      "loss": 3.7405,
+      "step": 3226
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.307658599218352e-06,
+      "loss": 3.6889,
+      "step": 3227
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.30588371937335e-06,
+      "loss": 3.7572,
+      "step": 3228
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.304108470400711e-06,
+      "loss": 3.7611,
+      "step": 3229
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.302332852584619e-06,
+      "loss": 3.6886,
+      "step": 3230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.300556866209319e-06,
+      "loss": 3.7076,
+      "step": 3231
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.298780511559108e-06,
+      "loss": 3.613,
+      "step": 3232
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.29700378891835e-06,
+      "loss": 3.7438,
+      "step": 3233
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.295226698571461e-06,
+      "loss": 3.738,
+      "step": 3234
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.293449240802922e-06,
+      "loss": 3.6918,
+      "step": 3235
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.291671415897268e-06,
+      "loss": 3.7187,
+      "step": 3236
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.289893224139095e-06,
+      "loss": 3.7507,
+      "step": 3237
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2881146658130575e-06,
+      "loss": 3.7091,
+      "step": 3238
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.286335741203867e-06,
+      "loss": 3.6872,
+      "step": 3239
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.284556450596298e-06,
+      "loss": 3.7959,
+      "step": 3240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.282776794275179e-06,
+      "loss": 3.8619,
+      "step": 3241
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.280996772525398e-06,
+      "loss": 3.7894,
+      "step": 3242
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.279216385631903e-06,
+      "loss": 3.6303,
+      "step": 3243
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2774356338797004e-06,
+      "loss": 3.7214,
+      "step": 3244
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.275654517553854e-06,
+      "loss": 3.7517,
+      "step": 3245
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.273873036939485e-06,
+      "loss": 3.6454,
+      "step": 3246
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2720911923217756e-06,
+      "loss": 3.6922,
+      "step": 3247
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.270308983985963e-06,
+      "loss": 3.694,
+      "step": 3248
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2685264122173474e-06,
+      "loss": 3.7453,
+      "step": 3249
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.266743477301283e-06,
+      "loss": 3.9046,
+      "step": 3250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.264960179523183e-06,
+      "loss": 3.6867,
+      "step": 3251
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2631765191685174e-06,
+      "loss": 3.6582,
+      "step": 3252
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.26139249652282e-06,
+      "loss": 3.6914,
+      "step": 3253
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.259608111871675e-06,
+      "loss": 3.6653,
+      "step": 3254
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2578233655007305e-06,
+      "loss": 3.7125,
+      "step": 3255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2560382576956875e-06,
+      "loss": 3.7468,
+      "step": 3256
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.25425278874231e-06,
+      "loss": 3.6898,
+      "step": 3257
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.252466958926414e-06,
+      "loss": 3.6865,
+      "step": 3258
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.250680768533881e-06,
+      "loss": 3.6676,
+      "step": 3259
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.24889421785064e-06,
+      "loss": 3.5099,
+      "step": 3260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.24710730716269e-06,
+      "loss": 3.7581,
+      "step": 3261
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.245320036756076e-06,
+      "loss": 3.7268,
+      "step": 3262
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2435324069169075e-06,
+      "loss": 3.6898,
+      "step": 3263
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.24174441793135e-06,
+      "loss": 3.7473,
+      "step": 3264
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.239956070085625e-06,
+      "loss": 3.7994,
+      "step": 3265
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.238167363666013e-06,
+      "loss": 3.7084,
+      "step": 3266
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.236378298958853e-06,
+      "loss": 3.8472,
+      "step": 3267
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.234588876250538e-06,
+      "loss": 3.7678,
+      "step": 3268
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.232799095827521e-06,
+      "loss": 3.7379,
+      "step": 3269
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.231008957976311e-06,
+      "loss": 3.7029,
+      "step": 3270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2292184629834735e-06,
+      "loss": 3.6942,
+      "step": 3271
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.227427611135634e-06,
+      "loss": 3.6598,
+      "step": 3272
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.225636402719474e-06,
+      "loss": 3.6847,
+      "step": 3273
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.223844838021731e-06,
+      "loss": 3.7188,
+      "step": 3274
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2220529173291986e-06,
+      "loss": 3.7156,
+      "step": 3275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.22026064092873e-06,
+      "loss": 3.7998,
+      "step": 3276
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.218468009107234e-06,
+      "loss": 3.778,
+      "step": 3277
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.216675022151677e-06,
+      "loss": 3.7462,
+      "step": 3278
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.214881680349081e-06,
+      "loss": 3.6566,
+      "step": 3279
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2130879839865255e-06,
+      "loss": 3.7942,
+      "step": 3280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.211293933351147e-06,
+      "loss": 3.574,
+      "step": 3281
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.209499528730138e-06,
+      "loss": 3.7413,
+      "step": 3282
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.20770477041075e-06,
+      "loss": 3.7871,
+      "step": 3283
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.205909658680287e-06,
+      "loss": 3.6767,
+      "step": 3284
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.204114193826112e-06,
+      "loss": 3.6131,
+      "step": 3285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.202318376135648e-06,
+      "loss": 3.6775,
+      "step": 3286
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2005222058963674e-06,
+      "loss": 3.7567,
+      "step": 3287
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.198725683395802e-06,
+      "loss": 3.824,
+      "step": 3288
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.196928808921543e-06,
+      "loss": 3.633,
+      "step": 3289
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.195131582761233e-06,
+      "loss": 3.6929,
+      "step": 3290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.193334005202575e-06,
+      "loss": 3.6991,
+      "step": 3291
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.191536076533326e-06,
+      "loss": 3.6945,
+      "step": 3292
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.189737797041301e-06,
+      "loss": 3.7189,
+      "step": 3293
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.187939167014367e-06,
+      "loss": 3.635,
+      "step": 3294
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.186140186740454e-06,
+      "loss": 3.7058,
+      "step": 3295
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.184340856507541e-06,
+      "loss": 3.7,
+      "step": 3296
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.182541176603668e-06,
+      "loss": 3.7157,
+      "step": 3297
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.1807411473169275e-06,
+      "loss": 3.8347,
+      "step": 3298
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.1789407689354705e-06,
+      "loss": 3.7398,
+      "step": 3299
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.177140041747503e-06,
+      "loss": 3.6483,
+      "step": 3300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.175338966041288e-06,
+      "loss": 3.7599,
+      "step": 3301
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.173537542105141e-06,
+      "loss": 3.6706,
+      "step": 3302
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.171735770227436e-06,
+      "loss": 3.8019,
+      "step": 3303
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.169933650696602e-06,
+      "loss": 3.6598,
+      "step": 3304
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.168131183801124e-06,
+      "loss": 3.6769,
+      "step": 3305
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.166328369829543e-06,
+      "loss": 3.6927,
+      "step": 3306
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.164525209070452e-06,
+      "loss": 3.6944,
+      "step": 3307
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.162721701812506e-06,
+      "loss": 3.7004,
+      "step": 3308
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.160917848344409e-06,
+      "loss": 3.6569,
+      "step": 3309
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1591136489549265e-06,
+      "loss": 3.7229,
+      "step": 3310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.157309103932873e-06,
+      "loss": 3.7048,
+      "step": 3311
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.155504213567122e-06,
+      "loss": 3.7124,
+      "step": 3312
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.153698978146601e-06,
+      "loss": 3.6298,
+      "step": 3313
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.151893397960297e-06,
+      "loss": 3.6214,
+      "step": 3314
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.150087473297246e-06,
+      "loss": 3.768,
+      "step": 3315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.148281204446541e-06,
+      "loss": 3.711,
+      "step": 3316
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1464745916973325e-06,
+      "loss": 3.627,
+      "step": 3317
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.144667635338824e-06,
+      "loss": 3.7193,
+      "step": 3318
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.142860335660275e-06,
+      "loss": 3.6731,
+      "step": 3319
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.141052692950999e-06,
+      "loss": 3.7275,
+      "step": 3320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.139244707500363e-06,
+      "loss": 3.5925,
+      "step": 3321
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.137436379597796e-06,
+      "loss": 3.7571,
+      "step": 3322
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.13562770953277e-06,
+      "loss": 3.6754,
+      "step": 3323
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.133818697594821e-06,
+      "loss": 3.8204,
+      "step": 3324
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.132009344073538e-06,
+      "loss": 3.7774,
+      "step": 3325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.130199649258564e-06,
+      "loss": 3.662,
+      "step": 3326
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1283896134395946e-06,
+      "loss": 3.8513,
+      "step": 3327
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.126579236906382e-06,
+      "loss": 3.7198,
+      "step": 3328
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.124768519948733e-06,
+      "loss": 3.848,
+      "step": 3329
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.122957462856509e-06,
+      "loss": 3.659,
+      "step": 3330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.121146065919625e-06,
+      "loss": 3.81,
+      "step": 3331
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1193343294280516e-06,
+      "loss": 3.6753,
+      "step": 3332
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.117522253671812e-06,
+      "loss": 3.6575,
+      "step": 3333
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.115709838940983e-06,
+      "loss": 3.6512,
+      "step": 3334
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.113897085525702e-06,
+      "loss": 3.7728,
+      "step": 3335
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.112083993716151e-06,
+      "loss": 3.7824,
+      "step": 3336
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.110270563802575e-06,
+      "loss": 3.6601,
+      "step": 3337
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.108456796075267e-06,
+      "loss": 3.6545,
+      "step": 3338
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.106642690824577e-06,
+      "loss": 3.6521,
+      "step": 3339
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.104828248340907e-06,
+      "loss": 3.7491,
+      "step": 3340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1030134689147165e-06,
+      "loss": 3.7367,
+      "step": 3341
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.101198352836516e-06,
+      "loss": 3.7105,
+      "step": 3342
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.099382900396871e-06,
+      "loss": 3.6405,
+      "step": 3343
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0975671118864e-06,
+      "loss": 3.6876,
+      "step": 3344
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.095750987595775e-06,
+      "loss": 3.6364,
+      "step": 3345
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0939345278157244e-06,
+      "loss": 3.7111,
+      "step": 3346
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0921177328370276e-06,
+      "loss": 3.814,
+      "step": 3347
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.090300602950518e-06,
+      "loss": 3.6871,
+      "step": 3348
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.088483138447084e-06,
+      "loss": 3.8209,
+      "step": 3349
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.086665339617667e-06,
+      "loss": 3.6987,
+      "step": 3350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.08484720675326e-06,
+      "loss": 3.7323,
+      "step": 3351
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.083028740144912e-06,
+      "loss": 3.6343,
+      "step": 3352
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.081209940083725e-06,
+      "loss": 3.7048,
+      "step": 3353
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.079390806860853e-06,
+      "loss": 3.6274,
+      "step": 3354
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.077571340767506e-06,
+      "loss": 3.7348,
+      "step": 3355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.075751542094944e-06,
+      "loss": 3.8084,
+      "step": 3356
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.073931411134482e-06,
+      "loss": 3.6836,
+      "step": 3357
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.072110948177487e-06,
+      "loss": 3.7731,
+      "step": 3358
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.070290153515382e-06,
+      "loss": 3.7874,
+      "step": 3359
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.068469027439642e-06,
+      "loss": 3.7154,
+      "step": 3360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.066647570241791e-06,
+      "loss": 3.6056,
+      "step": 3361
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.06482578221341e-06,
+      "loss": 3.7165,
+      "step": 3362
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.063003663646133e-06,
+      "loss": 3.7916,
+      "step": 3363
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0611812148316475e-06,
+      "loss": 3.7138,
+      "step": 3364
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.05935843606169e-06,
+      "loss": 3.6203,
+      "step": 3365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.057535327628056e-06,
+      "loss": 3.7024,
+      "step": 3366
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.055711889822585e-06,
+      "loss": 3.6853,
+      "step": 3367
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0538881229371795e-06,
+      "loss": 3.6951,
+      "step": 3368
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.052064027263785e-06,
+      "loss": 3.7579,
+      "step": 3369
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0502396030944065e-06,
+      "loss": 3.7967,
+      "step": 3370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.048414850721099e-06,
+      "loss": 3.7775,
+      "step": 3371
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.046589770435971e-06,
+      "loss": 3.776,
+      "step": 3372
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.044764362531181e-06,
+      "loss": 3.5891,
+      "step": 3373
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.042938627298942e-06,
+      "loss": 3.6799,
+      "step": 3374
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.041112565031521e-06,
+      "loss": 3.7097,
+      "step": 3375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.039286176021233e-06,
+      "loss": 3.7468,
+      "step": 3376
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.037459460560449e-06,
+      "loss": 3.6776,
+      "step": 3377
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.035632418941591e-06,
+      "loss": 3.583,
+      "step": 3378
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.033805051457134e-06,
+      "loss": 3.7957,
+      "step": 3379
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.031977358399602e-06,
+      "loss": 3.803,
+      "step": 3380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.030149340061576e-06,
+      "loss": 3.7772,
+      "step": 3381
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.028320996735686e-06,
+      "loss": 3.5963,
+      "step": 3382
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.0264923287146134e-06,
+      "loss": 3.7625,
+      "step": 3383
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.024663336291093e-06,
+      "loss": 3.6885,
+      "step": 3384
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.022834019757913e-06,
+      "loss": 3.8553,
+      "step": 3385
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.02100437940791e-06,
+      "loss": 3.6851,
+      "step": 3386
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.019174415533974e-06,
+      "loss": 3.5379,
+      "step": 3387
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.017344128429049e-06,
+      "loss": 3.7985,
+      "step": 3388
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.015513518386128e-06,
+      "loss": 3.7236,
+      "step": 3389
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.013682585698256e-06,
+      "loss": 3.6571,
+      "step": 3390
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.01185133065853e-06,
+      "loss": 3.6801,
+      "step": 3391
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.0100197535601e-06,
+      "loss": 3.564,
+      "step": 3392
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.008187854696163e-06,
+      "loss": 3.679,
+      "step": 3393
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.006355634359976e-06,
+      "loss": 3.7316,
+      "step": 3394
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.004523092844837e-06,
+      "loss": 3.7426,
+      "step": 3395
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.002690230444104e-06,
+      "loss": 3.7464,
+      "step": 3396
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.0008570474511805e-06,
+      "loss": 3.6962,
+      "step": 3397
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.999023544159525e-06,
+      "loss": 3.6609,
+      "step": 3398
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9971897208626485e-06,
+      "loss": 3.7273,
+      "step": 3399
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9953555778541074e-06,
+      "loss": 3.7914,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.993521115427515e-06,
+      "loss": 3.7171,
+      "step": 3401
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9916863338765315e-06,
+      "loss": 3.7876,
+      "step": 3402
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.989851233494871e-06,
+      "loss": 3.7094,
+      "step": 3403
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.988015814576299e-06,
+      "loss": 3.777,
+      "step": 3404
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.986180077414629e-06,
+      "loss": 3.7034,
+      "step": 3405
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.984344022303728e-06,
+      "loss": 3.7148,
+      "step": 3406
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.982507649537513e-06,
+      "loss": 3.7043,
+      "step": 3407
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.980670959409952e-06,
+      "loss": 3.7702,
+      "step": 3408
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.978833952215064e-06,
+      "loss": 3.7355,
+      "step": 3409
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.976996628246917e-06,
+      "loss": 3.618,
+      "step": 3410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.975158987799633e-06,
+      "loss": 3.6004,
+      "step": 3411
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9733210311673826e-06,
+      "loss": 3.6975,
+      "step": 3412
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9714827586443875e-06,
+      "loss": 3.7913,
+      "step": 3413
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.969644170524918e-06,
+      "loss": 3.702,
+      "step": 3414
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9678052671033e-06,
+      "loss": 3.6694,
+      "step": 3415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.965966048673905e-06,
+      "loss": 3.8384,
+      "step": 3416
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.964126515531153e-06,
+      "loss": 3.6294,
+      "step": 3417
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.962286667969524e-06,
+      "loss": 3.7518,
+      "step": 3418
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.960446506283538e-06,
+      "loss": 3.7379,
+      "step": 3419
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.958606030767771e-06,
+      "loss": 3.7492,
+      "step": 3420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.956765241716847e-06,
+      "loss": 3.655,
+      "step": 3421
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.954924139425441e-06,
+      "loss": 3.6733,
+      "step": 3422
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9530827241882785e-06,
+      "loss": 3.6054,
+      "step": 3423
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.951240996300135e-06,
+      "loss": 3.7125,
+      "step": 3424
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.949398956055836e-06,
+      "loss": 3.7939,
+      "step": 3425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.947556603750255e-06,
+      "loss": 3.7473,
+      "step": 3426
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.945713939678319e-06,
+      "loss": 3.6618,
+      "step": 3427
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9438709641350025e-06,
+      "loss": 3.7733,
+      "step": 3428
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.942027677415329e-06,
+      "loss": 3.7339,
+      "step": 3429
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.940184079814375e-06,
+      "loss": 3.6403,
+      "step": 3430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.938340171627265e-06,
+      "loss": 3.7054,
+      "step": 3431
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9364959531491714e-06,
+      "loss": 3.7962,
+      "step": 3432
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9346514246753206e-06,
+      "loss": 3.7126,
+      "step": 3433
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.932806586500984e-06,
+      "loss": 3.5225,
+      "step": 3434
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.930961438921487e-06,
+      "loss": 3.737,
+      "step": 3435
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9291159822322e-06,
+      "loss": 3.7929,
+      "step": 3436
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.927270216728547e-06,
+      "loss": 3.7622,
+      "step": 3437
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.925424142705997e-06,
+      "loss": 3.7103,
+      "step": 3438
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.923577760460074e-06,
+      "loss": 3.6708,
+      "step": 3439
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.921731070286346e-06,
+      "loss": 3.6823,
+      "step": 3440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.919884072480434e-06,
+      "loss": 3.6601,
+      "step": 3441
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.918036767338006e-06,
+      "loss": 3.6499,
+      "step": 3442
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.91618915515478e-06,
+      "loss": 3.8081,
+      "step": 3443
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.914341236226524e-06,
+      "loss": 3.5629,
+      "step": 3444
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.912493010849053e-06,
+      "loss": 3.6668,
+      "step": 3445
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.910644479318233e-06,
+      "loss": 3.6117,
+      "step": 3446
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.90879564192998e-06,
+      "loss": 3.8553,
+      "step": 3447
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.906946498980254e-06,
+      "loss": 3.7283,
+      "step": 3448
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.905097050765069e-06,
+      "loss": 3.7259,
+      "step": 3449
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.903247297580486e-06,
+      "loss": 3.7238,
+      "step": 3450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.901397239722616e-06,
+      "loss": 3.7309,
+      "step": 3451
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.899546877487616e-06,
+      "loss": 3.7358,
+      "step": 3452
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.8976962111716935e-06,
+      "loss": 3.6716,
+      "step": 3453
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.895845241071106e-06,
+      "loss": 3.731,
+      "step": 3454
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.893993967482158e-06,
+      "loss": 3.7386,
+      "step": 3455
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.8921423907012e-06,
+      "loss": 3.7221,
+      "step": 3456
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.890290511024637e-06,
+      "loss": 3.7494,
+      "step": 3457
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.888438328748918e-06,
+      "loss": 3.7314,
+      "step": 3458
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.886585844170543e-06,
+      "loss": 3.6436,
+      "step": 3459
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.884733057586057e-06,
+      "loss": 3.6872,
+      "step": 3460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.882879969292056e-06,
+      "loss": 3.7235,
+      "step": 3461
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.881026579585186e-06,
+      "loss": 3.763,
+      "step": 3462
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.879172888762138e-06,
+      "loss": 3.6615,
+      "step": 3463
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.8773188971196515e-06,
+      "loss": 3.7205,
+      "step": 3464
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.875464604954513e-06,
+      "loss": 3.7111,
+      "step": 3465
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.873610012563564e-06,
+      "loss": 3.6877,
+      "step": 3466
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.871755120243685e-06,
+      "loss": 3.6238,
+      "step": 3467
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.869899928291809e-06,
+      "loss": 3.6644,
+      "step": 3468
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.868044437004918e-06,
+      "loss": 3.7448,
+      "step": 3469
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8661886466800405e-06,
+      "loss": 3.6985,
+      "step": 3470
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.864332557614251e-06,
+      "loss": 3.7607,
+      "step": 3471
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.862476170104677e-06,
+      "loss": 3.7636,
+      "step": 3472
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.860619484448486e-06,
+      "loss": 3.8751,
+      "step": 3473
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.858762500942901e-06,
+      "loss": 3.824,
+      "step": 3474
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.856905219885189e-06,
+      "loss": 3.702,
+      "step": 3475
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8550476415726645e-06,
+      "loss": 3.7356,
+      "step": 3476
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.853189766302688e-06,
+      "loss": 3.5674,
+      "step": 3477
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8513315943726725e-06,
+      "loss": 3.7837,
+      "step": 3478
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8494731260800755e-06,
+      "loss": 3.7119,
+      "step": 3479
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.847614361722399e-06,
+      "loss": 3.7165,
+      "step": 3480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.845755301597198e-06,
+      "loss": 3.7317,
+      "step": 3481
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.843895946002073e-06,
+      "loss": 3.7054,
+      "step": 3482
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.842036295234669e-06,
+      "loss": 3.6656,
+      "step": 3483
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.840176349592681e-06,
+      "loss": 3.6744,
+      "step": 3484
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.83831610937385e-06,
+      "loss": 3.6657,
+      "step": 3485
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.836455574875967e-06,
+      "loss": 3.7562,
+      "step": 3486
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.834594746396866e-06,
+      "loss": 3.5528,
+      "step": 3487
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.83273362423443e-06,
+      "loss": 3.663,
+      "step": 3488
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.830872208686589e-06,
+      "loss": 3.6356,
+      "step": 3489
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.829010500051319e-06,
+      "loss": 3.683,
+      "step": 3490
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.827148498626643e-06,
+      "loss": 3.7821,
+      "step": 3491
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.825286204710635e-06,
+      "loss": 3.7249,
+      "step": 3492
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.82342361860141e-06,
+      "loss": 3.7084,
+      "step": 3493
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.821560740597133e-06,
+      "loss": 3.7145,
+      "step": 3494
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.819697570996015e-06,
+      "loss": 3.7459,
+      "step": 3495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.817834110096313e-06,
+      "loss": 3.6899,
+      "step": 3496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.815970358196331e-06,
+      "loss": 3.7373,
+      "step": 3497
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.814106315594421e-06,
+      "loss": 3.6005,
+      "step": 3498
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.81224198258898e-06,
+      "loss": 3.7526,
+      "step": 3499
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8103773594784506e-06,
+      "loss": 3.6792,
+      "step": 3500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.808512446561324e-06,
+      "loss": 3.75,
+      "step": 3501
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.806647244136138e-06,
+      "loss": 3.7127,
+      "step": 3502
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.804781752501474e-06,
+      "loss": 3.8362,
+      "step": 3503
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8029159719559615e-06,
+      "loss": 3.6433,
+      "step": 3504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.801049902798278e-06,
+      "loss": 3.6479,
+      "step": 3505
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.799183545327143e-06,
+      "loss": 3.7589,
+      "step": 3506
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7973168998413245e-06,
+      "loss": 3.6203,
+      "step": 3507
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.795449966639638e-06,
+      "loss": 3.804,
+      "step": 3508
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.793582746020943e-06,
+      "loss": 3.6987,
+      "step": 3509
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.791715238284146e-06,
+      "loss": 3.8793,
+      "step": 3510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.789847443728198e-06,
+      "loss": 3.688,
+      "step": 3511
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.787979362652097e-06,
+      "loss": 3.6284,
+      "step": 3512
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7861109953548885e-06,
+      "loss": 3.756,
+      "step": 3513
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.784242342135661e-06,
+      "loss": 3.6605,
+      "step": 3514
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.782373403293549e-06,
+      "loss": 3.6548,
+      "step": 3515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.780504179127735e-06,
+      "loss": 3.7638,
+      "step": 3516
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.778634669937446e-06,
+      "loss": 3.6864,
+      "step": 3517
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.776764876021954e-06,
+      "loss": 3.7272,
+      "step": 3518
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.774894797680576e-06,
+      "loss": 3.7233,
+      "step": 3519
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.773024435212678e-06,
+      "loss": 3.6186,
+      "step": 3520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.771153788917668e-06,
+      "loss": 3.6352,
+      "step": 3521
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7692828590949985e-06,
+      "loss": 3.6745,
+      "step": 3522
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.767411646044173e-06,
+      "loss": 3.6894,
+      "step": 3523
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.765540150064734e-06,
+      "loss": 3.6858,
+      "step": 3524
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.763668371456272e-06,
+      "loss": 3.723,
+      "step": 3525
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7617963105184246e-06,
+      "loss": 3.6598,
+      "step": 3526
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.75992396755087e-06,
+      "loss": 3.7186,
+      "step": 3527
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.758051342853337e-06,
+      "loss": 3.7213,
+      "step": 3528
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7561784367255954e-06,
+      "loss": 3.7519,
+      "step": 3529
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7543052494674645e-06,
+      "loss": 3.719,
+      "step": 3530
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.752431781378801e-06,
+      "loss": 3.6413,
+      "step": 3531
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.750558032759512e-06,
+      "loss": 3.7984,
+      "step": 3532
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7486840039095516e-06,
+      "loss": 3.7922,
+      "step": 3533
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.746809695128912e-06,
+      "loss": 3.6711,
+      "step": 3534
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7449351067176365e-06,
+      "loss": 3.6588,
+      "step": 3535
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.743060238975811e-06,
+      "loss": 3.6562,
+      "step": 3536
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.741185092203561e-06,
+      "loss": 3.711,
+      "step": 3537
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.739309666701068e-06,
+      "loss": 3.6536,
+      "step": 3538
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.737433962768549e-06,
+      "loss": 3.7522,
+      "step": 3539
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.735557980706268e-06,
+      "loss": 3.8217,
+      "step": 3540
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7336817208145325e-06,
+      "loss": 3.6937,
+      "step": 3541
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.731805183393696e-06,
+      "loss": 3.7409,
+      "step": 3542
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.729928368744158e-06,
+      "loss": 3.6562,
+      "step": 3543
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.728051277166358e-06,
+      "loss": 3.7344,
+      "step": 3544
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.726173908960785e-06,
+      "loss": 3.6436,
+      "step": 3545
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.724296264427966e-06,
+      "loss": 3.7929,
+      "step": 3546
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7224183438684785e-06,
+      "loss": 3.828,
+      "step": 3547
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.720540147582941e-06,
+      "loss": 3.6733,
+      "step": 3548
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.718661675872015e-06,
+      "loss": 3.6892,
+      "step": 3549
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.71678292903641e-06,
+      "loss": 3.8009,
+      "step": 3550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.714903907376875e-06,
+      "loss": 3.793,
+      "step": 3551
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.713024611194208e-06,
+      "loss": 3.7695,
+      "step": 3552
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.711145040789246e-06,
+      "loss": 3.6917,
+      "step": 3553
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.709265196462872e-06,
+      "loss": 3.7336,
+      "step": 3554
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.707385078516013e-06,
+      "loss": 3.6575,
+      "step": 3555
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.705504687249642e-06,
+      "loss": 3.6495,
+      "step": 3556
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.703624022964771e-06,
+      "loss": 3.7376,
+      "step": 3557
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.70174308596246e-06,
+      "loss": 3.6516,
+      "step": 3558
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.699861876543808e-06,
+      "loss": 3.8047,
+      "step": 3559
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.697980395009963e-06,
+      "loss": 3.6957,
+      "step": 3560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.696098641662112e-06,
+      "loss": 3.6429,
+      "step": 3561
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.69421661680149e-06,
+      "loss": 3.6788,
+      "step": 3562
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.692334320729372e-06,
+      "loss": 3.7384,
+      "step": 3563
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.690451753747076e-06,
+      "loss": 3.7989,
+      "step": 3564
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.688568916155968e-06,
+      "loss": 3.6721,
+      "step": 3565
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.686685808257451e-06,
+      "loss": 3.6058,
+      "step": 3566
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.684802430352974e-06,
+      "loss": 3.6349,
+      "step": 3567
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.682918782744033e-06,
+      "loss": 3.7674,
+      "step": 3568
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.68103486573216e-06,
+      "loss": 3.6662,
+      "step": 3569
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.679150679618935e-06,
+      "loss": 3.6113,
+      "step": 3570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.677266224705982e-06,
+      "loss": 3.7242,
+      "step": 3571
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.675381501294965e-06,
+      "loss": 3.6906,
+      "step": 3572
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.673496509687589e-06,
+      "loss": 3.7245,
+      "step": 3573
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.67161125018561e-06,
+      "loss": 3.5998,
+      "step": 3574
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.669725723090819e-06,
+      "loss": 3.8588,
+      "step": 3575
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.667839928705051e-06,
+      "loss": 3.5567,
+      "step": 3576
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6659538673301885e-06,
+      "loss": 3.6894,
+      "step": 3577
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.664067539268152e-06,
+      "loss": 3.6703,
+      "step": 3578
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.662180944820906e-06,
+      "loss": 3.6357,
+      "step": 3579
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.660294084290461e-06,
+      "loss": 3.7095,
+      "step": 3580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.658406957978862e-06,
+      "loss": 3.7224,
+      "step": 3581
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.656519566188206e-06,
+      "loss": 3.6643,
+      "step": 3582
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.654631909220627e-06,
+      "loss": 3.733,
+      "step": 3583
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6527439873783004e-06,
+      "loss": 3.577,
+      "step": 3584
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.650855800963448e-06,
+      "loss": 3.712,
+      "step": 3585
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.648967350278332e-06,
+      "loss": 3.7323,
+      "step": 3586
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.647078635625259e-06,
+      "loss": 3.7788,
+      "step": 3587
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.645189657306572e-06,
+      "loss": 3.7866,
+      "step": 3588
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.643300415624664e-06,
+      "loss": 3.8395,
+      "step": 3589
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6414109108819625e-06,
+      "loss": 3.6896,
+      "step": 3590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.639521143380944e-06,
+      "loss": 3.7609,
+      "step": 3591
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6376311134241216e-06,
+      "loss": 3.6265,
+      "step": 3592
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.635740821314054e-06,
+      "loss": 3.6623,
+      "step": 3593
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.63385026735334e-06,
+      "loss": 3.6919,
+      "step": 3594
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.631959451844622e-06,
+      "loss": 3.7519,
+      "step": 3595
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.630068375090581e-06,
+      "loss": 3.6627,
+      "step": 3596
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.628177037393945e-06,
+      "loss": 3.5737,
+      "step": 3597
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.626285439057479e-06,
+      "loss": 3.774,
+      "step": 3598
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.624393580383991e-06,
+      "loss": 3.808,
+      "step": 3599
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.622501461676333e-06,
+      "loss": 3.5829,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.620609083237395e-06,
+      "loss": 3.8038,
+      "step": 3601
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.618716445370113e-06,
+      "loss": 3.6758,
+      "step": 3602
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6168235483774576e-06,
+      "loss": 3.679,
+      "step": 3603
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6149303925624495e-06,
+      "loss": 3.6458,
+      "step": 3604
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.613036978228144e-06,
+      "loss": 3.7047,
+      "step": 3605
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.611143305677641e-06,
+      "loss": 3.666,
+      "step": 3606
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.609249375214082e-06,
+      "loss": 3.668,
+      "step": 3607
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.607355187140647e-06,
+      "loss": 3.7641,
+      "step": 3608
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.60546074176056e-06,
+      "loss": 3.6164,
+      "step": 3609
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6035660393770854e-06,
+      "loss": 3.7676,
+      "step": 3610
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.601671080293529e-06,
+      "loss": 3.746,
+      "step": 3611
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.599775864813237e-06,
+      "loss": 3.8062,
+      "step": 3612
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.597880393239596e-06,
+      "loss": 3.6239,
+      "step": 3613
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.595984665876036e-06,
+      "loss": 3.764,
+      "step": 3614
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.594088683026026e-06,
+      "loss": 3.7073,
+      "step": 3615
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.592192444993078e-06,
+      "loss": 3.673,
+      "step": 3616
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.590295952080738e-06,
+      "loss": 3.572,
+      "step": 3617
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.5883992045926024e-06,
+      "loss": 3.6987,
+      "step": 3618
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.586502202832304e-06,
+      "loss": 3.7076,
+      "step": 3619
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.584604947103515e-06,
+      "loss": 3.7113,
+      "step": 3620
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.582707437709949e-06,
+      "loss": 3.8025,
+      "step": 3621
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.580809674955363e-06,
+      "loss": 3.7099,
+      "step": 3622
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.578911659143551e-06,
+      "loss": 3.783,
+      "step": 3623
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.5770133905783485e-06,
+      "loss": 3.7241,
+      "step": 3624
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.575114869563632e-06,
+      "loss": 3.6125,
+      "step": 3625
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.573216096403319e-06,
+      "loss": 3.6182,
+      "step": 3626
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.571317071401364e-06,
+      "loss": 3.826,
+      "step": 3627
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.569417794861768e-06,
+      "loss": 3.6841,
+      "step": 3628
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.567518267088566e-06,
+      "loss": 3.7196,
+      "step": 3629
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.565618488385837e-06,
+      "loss": 3.6472,
+      "step": 3630
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.5637184590576995e-06,
+      "loss": 3.7721,
+      "step": 3631
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.56181817940831e-06,
+      "loss": 3.7247,
+      "step": 3632
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.559917649741869e-06,
+      "loss": 3.7494,
+      "step": 3633
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.558016870362612e-06,
+      "loss": 3.6484,
+      "step": 3634
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.55611584157482e-06,
+      "loss": 3.7086,
+      "step": 3635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.554214563682809e-06,
+      "loss": 3.7441,
+      "step": 3636
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.552313036990938e-06,
+      "loss": 3.6805,
+      "step": 3637
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.550411261803606e-06,
+      "loss": 3.764,
+      "step": 3638
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.548509238425248e-06,
+      "loss": 3.6301,
+      "step": 3639
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.546606967160343e-06,
+      "loss": 3.6867,
+      "step": 3640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.544704448313407e-06,
+      "loss": 3.7253,
+      "step": 3641
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.542801682188997e-06,
+      "loss": 3.646,
+      "step": 3642
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.540898669091711e-06,
+      "loss": 3.7639,
+      "step": 3643
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.538995409326183e-06,
+      "loss": 3.7279,
+      "step": 3644
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.537091903197089e-06,
+      "loss": 3.6663,
+      "step": 3645
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.535188151009143e-06,
+      "loss": 3.7469,
+      "step": 3646
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.5332841530671e-06,
+      "loss": 3.7011,
+      "step": 3647
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.531379909675753e-06,
+      "loss": 3.7157,
+      "step": 3648
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.5294754211399345e-06,
+      "loss": 3.6656,
+      "step": 3649
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.527570687764518e-06,
+      "loss": 3.7799,
+      "step": 3650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.5256657098544115e-06,
+      "loss": 3.6649,
+      "step": 3651
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.5237604877145685e-06,
+      "loss": 3.738,
+      "step": 3652
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.521855021649978e-06,
+      "loss": 3.6555,
+      "step": 3653
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.519949311965666e-06,
+      "loss": 3.7975,
+      "step": 3654
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.518043358966701e-06,
+      "loss": 3.6918,
+      "step": 3655
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.516137162958192e-06,
+      "loss": 3.7826,
+      "step": 3656
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.514230724245281e-06,
+      "loss": 3.7419,
+      "step": 3657
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.512324043133155e-06,
+      "loss": 3.6324,
+      "step": 3658
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.510417119927034e-06,
+      "loss": 3.6803,
+      "step": 3659
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.5085099549321826e-06,
+      "loss": 3.6211,
+      "step": 3660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.506602548453899e-06,
+      "loss": 3.6956,
+      "step": 3661
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.504694900797523e-06,
+      "loss": 3.755,
+      "step": 3662
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.502787012268433e-06,
+      "loss": 3.6345,
+      "step": 3663
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.500878883172045e-06,
+      "loss": 3.7468,
+      "step": 3664
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.498970513813814e-06,
+      "loss": 3.7298,
+      "step": 3665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.49706190449923e-06,
+      "loss": 3.7989,
+      "step": 3666
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4951530555338295e-06,
+      "loss": 3.7005,
+      "step": 3667
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.49324396722318e-06,
+      "loss": 3.8581,
+      "step": 3668
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.49133463987289e-06,
+      "loss": 3.7336,
+      "step": 3669
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.489425073788608e-06,
+      "loss": 3.7439,
+      "step": 3670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.487515269276015e-06,
+      "loss": 3.7396,
+      "step": 3671
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4856052266408375e-06,
+      "loss": 3.743,
+      "step": 3672
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4836949461888345e-06,
+      "loss": 3.5536,
+      "step": 3673
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.481784428225806e-06,
+      "loss": 3.6926,
+      "step": 3674
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.479873673057588e-06,
+      "loss": 3.6384,
+      "step": 3675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4779626809900585e-06,
+      "loss": 3.717,
+      "step": 3676
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.476051452329128e-06,
+      "loss": 3.5967,
+      "step": 3677
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.474139987380748e-06,
+      "loss": 3.7512,
+      "step": 3678
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.472228286450907e-06,
+      "loss": 3.7176,
+      "step": 3679
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.470316349845632e-06,
+      "loss": 3.8267,
+      "step": 3680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.468404177870988e-06,
+      "loss": 3.7137,
+      "step": 3681
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.466491770833077e-06,
+      "loss": 3.6725,
+      "step": 3682
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.464579129038037e-06,
+      "loss": 3.6775,
+      "step": 3683
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.462666252792044e-06,
+      "loss": 3.8296,
+      "step": 3684
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.460753142401318e-06,
+      "loss": 3.6604,
+      "step": 3685
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4588397981721066e-06,
+      "loss": 3.6687,
+      "step": 3686
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.456926220410701e-06,
+      "loss": 3.6281,
+      "step": 3687
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.455012409423427e-06,
+      "loss": 3.725,
+      "step": 3688
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.453098365516651e-06,
+      "loss": 3.6913,
+      "step": 3689
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4511840889967715e-06,
+      "loss": 3.7276,
+      "step": 3690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.449269580170231e-06,
+      "loss": 3.7413,
+      "step": 3691
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.447354839343504e-06,
+      "loss": 3.6254,
+      "step": 3692
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.445439866823102e-06,
+      "loss": 3.7205,
+      "step": 3693
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.443524662915579e-06,
+      "loss": 3.6268,
+      "step": 3694
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.441609227927519e-06,
+      "loss": 3.7356,
+      "step": 3695
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.439693562165546e-06,
+      "loss": 3.7334,
+      "step": 3696
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.437777665936325e-06,
+      "loss": 3.6523,
+      "step": 3697
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4358615395465506e-06,
+      "loss": 3.7367,
+      "step": 3698
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.433945183302959e-06,
+      "loss": 3.7237,
+      "step": 3699
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.432028597512322e-06,
+      "loss": 3.6975,
+      "step": 3700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.430111782481449e-06,
+      "loss": 3.8091,
+      "step": 3701
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.428194738517183e-06,
+      "loss": 3.6519,
+      "step": 3702
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.426277465926408e-06,
+      "loss": 3.7556,
+      "step": 3703
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.424359965016041e-06,
+      "loss": 3.6974,
+      "step": 3704
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.422442236093038e-06,
+      "loss": 3.5542,
+      "step": 3705
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.42052427946439e-06,
+      "loss": 3.7396,
+      "step": 3706
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.418606095437125e-06,
+      "loss": 3.6031,
+      "step": 3707
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.416687684318309e-06,
+      "loss": 3.8136,
+      "step": 3708
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.414769046415039e-06,
+      "loss": 3.6649,
+      "step": 3709
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4128501820344556e-06,
+      "loss": 3.7847,
+      "step": 3710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.41093109148373e-06,
+      "loss": 3.5664,
+      "step": 3711
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.409011775070073e-06,
+      "loss": 3.6976,
+      "step": 3712
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.407092233100729e-06,
+      "loss": 3.8125,
+      "step": 3713
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4051724658829815e-06,
+      "loss": 3.6809,
+      "step": 3714
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.403252473724146e-06,
+      "loss": 3.7574,
+      "step": 3715
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.40133225693158e-06,
+      "loss": 3.6718,
+      "step": 3716
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.399411815812668e-06,
+      "loss": 3.6849,
+      "step": 3717
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.397491150674841e-06,
+      "loss": 3.7876,
+      "step": 3718
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.395570261825556e-06,
+      "loss": 3.6306,
+      "step": 3719
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.393649149572313e-06,
+      "loss": 3.6503,
+      "step": 3720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.391727814222645e-06,
+      "loss": 3.6915,
+      "step": 3721
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.38980625608412e-06,
+      "loss": 3.6727,
+      "step": 3722
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.387884475464342e-06,
+      "loss": 3.691,
+      "step": 3723
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.385962472670953e-06,
+      "loss": 3.6275,
+      "step": 3724
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.384040248011628e-06,
+      "loss": 3.7549,
+      "step": 3725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.382117801794078e-06,
+      "loss": 3.647,
+      "step": 3726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.380195134326049e-06,
+      "loss": 3.7386,
+      "step": 3727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3782722459153246e-06,
+      "loss": 3.7524,
+      "step": 3728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3763491368697205e-06,
+      "loss": 3.7764,
+      "step": 3729
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.37442580749709e-06,
+      "loss": 3.7745,
+      "step": 3730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.372502258105322e-06,
+      "loss": 3.6957,
+      "step": 3731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.370578489002339e-06,
+      "loss": 3.8293,
+      "step": 3732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3686545004961e-06,
+      "loss": 3.6797,
+      "step": 3733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.366730292894597e-06,
+      "loss": 3.6572,
+      "step": 3734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3648058665058594e-06,
+      "loss": 3.698,
+      "step": 3735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.362881221637953e-06,
+      "loss": 3.7753,
+      "step": 3736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.360956358598974e-06,
+      "loss": 3.5675,
+      "step": 3737
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.359031277697057e-06,
+      "loss": 3.6242,
+      "step": 3738
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.35710597924037e-06,
+      "loss": 3.6971,
+      "step": 3739
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.355180463537116e-06,
+      "loss": 3.6399,
+      "step": 3740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.353254730895533e-06,
+      "loss": 3.6672,
+      "step": 3741
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.351328781623895e-06,
+      "loss": 3.7436,
+      "step": 3742
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3494026160305075e-06,
+      "loss": 3.656,
+      "step": 3743
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.347476234423715e-06,
+      "loss": 3.7569,
+      "step": 3744
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.345549637111892e-06,
+      "loss": 3.6836,
+      "step": 3745
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.343622824403451e-06,
+      "loss": 3.7002,
+      "step": 3746
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.341695796606835e-06,
+      "loss": 3.7639,
+      "step": 3747
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.339768554030528e-06,
+      "loss": 3.5775,
+      "step": 3748
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3378410969830416e-06,
+      "loss": 3.6944,
+      "step": 3749
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.335913425772926e-06,
+      "loss": 3.7449,
+      "step": 3750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3339855407087645e-06,
+      "loss": 3.6903,
+      "step": 3751
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3320574420991735e-06,
+      "loss": 3.721,
+      "step": 3752
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.330129130252805e-06,
+      "loss": 3.6974,
+      "step": 3753
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.328200605478345e-06,
+      "loss": 3.6731,
+      "step": 3754
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.326271868084512e-06,
+      "loss": 3.6927,
+      "step": 3755
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.324342918380061e-06,
+      "loss": 3.7086,
+      "step": 3756
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3224137566737784e-06,
+      "loss": 3.6846,
+      "step": 3757
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3204843832744875e-06,
+      "loss": 3.7193,
+      "step": 3758
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.318554798491041e-06,
+      "loss": 3.6508,
+      "step": 3759
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.316625002632333e-06,
+      "loss": 3.8575,
+      "step": 3760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3146949960072836e-06,
+      "loss": 3.6145,
+      "step": 3761
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3127647789248506e-06,
+      "loss": 3.6178,
+      "step": 3762
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.310834351694023e-06,
+      "loss": 3.7988,
+      "step": 3763
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3089037146238275e-06,
+      "loss": 3.6194,
+      "step": 3764
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.30697286802332e-06,
+      "loss": 3.857,
+      "step": 3765
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.305041812201593e-06,
+      "loss": 3.7097,
+      "step": 3766
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.303110547467772e-06,
+      "loss": 3.6304,
+      "step": 3767
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.301179074131014e-06,
+      "loss": 3.7305,
+      "step": 3768
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2992473925005115e-06,
+      "loss": 3.6459,
+      "step": 3769
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.297315502885489e-06,
+      "loss": 3.8196,
+      "step": 3770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.295383405595204e-06,
+      "loss": 3.6211,
+      "step": 3771
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.29345110093895e-06,
+      "loss": 3.7244,
+      "step": 3772
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.291518589226051e-06,
+      "loss": 3.8551,
+      "step": 3773
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.289585870765865e-06,
+      "loss": 3.6839,
+      "step": 3774
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.287652945867783e-06,
+      "loss": 3.7677,
+      "step": 3775
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.28571981484123e-06,
+      "loss": 3.6045,
+      "step": 3776
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2837864779956625e-06,
+      "loss": 3.7665,
+      "step": 3777
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.28185293564057e-06,
+      "loss": 3.7565,
+      "step": 3778
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2799191880854745e-06,
+      "loss": 3.6453,
+      "step": 3779
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.277985235639935e-06,
+      "loss": 3.6914,
+      "step": 3780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.276051078613538e-06,
+      "loss": 3.5782,
+      "step": 3781
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.274116717315904e-06,
+      "loss": 3.671,
+      "step": 3782
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.272182152056687e-06,
+      "loss": 3.6711,
+      "step": 3783
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2702473831455755e-06,
+      "loss": 3.7226,
+      "step": 3784
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.26831241089229e-06,
+      "loss": 3.7027,
+      "step": 3785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.26637723560658e-06,
+      "loss": 3.63,
+      "step": 3786
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.264441857598229e-06,
+      "loss": 3.7237,
+      "step": 3787
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.262506277177055e-06,
+      "loss": 3.7897,
+      "step": 3788
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.260570494652908e-06,
+      "loss": 3.6796,
+      "step": 3789
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.258634510335668e-06,
+      "loss": 3.7858,
+      "step": 3790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2566983245352495e-06,
+      "loss": 3.6105,
+      "step": 3791
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.254761937561599e-06,
+      "loss": 3.6646,
+      "step": 3792
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2528253497246935e-06,
+      "loss": 3.6325,
+      "step": 3793
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.250888561334547e-06,
+      "loss": 3.6268,
+      "step": 3794
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.248951572701198e-06,
+      "loss": 3.666,
+      "step": 3795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.247014384134723e-06,
+      "loss": 3.7752,
+      "step": 3796
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.245076995945228e-06,
+      "loss": 3.7849,
+      "step": 3797
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.243139408442853e-06,
+      "loss": 3.6972,
+      "step": 3798
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.241201621937767e-06,
+      "loss": 3.6878,
+      "step": 3799
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2392636367401725e-06,
+      "loss": 3.811,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.237325453160303e-06,
+      "loss": 3.6199,
+      "step": 3801
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.235387071508427e-06,
+      "loss": 3.7584,
+      "step": 3802
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.23344849209484e-06,
+      "loss": 3.784,
+      "step": 3803
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.2315097152298705e-06,
+      "loss": 3.6461,
+      "step": 3804
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.229570741223882e-06,
+      "loss": 3.8595,
+      "step": 3805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.227631570387265e-06,
+      "loss": 3.6433,
+      "step": 3806
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.225692203030444e-06,
+      "loss": 3.7085,
+      "step": 3807
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.223752639463876e-06,
+      "loss": 3.6797,
+      "step": 3808
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.221812879998045e-06,
+      "loss": 3.711,
+      "step": 3809
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.219872924943471e-06,
+      "loss": 3.7098,
+      "step": 3810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.217932774610702e-06,
+      "loss": 3.7204,
+      "step": 3811
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.2159924293103205e-06,
+      "loss": 3.7824,
+      "step": 3812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.214051889352936e-06,
+      "loss": 3.6153,
+      "step": 3813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.212111155049197e-06,
+      "loss": 3.6499,
+      "step": 3814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.210170226709771e-06,
+      "loss": 3.7434,
+      "step": 3815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.208229104645366e-06,
+      "loss": 3.7965,
+      "step": 3816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.2062877891667205e-06,
+      "loss": 3.7459,
+      "step": 3817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.204346280584599e-06,
+      "loss": 3.7681,
+      "step": 3818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.202404579209801e-06,
+      "loss": 3.7637,
+      "step": 3819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.2004626853531545e-06,
+      "loss": 3.7793,
+      "step": 3820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1985205993255214e-06,
+      "loss": 3.5985,
+      "step": 3821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1965783214377895e-06,
+      "loss": 3.625,
+      "step": 3822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.194635852000882e-06,
+      "loss": 3.7109,
+      "step": 3823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1926931913257515e-06,
+      "loss": 3.6842,
+      "step": 3824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1907503397233785e-06,
+      "loss": 3.5969,
+      "step": 3825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.188807297504778e-06,
+      "loss": 3.8333,
+      "step": 3826
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.186864064980992e-06,
+      "loss": 3.6232,
+      "step": 3827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.184920642463095e-06,
+      "loss": 3.639,
+      "step": 3828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.182977030262193e-06,
+      "loss": 3.6509,
+      "step": 3829
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.18103322868942e-06,
+      "loss": 3.7367,
+      "step": 3830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.179089238055942e-06,
+      "loss": 3.7953,
+      "step": 3831
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.177145058672954e-06,
+      "loss": 3.6613,
+      "step": 3832
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.175200690851682e-06,
+      "loss": 3.5828,
+      "step": 3833
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1732561349033794e-06,
+      "loss": 3.6719,
+      "step": 3834
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.171311391139336e-06,
+      "loss": 3.5912,
+      "step": 3835
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.169366459870866e-06,
+      "loss": 3.6962,
+      "step": 3836
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.167421341409314e-06,
+      "loss": 3.6656,
+      "step": 3837
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1654760360660605e-06,
+      "loss": 3.7507,
+      "step": 3838
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.163530544152506e-06,
+      "loss": 3.5973,
+      "step": 3839
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.16158486598009e-06,
+      "loss": 3.6731,
+      "step": 3840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.159639001860277e-06,
+      "loss": 3.6629,
+      "step": 3841
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.157692952104564e-06,
+      "loss": 3.7681,
+      "step": 3842
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1557467170244735e-06,
+      "loss": 3.754,
+      "step": 3843
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.153800296931561e-06,
+      "loss": 3.7352,
+      "step": 3844
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.151853692137413e-06,
+      "loss": 3.5703,
+      "step": 3845
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.149906902953641e-06,
+      "loss": 3.725,
+      "step": 3846
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1479599296918905e-06,
+      "loss": 3.7436,
+      "step": 3847
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.146012772663832e-06,
+      "loss": 3.6909,
+      "step": 3848
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.14406543218117e-06,
+      "loss": 3.7393,
+      "step": 3849
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.142117908555636e-06,
+      "loss": 3.732,
+      "step": 3850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1401702020989896e-06,
+      "loss": 3.6724,
+      "step": 3851
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.138222313123022e-06,
+      "loss": 3.6928,
+      "step": 3852
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1362742419395536e-06,
+      "loss": 3.6945,
+      "step": 3853
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1343259888604335e-06,
+      "loss": 3.697,
+      "step": 3854
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.132377554197537e-06,
+      "loss": 3.6791,
+      "step": 3855
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.130428938262774e-06,
+      "loss": 3.7083,
+      "step": 3856
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.128480141368079e-06,
+      "loss": 3.6132,
+      "step": 3857
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.126531163825415e-06,
+      "loss": 3.6851,
+      "step": 3858
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.12458200594678e-06,
+      "loss": 3.8025,
+      "step": 3859
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.122632668044193e-06,
+      "loss": 3.7002,
+      "step": 3860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.120683150429707e-06,
+      "loss": 3.6984,
+      "step": 3861
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.118733453415403e-06,
+      "loss": 3.7837,
+      "step": 3862
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.116783577313389e-06,
+      "loss": 3.8129,
+      "step": 3863
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1148335224358016e-06,
+      "loss": 3.8042,
+      "step": 3864
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.112883289094809e-06,
+      "loss": 3.7531,
+      "step": 3865
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1109328776026054e-06,
+      "loss": 3.7047,
+      "step": 3866
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.108982288271414e-06,
+      "loss": 3.7618,
+      "step": 3867
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.107031521413488e-06,
+      "loss": 3.7772,
+      "step": 3868
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.105080577341105e-06,
+      "loss": 3.6567,
+      "step": 3869
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1031294563665766e-06,
+      "loss": 3.7734,
+      "step": 3870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.101178158802237e-06,
+      "loss": 3.7727,
+      "step": 3871
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.099226684960454e-06,
+      "loss": 3.6328,
+      "step": 3872
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.097275035153618e-06,
+      "loss": 3.6688,
+      "step": 3873
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.0953232096941525e-06,
+      "loss": 3.7043,
+      "step": 3874
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.093371208894507e-06,
+      "loss": 3.6284,
+      "step": 3875
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.091419033067159e-06,
+      "loss": 3.6313,
+      "step": 3876
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.0894666825246166e-06,
+      "loss": 3.6854,
+      "step": 3877
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.08751415757941e-06,
+      "loss": 3.6484,
+      "step": 3878
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.085561458544102e-06,
+      "loss": 3.7014,
+      "step": 3879
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.083608585731283e-06,
+      "loss": 3.7266,
+      "step": 3880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.08165553945357e-06,
+      "loss": 3.7916,
+      "step": 3881
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.079702320023608e-06,
+      "loss": 3.8293,
+      "step": 3882
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.0777489277540715e-06,
+      "loss": 3.5638,
+      "step": 3883
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.075795362957657e-06,
+      "loss": 3.7039,
+      "step": 3884
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.073841625947097e-06,
+      "loss": 3.8609,
+      "step": 3885
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.071887717035143e-06,
+      "loss": 3.6675,
+      "step": 3886
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0699336365345825e-06,
+      "loss": 3.7409,
+      "step": 3887
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.067979384758225e-06,
+      "loss": 3.7046,
+      "step": 3888
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.066024962018907e-06,
+      "loss": 3.6994,
+      "step": 3889
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.064070368629496e-06,
+      "loss": 3.6736,
+      "step": 3890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0621156049028814e-06,
+      "loss": 3.7665,
+      "step": 3891
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0601606711519875e-06,
+      "loss": 3.7105,
+      "step": 3892
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.058205567689759e-06,
+      "loss": 3.6841,
+      "step": 3893
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.056250294829171e-06,
+      "loss": 3.6964,
+      "step": 3894
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.054294852883227e-06,
+      "loss": 3.7158,
+      "step": 3895
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0523392421649515e-06,
+      "loss": 3.6983,
+      "step": 3896
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.050383462987403e-06,
+      "loss": 3.687,
+      "step": 3897
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.048427515663664e-06,
+      "loss": 3.6399,
+      "step": 3898
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.046471400506843e-06,
+      "loss": 3.722,
+      "step": 3899
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.044515117830078e-06,
+      "loss": 3.6801,
+      "step": 3900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.042558667946531e-06,
+      "loss": 3.6317,
+      "step": 3901
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.040602051169393e-06,
+      "loss": 3.7451,
+      "step": 3902
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.038645267811879e-06,
+      "loss": 3.5938,
+      "step": 3903
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.036688318187234e-06,
+      "loss": 3.7468,
+      "step": 3904
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0347312026087265e-06,
+      "loss": 3.6487,
+      "step": 3905
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.032773921389655e-06,
+      "loss": 3.6579,
+      "step": 3906
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.030816474843342e-06,
+      "loss": 3.6506,
+      "step": 3907
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.028858863283135e-06,
+      "loss": 3.6221,
+      "step": 3908
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.026901087022411e-06,
+      "loss": 3.7477,
+      "step": 3909
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.024943146374574e-06,
+      "loss": 3.7567,
+      "step": 3910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.022985041653052e-06,
+      "loss": 3.7408,
+      "step": 3911
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.021026773171299e-06,
+      "loss": 3.6715,
+      "step": 3912
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.019068341242798e-06,
+      "loss": 3.6312,
+      "step": 3913
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.017109746181052e-06,
+      "loss": 3.7498,
+      "step": 3914
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0151509882996e-06,
+      "loss": 3.8014,
+      "step": 3915
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.013192067911997e-06,
+      "loss": 3.6529,
+      "step": 3916
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.011232985331831e-06,
+      "loss": 3.7359,
+      "step": 3917
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.009273740872712e-06,
+      "loss": 3.7843,
+      "step": 3918
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.007314334848279e-06,
+      "loss": 3.579,
+      "step": 3919
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.005354767572194e-06,
+      "loss": 3.6859,
+      "step": 3920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0033950393581445e-06,
+      "loss": 3.6664,
+      "step": 3921
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0014351505198475e-06,
+      "loss": 3.721,
+      "step": 3922
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.999475101371044e-06,
+      "loss": 3.7408,
+      "step": 3923
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.997514892225499e-06,
+      "loss": 3.7357,
+      "step": 3924
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9955545233970045e-06,
+      "loss": 3.7419,
+      "step": 3925
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.993593995199377e-06,
+      "loss": 3.7712,
+      "step": 3926
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.991633307946462e-06,
+      "loss": 3.6963,
+      "step": 3927
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9896724619521256e-06,
+      "loss": 3.6938,
+      "step": 3928
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.987711457530261e-06,
+      "loss": 3.7193,
+      "step": 3929
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9857502949947875e-06,
+      "loss": 3.5801,
+      "step": 3930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9837889746596525e-06,
+      "loss": 3.6659,
+      "step": 3931
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9818274968388225e-06,
+      "loss": 3.608,
+      "step": 3932
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.979865861846292e-06,
+      "loss": 3.6195,
+      "step": 3933
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.977904069996083e-06,
+      "loss": 3.7098,
+      "step": 3934
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.975942121602241e-06,
+      "loss": 3.7982,
+      "step": 3935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.973980016978834e-06,
+      "loss": 3.677,
+      "step": 3936
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.972017756439959e-06,
+      "loss": 3.6758,
+      "step": 3937
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.970055340299735e-06,
+      "loss": 3.7484,
+      "step": 3938
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.968092768872309e-06,
+      "loss": 3.6321,
+      "step": 3939
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.966130042471848e-06,
+      "loss": 3.8344,
+      "step": 3940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.964167161412549e-06,
+      "loss": 3.7126,
+      "step": 3941
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9622041260086315e-06,
+      "loss": 3.6906,
+      "step": 3942
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.960240936574339e-06,
+      "loss": 3.761,
+      "step": 3943
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.95827759342394e-06,
+      "loss": 3.7768,
+      "step": 3944
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.956314096871727e-06,
+      "loss": 3.7698,
+      "step": 3945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9543504472320206e-06,
+      "loss": 3.7312,
+      "step": 3946
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.952386644819162e-06,
+      "loss": 3.7588,
+      "step": 3947
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.950422689947519e-06,
+      "loss": 3.7885,
+      "step": 3948
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.948458582931482e-06,
+      "loss": 3.638,
+      "step": 3949
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.946494324085468e-06,
+      "loss": 3.6326,
+      "step": 3950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.944529913723915e-06,
+      "loss": 3.6433,
+      "step": 3951
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.942565352161289e-06,
+      "loss": 3.6569,
+      "step": 3952
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9406006397120775e-06,
+      "loss": 3.6675,
+      "step": 3953
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.938635776690794e-06,
+      "loss": 3.6228,
+      "step": 3954
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.936670763411976e-06,
+      "loss": 3.6417,
+      "step": 3955
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.934705600190183e-06,
+      "loss": 3.779,
+      "step": 3956
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.932740287339998e-06,
+      "loss": 3.7505,
+      "step": 3957
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.930774825176034e-06,
+      "loss": 3.6955,
+      "step": 3958
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.928809214012922e-06,
+      "loss": 3.7214,
+      "step": 3959
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.926843454165317e-06,
+      "loss": 3.6494,
+      "step": 3960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.924877545947901e-06,
+      "loss": 3.7121,
+      "step": 3961
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9229114896753785e-06,
+      "loss": 3.7493,
+      "step": 3962
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.920945285662475e-06,
+      "loss": 3.7294,
+      "step": 3963
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.918978934223945e-06,
+      "loss": 3.8244,
+      "step": 3964
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.917012435674561e-06,
+      "loss": 3.6891,
+      "step": 3965
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.915045790329121e-06,
+      "loss": 3.7348,
+      "step": 3966
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.913078998502449e-06,
+      "loss": 3.7372,
+      "step": 3967
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.911112060509392e-06,
+      "loss": 3.7163,
+      "step": 3968
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.909144976664817e-06,
+      "loss": 3.8307,
+      "step": 3969
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.907177747283614e-06,
+      "loss": 3.6695,
+      "step": 3970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.905210372680704e-06,
+      "loss": 3.5952,
+      "step": 3971
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.903242853171022e-06,
+      "loss": 3.6695,
+      "step": 3972
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.90127518906953e-06,
+      "loss": 3.7634,
+      "step": 3973
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.899307380691214e-06,
+      "loss": 3.6986,
+      "step": 3974
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.897339428351083e-06,
+      "loss": 3.6481,
+      "step": 3975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.895371332364167e-06,
+      "loss": 3.6508,
+      "step": 3976
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.893403093045522e-06,
+      "loss": 3.7343,
+      "step": 3977
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8914347107102235e-06,
+      "loss": 3.7144,
+      "step": 3978
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8894661856733714e-06,
+      "loss": 3.666,
+      "step": 3979
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.88749751825009e-06,
+      "loss": 3.6517,
+      "step": 3980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.885528708755526e-06,
+      "loss": 3.6432,
+      "step": 3981
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.883559757504846e-06,
+      "loss": 3.7078,
+      "step": 3982
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.881590664813243e-06,
+      "loss": 3.6878,
+      "step": 3983
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.879621430995927e-06,
+      "loss": 3.6967,
+      "step": 3984
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.87765205636814e-06,
+      "loss": 3.6609,
+      "step": 3985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.875682541245138e-06,
+      "loss": 3.7535,
+      "step": 3986
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.873712885942202e-06,
+      "loss": 3.6768,
+      "step": 3987
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8717430907746365e-06,
+      "loss": 3.6491,
+      "step": 3988
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.869773156057767e-06,
+      "loss": 3.6733,
+      "step": 3989
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8678030821069455e-06,
+      "loss": 3.6866,
+      "step": 3990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.865832869237539e-06,
+      "loss": 3.6808,
+      "step": 3991
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.863862517764942e-06,
+      "loss": 3.6855,
+      "step": 3992
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.861892028004571e-06,
+      "loss": 3.6439,
+      "step": 3993
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.859921400271863e-06,
+      "loss": 3.7716,
+      "step": 3994
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.857950634882277e-06,
+      "loss": 3.5935,
+      "step": 3995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8559797321512946e-06,
+      "loss": 3.7299,
+      "step": 3996
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.854008692394421e-06,
+      "loss": 3.7527,
+      "step": 3997
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.852037515927179e-06,
+      "loss": 3.7411,
+      "step": 3998
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.850066203065118e-06,
+      "loss": 3.7316,
+      "step": 3999
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.848094754123808e-06,
+      "loss": 3.5845,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.846123169418837e-06,
+      "loss": 3.6617,
+      "step": 4001
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.84415144926582e-06,
+      "loss": 3.7236,
+      "step": 4002
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.84217959398039e-06,
+      "loss": 3.6676,
+      "step": 4003
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8402076038782065e-06,
+      "loss": 3.7235,
+      "step": 4004
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.838235479274944e-06,
+      "loss": 3.6979,
+      "step": 4005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.836263220486303e-06,
+      "loss": 3.6757,
+      "step": 4006
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.834290827828004e-06,
+      "loss": 3.7218,
+      "step": 4007
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.832318301615789e-06,
+      "loss": 3.6229,
+      "step": 4008
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.830345642165421e-06,
+      "loss": 3.631,
+      "step": 4009
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8283728497926865e-06,
+      "loss": 3.7309,
+      "step": 4010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.82639992481339e-06,
+      "loss": 3.741,
+      "step": 4011
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.824426867543358e-06,
+      "loss": 3.6265,
+      "step": 4012
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.822453678298441e-06,
+      "loss": 3.7477,
+      "step": 4013
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.820480357394509e-06,
+      "loss": 3.6715,
+      "step": 4014
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.81850690514745e-06,
+      "loss": 3.8103,
+      "step": 4015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.816533321873178e-06,
+      "loss": 3.8507,
+      "step": 4016
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.814559607887625e-06,
+      "loss": 3.7837,
+      "step": 4017
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.812585763506746e-06,
+      "loss": 3.6063,
+      "step": 4018
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.810611789046514e-06,
+      "loss": 3.6053,
+      "step": 4019
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.808637684822924e-06,
+      "loss": 3.7338,
+      "step": 4020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.806663451151994e-06,
+      "loss": 3.6023,
+      "step": 4021
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.804689088349758e-06,
+      "loss": 3.7726,
+      "step": 4022
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8027145967322764e-06,
+      "loss": 3.6921,
+      "step": 4023
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.800739976615626e-06,
+      "loss": 3.6574,
+      "step": 4024
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.798765228315904e-06,
+      "loss": 3.6528,
+      "step": 4025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.796790352149233e-06,
+      "loss": 3.7831,
+      "step": 4026
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.794815348431751e-06,
+      "loss": 3.6588,
+      "step": 4027
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.792840217479616e-06,
+      "loss": 3.6047,
+      "step": 4028
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.790864959609013e-06,
+      "loss": 3.6751,
+      "step": 4029
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.788889575136138e-06,
+      "loss": 3.6812,
+      "step": 4030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7869140643772174e-06,
+      "loss": 3.7986,
+      "step": 4031
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.784938427648488e-06,
+      "loss": 3.7096,
+      "step": 4032
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.782962665266212e-06,
+      "loss": 3.7375,
+      "step": 4033
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7809867775466725e-06,
+      "loss": 3.8208,
+      "step": 4034
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.779010764806169e-06,
+      "loss": 3.7556,
+      "step": 4035
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.777034627361025e-06,
+      "loss": 3.771,
+      "step": 4036
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.775058365527582e-06,
+      "loss": 3.8248,
+      "step": 4037
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.773081979622201e-06,
+      "loss": 3.7362,
+      "step": 4038
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.771105469961262e-06,
+      "loss": 3.7058,
+      "step": 4039
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.769128836861169e-06,
+      "loss": 3.7214,
+      "step": 4040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.76715208063834e-06,
+      "loss": 3.7235,
+      "step": 4041
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.765175201609221e-06,
+      "loss": 3.5725,
+      "step": 4042
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.763198200090265e-06,
+      "loss": 3.7989,
+      "step": 4043
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.761221076397957e-06,
+      "loss": 3.7715,
+      "step": 4044
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.759243830848794e-06,
+      "loss": 3.7509,
+      "step": 4045
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.757266463759296e-06,
+      "loss": 3.7818,
+      "step": 4046
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.755288975446002e-06,
+      "loss": 3.7761,
+      "step": 4047
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.753311366225467e-06,
+      "loss": 3.6569,
+      "step": 4048
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.751333636414273e-06,
+      "loss": 3.7923,
+      "step": 4049
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.749355786329012e-06,
+      "loss": 3.6777,
+      "step": 4050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7473778162863024e-06,
+      "loss": 3.6643,
+      "step": 4051
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7453997266027784e-06,
+      "loss": 3.6211,
+      "step": 4052
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.743421517595094e-06,
+      "loss": 3.7809,
+      "step": 4053
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7414431895799224e-06,
+      "loss": 3.8135,
+      "step": 4054
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.739464742873957e-06,
+      "loss": 3.63,
+      "step": 4055
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.737486177793907e-06,
+      "loss": 3.6947,
+      "step": 4056
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.735507494656504e-06,
+      "loss": 3.7829,
+      "step": 4057
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.733528693778496e-06,
+      "loss": 3.7027,
+      "step": 4058
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7315497754766525e-06,
+      "loss": 3.7704,
+      "step": 4059
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.729570740067758e-06,
+      "loss": 3.7274,
+      "step": 4060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.727591587868621e-06,
+      "loss": 3.7752,
+      "step": 4061
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.725612319196065e-06,
+      "loss": 3.6937,
+      "step": 4062
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.72363293436693e-06,
+      "loss": 3.603,
+      "step": 4063
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.721653433698082e-06,
+      "loss": 3.6662,
+      "step": 4064
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.719673817506397e-06,
+      "loss": 3.7208,
+      "step": 4065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7176940861087745e-06,
+      "loss": 3.7371,
+      "step": 4066
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.715714239822133e-06,
+      "loss": 3.6764,
+      "step": 4067
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.713734278963407e-06,
+      "loss": 3.6704,
+      "step": 4068
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.711754203849549e-06,
+      "loss": 3.7745,
+      "step": 4069
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.709774014797531e-06,
+      "loss": 3.6974,
+      "step": 4070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.707793712124345e-06,
+      "loss": 3.6905,
+      "step": 4071
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.705813296146998e-06,
+      "loss": 3.6874,
+      "step": 4072
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.703832767182516e-06,
+      "loss": 3.642,
+      "step": 4073
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.701852125547947e-06,
+      "loss": 3.7294,
+      "step": 4074
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.699871371560349e-06,
+      "loss": 3.7708,
+      "step": 4075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.697890505536805e-06,
+      "loss": 3.5606,
+      "step": 4076
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.695909527794414e-06,
+      "loss": 3.6773,
+      "step": 4077
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.693928438650292e-06,
+      "loss": 3.7291,
+      "step": 4078
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.691947238421573e-06,
+      "loss": 3.6658,
+      "step": 4079
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.689965927425407e-06,
+      "loss": 3.6483,
+      "step": 4080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.687984505978967e-06,
+      "loss": 3.8368,
+      "step": 4081
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.686002974399441e-06,
+      "loss": 3.7093,
+      "step": 4082
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6840213330040305e-06,
+      "loss": 3.6507,
+      "step": 4083
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.682039582109959e-06,
+      "loss": 3.7252,
+      "step": 4084
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.680057722034469e-06,
+      "loss": 3.755,
+      "step": 4085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.678075753094817e-06,
+      "loss": 3.8052,
+      "step": 4086
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.676093675608279e-06,
+      "loss": 3.7089,
+      "step": 4087
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.674111489892144e-06,
+      "loss": 3.6788,
+      "step": 4088
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.672129196263727e-06,
+      "loss": 3.7359,
+      "step": 4089
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.670146795040352e-06,
+      "loss": 3.6633,
+      "step": 4090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.668164286539363e-06,
+      "loss": 3.9248,
+      "step": 4091
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.666181671078123e-06,
+      "loss": 3.8652,
+      "step": 4092
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.664198948974009e-06,
+      "loss": 3.7496,
+      "step": 4093
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.662216120544417e-06,
+      "loss": 3.742,
+      "step": 4094
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.660233186106761e-06,
+      "loss": 3.6715,
+      "step": 4095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.658250145978469e-06,
+      "loss": 3.7909,
+      "step": 4096
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.656267000476989e-06,
+      "loss": 3.6724,
+      "step": 4097
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.654283749919784e-06,
+      "loss": 3.6942,
+      "step": 4098
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.652300394624335e-06,
+      "loss": 3.8413,
+      "step": 4099
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.650316934908138e-06,
+      "loss": 3.6383,
+      "step": 4100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.648333371088705e-06,
+      "loss": 3.7272,
+      "step": 4101
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.64634970348357e-06,
+      "loss": 3.7859,
+      "step": 4102
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.644365932410278e-06,
+      "loss": 3.7541,
+      "step": 4103
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.642382058186394e-06,
+      "loss": 3.6536,
+      "step": 4104
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6403980811294965e-06,
+      "loss": 3.699,
+      "step": 4105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.638414001557183e-06,
+      "loss": 3.5847,
+      "step": 4106
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.636429819787066e-06,
+      "loss": 3.6636,
+      "step": 4107
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.634445536136774e-06,
+      "loss": 3.6918,
+      "step": 4108
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6324611509239535e-06,
+      "loss": 3.6206,
+      "step": 4109
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.630476664466268e-06,
+      "loss": 3.766,
+      "step": 4110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6284920770813934e-06,
+      "loss": 3.71,
+      "step": 4111
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.626507389087026e-06,
+      "loss": 3.7081,
+      "step": 4112
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.624522600800874e-06,
+      "loss": 3.7257,
+      "step": 4113
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.622537712540664e-06,
+      "loss": 3.7372,
+      "step": 4114
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.62055272462414e-06,
+      "loss": 3.7345,
+      "step": 4115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.61856763736906e-06,
+      "loss": 3.7586,
+      "step": 4116
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.616582451093198e-06,
+      "loss": 3.7676,
+      "step": 4117
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.614597166114343e-06,
+      "loss": 3.6823,
+      "step": 4118
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.612611782750303e-06,
+      "loss": 3.7085,
+      "step": 4119
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.610626301318897e-06,
+      "loss": 3.729,
+      "step": 4120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.608640722137965e-06,
+      "loss": 3.6973,
+      "step": 4121
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.606655045525359e-06,
+      "loss": 3.7112,
+      "step": 4122
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6046692717989485e-06,
+      "loss": 3.7725,
+      "step": 4123
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6026834012766155e-06,
+      "loss": 3.6459,
+      "step": 4124
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.600697434276262e-06,
+      "loss": 3.7322,
+      "step": 4125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.598711371115802e-06,
+      "loss": 3.8093,
+      "step": 4126
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.596725212113167e-06,
+      "loss": 3.7232,
+      "step": 4127
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.594738957586302e-06,
+      "loss": 3.6951,
+      "step": 4128
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.592752607853167e-06,
+      "loss": 3.8012,
+      "step": 4129
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.59076616323174e-06,
+      "loss": 3.7518,
+      "step": 4130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.5887796240400125e-06,
+      "loss": 3.7489,
+      "step": 4131
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.586792990595992e-06,
+      "loss": 3.7181,
+      "step": 4132
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.584806263217697e-06,
+      "loss": 3.791,
+      "step": 4133
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.582819442223169e-06,
+      "loss": 3.6959,
+      "step": 4134
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5808325279304555e-06,
+      "loss": 3.6199,
+      "step": 4135
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.578845520657625e-06,
+      "loss": 3.7665,
+      "step": 4136
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.576858420722761e-06,
+      "loss": 3.6589,
+      "step": 4137
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.574871228443956e-06,
+      "loss": 3.6959,
+      "step": 4138
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.572883944139326e-06,
+      "loss": 3.69,
+      "step": 4139
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.570896568126994e-06,
+      "loss": 3.7093,
+      "step": 4140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5689091007251e-06,
+      "loss": 3.7311,
+      "step": 4141
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5669215422518005e-06,
+      "loss": 3.6435,
+      "step": 4142
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.564933893025266e-06,
+      "loss": 3.7462,
+      "step": 4143
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.562946153363681e-06,
+      "loss": 3.6925,
+      "step": 4144
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5609583235852425e-06,
+      "loss": 3.7344,
+      "step": 4145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.558970404008166e-06,
+      "loss": 3.5862,
+      "step": 4146
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.556982394950678e-06,
+      "loss": 3.7435,
+      "step": 4147
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5549942967310214e-06,
+      "loss": 3.7267,
+      "step": 4148
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.553006109667452e-06,
+      "loss": 3.733,
+      "step": 4149
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.551017834078239e-06,
+      "loss": 3.7052,
+      "step": 4150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.549029470281668e-06,
+      "loss": 3.6437,
+      "step": 4151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.547041018596039e-06,
+      "loss": 3.7419,
+      "step": 4152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.545052479339662e-06,
+      "loss": 3.6411,
+      "step": 4153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.543063852830867e-06,
+      "loss": 3.6947,
+      "step": 4154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.541075139387994e-06,
+      "loss": 3.7114,
+      "step": 4155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.539086339329398e-06,
+      "loss": 3.6772,
+      "step": 4156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.537097452973447e-06,
+      "loss": 3.6731,
+      "step": 4157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.535108480638522e-06,
+      "loss": 3.6406,
+      "step": 4158
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.533119422643023e-06,
+      "loss": 3.7514,
+      "step": 4159
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.531130279305357e-06,
+      "loss": 3.685,
+      "step": 4160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.529141050943947e-06,
+      "loss": 3.6945,
+      "step": 4161
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.527151737877232e-06,
+      "loss": 3.692,
+      "step": 4162
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.525162340423663e-06,
+      "loss": 3.6411,
+      "step": 4163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.523172858901703e-06,
+      "loss": 3.818,
+      "step": 4164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.521183293629829e-06,
+      "loss": 3.559,
+      "step": 4165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5191936449265345e-06,
+      "loss": 3.6791,
+      "step": 4166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.517203913110323e-06,
+      "loss": 3.6306,
+      "step": 4167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.515214098499712e-06,
+      "loss": 3.5631,
+      "step": 4168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5132242014132335e-06,
+      "loss": 3.6292,
+      "step": 4169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.511234222169429e-06,
+      "loss": 3.6378,
+      "step": 4170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.50924416108686e-06,
+      "loss": 3.632,
+      "step": 4171
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.507254018484094e-06,
+      "loss": 3.6947,
+      "step": 4172
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.505263794679715e-06,
+      "loss": 3.706,
+      "step": 4173
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5032734899923205e-06,
+      "loss": 3.7416,
+      "step": 4174
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.501283104740519e-06,
+      "loss": 3.7045,
+      "step": 4175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.499292639242933e-06,
+      "loss": 3.6808,
+      "step": 4176
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.497302093818199e-06,
+      "loss": 3.6508,
+      "step": 4177
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.495311468784964e-06,
+      "loss": 3.7528,
+      "step": 4178
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.493320764461888e-06,
+      "loss": 3.8076,
+      "step": 4179
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.491329981167647e-06,
+      "loss": 3.6819,
+      "step": 4180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.489339119220924e-06,
+      "loss": 3.6881,
+      "step": 4181
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.487348178940422e-06,
+      "loss": 3.7973,
+      "step": 4182
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.485357160644847e-06,
+      "loss": 3.693,
+      "step": 4183
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.483366064652925e-06,
+      "loss": 3.7172,
+      "step": 4184
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.481374891283395e-06,
+      "loss": 3.752,
+      "step": 4185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.479383640855e-06,
+      "loss": 3.665,
+      "step": 4186
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.477392313686505e-06,
+      "loss": 3.5335,
+      "step": 4187
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.475400910096682e-06,
+      "loss": 3.7089,
+      "step": 4188
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4734094304043164e-06,
+      "loss": 3.6406,
+      "step": 4189
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.471417874928205e-06,
+      "loss": 3.6867,
+      "step": 4190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.469426243987159e-06,
+      "loss": 3.5994,
+      "step": 4191
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4674345379e-06,
+      "loss": 3.6785,
+      "step": 4192
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4654427569855625e-06,
+      "loss": 3.5851,
+      "step": 4193
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.46345090156269e-06,
+      "loss": 3.6677,
+      "step": 4194
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.461458971950242e-06,
+      "loss": 3.5919,
+      "step": 4195
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4594669684670874e-06,
+      "loss": 3.695,
+      "step": 4196
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4574748914321074e-06,
+      "loss": 3.7527,
+      "step": 4197
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.455482741164196e-06,
+      "loss": 3.8014,
+      "step": 4198
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.453490517982258e-06,
+      "loss": 3.6481,
+      "step": 4199
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.451498222205211e-06,
+      "loss": 3.6806,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.449505854151978e-06,
+      "loss": 3.6679,
+      "step": 4201
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.447513414141507e-06,
+      "loss": 3.5971,
+      "step": 4202
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.445520902492743e-06,
+      "loss": 3.6454,
+      "step": 4203
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.443528319524653e-06,
+      "loss": 3.6948,
+      "step": 4204
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.441535665556208e-06,
+      "loss": 3.6709,
+      "step": 4205
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.439542940906395e-06,
+      "loss": 3.714,
+      "step": 4206
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.43755014589421e-06,
+      "loss": 3.7622,
+      "step": 4207
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.435557280838663e-06,
+      "loss": 3.7271,
+      "step": 4208
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.433564346058771e-06,
+      "loss": 3.7177,
+      "step": 4209
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.431571341873567e-06,
+      "loss": 3.6693,
+      "step": 4210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4295782686020905e-06,
+      "loss": 3.7507,
+      "step": 4211
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.427585126563395e-06,
+      "loss": 3.7313,
+      "step": 4212
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.425591916076545e-06,
+      "loss": 3.7231,
+      "step": 4213
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.423598637460614e-06,
+      "loss": 3.7635,
+      "step": 4214
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.421605291034688e-06,
+      "loss": 3.6346,
+      "step": 4215
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.419611877117864e-06,
+      "loss": 3.6421,
+      "step": 4216
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.417618396029249e-06,
+      "loss": 3.6843,
+      "step": 4217
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.415624848087959e-06,
+      "loss": 3.6606,
+      "step": 4218
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.413631233613126e-06,
+      "loss": 3.6625,
+      "step": 4219
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.411637552923887e-06,
+      "loss": 3.5939,
+      "step": 4220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.409643806339392e-06,
+      "loss": 3.7425,
+      "step": 4221
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.407649994178802e-06,
+      "loss": 3.7267,
+      "step": 4222
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.405656116761289e-06,
+      "loss": 3.7141,
+      "step": 4223
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.403662174406033e-06,
+      "loss": 3.6092,
+      "step": 4224
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.401668167432225e-06,
+      "loss": 3.7312,
+      "step": 4225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.399674096159071e-06,
+      "loss": 3.7115,
+      "step": 4226
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.39767996090578e-06,
+      "loss": 3.7077,
+      "step": 4227
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.395685761991576e-06,
+      "loss": 3.5797,
+      "step": 4228
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.393691499735691e-06,
+      "loss": 3.6895,
+      "step": 4229
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.391697174457369e-06,
+      "loss": 3.6794,
+      "step": 4230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.389702786475862e-06,
+      "loss": 3.7067,
+      "step": 4231
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.387708336110434e-06,
+      "loss": 3.7083,
+      "step": 4232
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.385713823680359e-06,
+      "loss": 3.7691,
+      "step": 4233
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.383719249504918e-06,
+      "loss": 3.7204,
+      "step": 4234
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.381724613903405e-06,
+      "loss": 3.7561,
+      "step": 4235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.379729917195124e-06,
+      "loss": 3.6524,
+      "step": 4236
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3777351596993845e-06,
+      "loss": 3.7005,
+      "step": 4237
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.375740341735512e-06,
+      "loss": 3.7417,
+      "step": 4238
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.373745463622837e-06,
+      "loss": 3.7391,
+      "step": 4239
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.371750525680703e-06,
+      "loss": 3.7432,
+      "step": 4240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.369755528228458e-06,
+      "loss": 3.7256,
+      "step": 4241
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.367760471585464e-06,
+      "loss": 3.7318,
+      "step": 4242
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.365765356071091e-06,
+      "loss": 3.7514,
+      "step": 4243
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.36377018200472e-06,
+      "loss": 3.7306,
+      "step": 4244
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.361774949705739e-06,
+      "loss": 3.7765,
+      "step": 4245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.359779659493546e-06,
+      "loss": 3.6852,
+      "step": 4246
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.357784311687549e-06,
+      "loss": 3.6416,
+      "step": 4247
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.355788906607164e-06,
+      "loss": 3.6636,
+      "step": 4248
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.35379344457182e-06,
+      "loss": 3.7131,
+      "step": 4249
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.351797925900949e-06,
+      "loss": 3.657,
+      "step": 4250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.349802350913995e-06,
+      "loss": 3.6774,
+      "step": 4251
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.347806719930414e-06,
+      "loss": 3.6903,
+      "step": 4252
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.345811033269666e-06,
+      "loss": 3.7356,
+      "step": 4253
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.343815291251221e-06,
+      "loss": 3.6989,
+      "step": 4254
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.341819494194561e-06,
+      "loss": 3.8059,
+      "step": 4255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.339823642419174e-06,
+      "loss": 3.613,
+      "step": 4256
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3378277362445585e-06,
+      "loss": 3.667,
+      "step": 4257
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.335831775990217e-06,
+      "loss": 3.7288,
+      "step": 4258
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.333835761975669e-06,
+      "loss": 3.7234,
+      "step": 4259
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.331839694520436e-06,
+      "loss": 3.695,
+      "step": 4260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.32984357394405e-06,
+      "loss": 3.6722,
+      "step": 4261
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.327847400566052e-06,
+      "loss": 3.6358,
+      "step": 4262
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3258511747059895e-06,
+      "loss": 3.8336,
+      "step": 4263
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.323854896683422e-06,
+      "loss": 3.6647,
+      "step": 4264
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.321858566817913e-06,
+      "loss": 3.6482,
+      "step": 4265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.319862185429039e-06,
+      "loss": 3.5895,
+      "step": 4266
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.31786575283638e-06,
+      "loss": 3.6803,
+      "step": 4267
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.315869269359527e-06,
+      "loss": 3.5865,
+      "step": 4268
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.313872735318079e-06,
+      "loss": 3.6551,
+      "step": 4269
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.311876151031642e-06,
+      "loss": 3.7189,
+      "step": 4270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.309879516819833e-06,
+      "loss": 3.5828,
+      "step": 4271
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.307882833002271e-06,
+      "loss": 3.8008,
+      "step": 4272
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.305886099898591e-06,
+      "loss": 3.6287,
+      "step": 4273
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.303889317828428e-06,
+      "loss": 3.6374,
+      "step": 4274
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.301892487111431e-06,
+      "loss": 3.6395,
+      "step": 4275
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.299895608067253e-06,
+      "loss": 3.6491,
+      "step": 4276
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.297898681015554e-06,
+      "loss": 3.6189,
+      "step": 4277
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.295901706276008e-06,
+      "loss": 3.6526,
+      "step": 4278
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2939046841682875e-06,
+      "loss": 3.7579,
+      "step": 4279
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.291907615012081e-06,
+      "loss": 3.6787,
+      "step": 4280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.289910499127079e-06,
+      "loss": 3.7292,
+      "step": 4281
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2879133368329806e-06,
+      "loss": 3.7504,
+      "step": 4282
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2859161284494955e-06,
+      "loss": 3.6146,
+      "step": 4283
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.283918874296338e-06,
+      "loss": 3.7627,
+      "step": 4284
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.281921574693227e-06,
+      "loss": 3.7489,
+      "step": 4285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2799242299598964e-06,
+      "loss": 3.8044,
+      "step": 4286
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2779268404160785e-06,
+      "loss": 3.6893,
+      "step": 4287
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2759294063815205e-06,
+      "loss": 3.6934,
+      "step": 4288
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.27393192817597e-06,
+      "loss": 3.5608,
+      "step": 4289
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.271934406119186e-06,
+      "loss": 3.7797,
+      "step": 4290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.269936840530934e-06,
+      "loss": 3.6824,
+      "step": 4291
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.267939231730985e-06,
+      "loss": 3.7053,
+      "step": 4292
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.265941580039119e-06,
+      "loss": 3.7512,
+      "step": 4293
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.263943885775119e-06,
+      "loss": 3.7718,
+      "step": 4294
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.261946149258781e-06,
+      "loss": 3.7114,
+      "step": 4295
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.259948370809902e-06,
+      "loss": 3.6583,
+      "step": 4296
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.257950550748288e-06,
+      "loss": 3.7158,
+      "step": 4297
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2559526893937525e-06,
+      "loss": 3.656,
+      "step": 4298
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.253954787066115e-06,
+      "loss": 3.6911,
+      "step": 4299
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.251956844085198e-06,
+      "loss": 3.7322,
+      "step": 4300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.249958860770837e-06,
+      "loss": 3.7595,
+      "step": 4301
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.24796083744287e-06,
+      "loss": 3.7452,
+      "step": 4302
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.245962774421141e-06,
+      "loss": 3.7952,
+      "step": 4303
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.243964672025502e-06,
+      "loss": 3.7568,
+      "step": 4304
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2419665305758125e-06,
+      "loss": 3.6443,
+      "step": 4305
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.239968350391934e-06,
+      "loss": 3.4912,
+      "step": 4306
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.237970131793739e-06,
+      "loss": 3.6386,
+      "step": 4307
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.235971875101101e-06,
+      "loss": 3.7827,
+      "step": 4308
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2339735806339056e-06,
+      "loss": 3.6357,
+      "step": 4309
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2319752487120405e-06,
+      "loss": 3.8261,
+      "step": 4310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.229976879655397e-06,
+      "loss": 3.6652,
+      "step": 4311
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.22797847378388e-06,
+      "loss": 3.6799,
+      "step": 4312
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.225980031417394e-06,
+      "loss": 3.7043,
+      "step": 4313
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.22398155287585e-06,
+      "loss": 3.6809,
+      "step": 4314
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.221983038479167e-06,
+      "loss": 3.7272,
+      "step": 4315
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.219984488547269e-06,
+      "loss": 3.769,
+      "step": 4316
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.217985903400086e-06,
+      "loss": 3.6441,
+      "step": 4317
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.215987283357552e-06,
+      "loss": 3.7355,
+      "step": 4318
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.213988628739608e-06,
+      "loss": 3.6864,
+      "step": 4319
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2119899398662e-06,
+      "loss": 3.7064,
+      "step": 4320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.20999121705728e-06,
+      "loss": 3.6814,
+      "step": 4321
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.207992460632805e-06,
+      "loss": 3.7104,
+      "step": 4322
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.205993670912736e-06,
+      "loss": 3.774,
+      "step": 4323
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.203994848217043e-06,
+      "loss": 3.6223,
+      "step": 4324
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.201995992865698e-06,
+      "loss": 3.649,
+      "step": 4325
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.19999710517868e-06,
+      "loss": 3.673,
+      "step": 4326
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.197998185475971e-06,
+      "loss": 3.7471,
+      "step": 4327
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.195999234077561e-06,
+      "loss": 3.744,
+      "step": 4328
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.194000251303443e-06,
+      "loss": 3.5626,
+      "step": 4329
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1920012374736165e-06,
+      "loss": 3.7613,
+      "step": 4330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1900021929080835e-06,
+      "loss": 3.7844,
+      "step": 4331
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.188003117926854e-06,
+      "loss": 3.7613,
+      "step": 4332
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.186004012849941e-06,
+      "loss": 3.7666,
+      "step": 4333
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.184004877997364e-06,
+      "loss": 3.7106,
+      "step": 4334
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.182005713689145e-06,
+      "loss": 3.6619,
+      "step": 4335
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1800065202453095e-06,
+      "loss": 3.6754,
+      "step": 4336
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.178007297985893e-06,
+      "loss": 3.7401,
+      "step": 4337
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.176008047230933e-06,
+      "loss": 3.863,
+      "step": 4338
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.174008768300467e-06,
+      "loss": 3.6794,
+      "step": 4339
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1720094615145455e-06,
+      "loss": 3.6974,
+      "step": 4340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.170010127193218e-06,
+      "loss": 3.7846,
+      "step": 4341
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.168010765656538e-06,
+      "loss": 3.6455,
+      "step": 4342
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.166011377224566e-06,
+      "loss": 3.6263,
+      "step": 4343
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.164011962217366e-06,
+      "loss": 3.798,
+      "step": 4344
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.162012520955005e-06,
+      "loss": 3.5898,
+      "step": 4345
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.160013053757554e-06,
+      "loss": 3.7066,
+      "step": 4346
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.158013560945091e-06,
+      "loss": 3.7598,
+      "step": 4347
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.156014042837696e-06,
+      "loss": 3.6948,
+      "step": 4348
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.154014499755455e-06,
+      "loss": 3.8137,
+      "step": 4349
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.152014932018452e-06,
+      "loss": 3.6345,
+      "step": 4350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.150015339946783e-06,
+      "loss": 3.6739,
+      "step": 4351
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.148015723860543e-06,
+      "loss": 3.7321,
+      "step": 4352
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.146016084079833e-06,
+      "loss": 3.7863,
+      "step": 4353
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.144016420924756e-06,
+      "loss": 3.6825,
+      "step": 4354
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.142016734715419e-06,
+      "loss": 3.6656,
+      "step": 4355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.140017025771936e-06,
+      "loss": 3.736,
+      "step": 4356
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.138017294414419e-06,
+      "loss": 3.7916,
+      "step": 4357
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.136017540962988e-06,
+      "loss": 3.7547,
+      "step": 4358
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1340177657377634e-06,
+      "loss": 3.717,
+      "step": 4359
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1320179690588735e-06,
+      "loss": 3.7402,
+      "step": 4360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.130018151246445e-06,
+      "loss": 3.6537,
+      "step": 4361
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.128018312620611e-06,
+      "loss": 3.6445,
+      "step": 4362
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.126018453501507e-06,
+      "loss": 3.6082,
+      "step": 4363
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.124018574209272e-06,
+      "loss": 3.758,
+      "step": 4364
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.12201867506405e-06,
+      "loss": 3.7798,
+      "step": 4365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1200187563859836e-06,
+      "loss": 3.6386,
+      "step": 4366
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.118018818495223e-06,
+      "loss": 3.7002,
+      "step": 4367
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.116018861711919e-06,
+      "loss": 3.7694,
+      "step": 4368
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.114018886356227e-06,
+      "loss": 3.7736,
+      "step": 4369
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.112018892748305e-06,
+      "loss": 3.5568,
+      "step": 4370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.110018881208312e-06,
+      "loss": 3.6667,
+      "step": 4371
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.108018852056411e-06,
+      "loss": 3.7561,
+      "step": 4372
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.10601880561277e-06,
+      "loss": 3.7296,
+      "step": 4373
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.104018742197557e-06,
+      "loss": 3.6421,
+      "step": 4374
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.102018662130944e-06,
+      "loss": 3.6641,
+      "step": 4375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.100018565733106e-06,
+      "loss": 3.6871,
+      "step": 4376
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.098018453324218e-06,
+      "loss": 3.6059,
+      "step": 4377
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.096018325224462e-06,
+      "loss": 3.738,
+      "step": 4378
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.094018181754018e-06,
+      "loss": 3.8014,
+      "step": 4379
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.092018023233072e-06,
+      "loss": 3.6054,
+      "step": 4380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.09001784998181e-06,
+      "loss": 3.7464,
+      "step": 4381
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.088017662320422e-06,
+      "loss": 3.823,
+      "step": 4382
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.086017460569097e-06,
+      "loss": 3.7406,
+      "step": 4383
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.084017245048034e-06,
+      "loss": 3.6473,
+      "step": 4384
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.082017016077423e-06,
+      "loss": 3.7594,
+      "step": 4385
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.080016773977466e-06,
+      "loss": 3.8061,
+      "step": 4386
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.078016519068364e-06,
+      "loss": 3.6824,
+      "step": 4387
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0760162516703156e-06,
+      "loss": 3.7316,
+      "step": 4388
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0740159721035286e-06,
+      "loss": 3.6997,
+      "step": 4389
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.072015680688208e-06,
+      "loss": 3.7537,
+      "step": 4390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.070015377744561e-06,
+      "loss": 3.631,
+      "step": 4391
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.068015063592799e-06,
+      "loss": 3.7303,
+      "step": 4392
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.066014738553135e-06,
+      "loss": 3.7432,
+      "step": 4393
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.064014402945778e-06,
+      "loss": 3.5933,
+      "step": 4394
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.062014057090948e-06,
+      "loss": 3.6197,
+      "step": 4395
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.06001370130886e-06,
+      "loss": 3.6482,
+      "step": 4396
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.058013335919732e-06,
+      "loss": 3.7048,
+      "step": 4397
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.056012961243786e-06,
+      "loss": 3.7004,
+      "step": 4398
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0540125776012405e-06,
+      "loss": 3.7225,
+      "step": 4399
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.052012185312322e-06,
+      "loss": 3.665,
+      "step": 4400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.050011784697252e-06,
+      "loss": 3.784,
+      "step": 4401
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.048011376076257e-06,
+      "loss": 3.7126,
+      "step": 4402
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.046010959769564e-06,
+      "loss": 3.655,
+      "step": 4403
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.044010536097402e-06,
+      "loss": 3.6869,
+      "step": 4404
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.042010105379998e-06,
+      "loss": 3.8237,
+      "step": 4405
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.040009667937586e-06,
+      "loss": 3.6487,
+      "step": 4406
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.038009224090395e-06,
+      "loss": 3.7004,
+      "step": 4407
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.036008774158658e-06,
+      "loss": 3.6797,
+      "step": 4408
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.034008318462607e-06,
+      "loss": 3.7281,
+      "step": 4409
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.032007857322481e-06,
+      "loss": 3.7675,
+      "step": 4410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.030007391058511e-06,
+      "loss": 3.6333,
+      "step": 4411
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.028006919990936e-06,
+      "loss": 3.7684,
+      "step": 4412
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.026006444439991e-06,
+      "loss": 3.7796,
+      "step": 4413
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.024005964725915e-06,
+      "loss": 3.6753,
+      "step": 4414
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.022005481168945e-06,
+      "loss": 3.7123,
+      "step": 4415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0200049940893225e-06,
+      "loss": 3.7421,
+      "step": 4416
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.018004503807283e-06,
+      "loss": 3.6449,
+      "step": 4417
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.016004010643071e-06,
+      "loss": 3.6927,
+      "step": 4418
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.014003514916923e-06,
+      "loss": 3.6253,
+      "step": 4419
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.012003016949082e-06,
+      "loss": 3.6473,
+      "step": 4420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.010002517059787e-06,
+      "loss": 3.7324,
+      "step": 4421
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.008002015569284e-06,
+      "loss": 3.5777,
+      "step": 4422
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.00600151279781e-06,
+      "loss": 3.7536,
+      "step": 4423
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.004001009065611e-06,
+      "loss": 3.6962,
+      "step": 4424
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.002000504692928e-06,
+      "loss": 3.6782,
+      "step": 4425
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5e-06,
+      "loss": 3.6731,
+      "step": 4426
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9979994953070755e-06,
+      "loss": 3.8393,
+      "step": 4427
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.99599899093439e-06,
+      "loss": 3.6714,
+      "step": 4428
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.993998487202192e-06,
+      "loss": 3.7424,
+      "step": 4429
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.991997984430718e-06,
+      "loss": 3.7327,
+      "step": 4430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.989997482940214e-06,
+      "loss": 3.6763,
+      "step": 4431
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.98799698305092e-06,
+      "loss": 3.7949,
+      "step": 4432
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.98599648508308e-06,
+      "loss": 3.6782,
+      "step": 4433
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.983995989356931e-06,
+      "loss": 3.6454,
+      "step": 4434
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.981995496192719e-06,
+      "loss": 3.6466,
+      "step": 4435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.979995005910679e-06,
+      "loss": 3.7403,
+      "step": 4436
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9779945188310565e-06,
+      "loss": 3.6712,
+      "step": 4437
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.975994035274086e-06,
+      "loss": 3.6599,
+      "step": 4438
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.973993555560011e-06,
+      "loss": 3.6412,
+      "step": 4439
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.971993080009065e-06,
+      "loss": 3.677,
+      "step": 4440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.969992608941491e-06,
+      "loss": 3.7177,
+      "step": 4441
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.967992142677521e-06,
+      "loss": 3.7522,
+      "step": 4442
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9659916815373935e-06,
+      "loss": 3.6822,
+      "step": 4443
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9639912258413435e-06,
+      "loss": 3.7207,
+      "step": 4444
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.961990775909608e-06,
+      "loss": 3.7286,
+      "step": 4445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.959990332062415e-06,
+      "loss": 3.7629,
+      "step": 4446
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.957989894620004e-06,
+      "loss": 3.7765,
+      "step": 4447
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.955989463902599e-06,
+      "loss": 3.71,
+      "step": 4448
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.953989040230438e-06,
+      "loss": 3.7775,
+      "step": 4449
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.951988623923745e-06,
+      "loss": 3.6952,
+      "step": 4450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.949988215302751e-06,
+      "loss": 3.8576,
+      "step": 4451
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.94798781468768e-06,
+      "loss": 3.7616,
+      "step": 4452
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.945987422398761e-06,
+      "loss": 3.625,
+      "step": 4453
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.943987038756216e-06,
+      "loss": 3.6478,
+      "step": 4454
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.941986664080269e-06,
+      "loss": 3.7907,
+      "step": 4455
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.939986298691141e-06,
+      "loss": 3.771,
+      "step": 4456
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9379859429090536e-06,
+      "loss": 3.6701,
+      "step": 4457
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.935985597054223e-06,
+      "loss": 3.7127,
+      "step": 4458
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.933985261446869e-06,
+      "loss": 3.6392,
+      "step": 4459
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.931984936407202e-06,
+      "loss": 3.6247,
+      "step": 4460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.929984622255441e-06,
+      "loss": 3.6723,
+      "step": 4461
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.927984319311794e-06,
+      "loss": 3.748,
+      "step": 4462
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.925984027896474e-06,
+      "loss": 3.6006,
+      "step": 4463
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.923983748329685e-06,
+      "loss": 3.6417,
+      "step": 4464
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9219834809316395e-06,
+      "loss": 3.7388,
+      "step": 4465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.919983226022535e-06,
+      "loss": 3.8644,
+      "step": 4466
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.917982983922578e-06,
+      "loss": 3.6645,
+      "step": 4467
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9159827549519676e-06,
+      "loss": 3.6862,
+      "step": 4468
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.913982539430904e-06,
+      "loss": 3.6175,
+      "step": 4469
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.911982337679579e-06,
+      "loss": 3.7053,
+      "step": 4470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.909982150018191e-06,
+      "loss": 3.732,
+      "step": 4471
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.907981976766928e-06,
+      "loss": 3.7281,
+      "step": 4472
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9059818182459835e-06,
+      "loss": 3.6476,
+      "step": 4473
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9039816747755385e-06,
+      "loss": 3.7119,
+      "step": 4474
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.901981546675783e-06,
+      "loss": 3.6681,
+      "step": 4475
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.899981434266895e-06,
+      "loss": 3.7526,
+      "step": 4476
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.897981337869057e-06,
+      "loss": 3.7239,
+      "step": 4477
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.895981257802444e-06,
+      "loss": 3.6996,
+      "step": 4478
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8939811943872315e-06,
+      "loss": 3.7071,
+      "step": 4479
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.891981147943589e-06,
+      "loss": 3.668,
+      "step": 4480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8899811187916905e-06,
+      "loss": 3.8737,
+      "step": 4481
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.887981107251696e-06,
+      "loss": 3.6349,
+      "step": 4482
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.885981113643774e-06,
+      "loss": 3.7176,
+      "step": 4483
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.883981138288081e-06,
+      "loss": 3.7257,
+      "step": 4484
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8819811815047785e-06,
+      "loss": 3.7137,
+      "step": 4485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.879981243614017e-06,
+      "loss": 3.6151,
+      "step": 4486
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.877981324935952e-06,
+      "loss": 3.6442,
+      "step": 4487
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8759814257907275e-06,
+      "loss": 3.6872,
+      "step": 4488
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.873981546498495e-06,
+      "loss": 3.5743,
+      "step": 4489
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.87198168737939e-06,
+      "loss": 3.7045,
+      "step": 4490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.869981848753556e-06,
+      "loss": 3.6503,
+      "step": 4491
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.867982030941127e-06,
+      "loss": 3.7285,
+      "step": 4492
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.865982234262237e-06,
+      "loss": 3.7243,
+      "step": 4493
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.863982459037012e-06,
+      "loss": 3.6386,
+      "step": 4494
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.861982705585583e-06,
+      "loss": 3.6387,
+      "step": 4495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.859982974228065e-06,
+      "loss": 3.7921,
+      "step": 4496
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8579832652845816e-06,
+      "loss": 3.6866,
+      "step": 4497
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.855983579075245e-06,
+      "loss": 3.7146,
+      "step": 4498
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.853983915920168e-06,
+      "loss": 3.6892,
+      "step": 4499
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.851984276139458e-06,
+      "loss": 3.781,
+      "step": 4500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.849984660053219e-06,
+      "loss": 3.7373,
+      "step": 4501
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8479850679815496e-06,
+      "loss": 3.6447,
+      "step": 4502
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.845985500244547e-06,
+      "loss": 3.7134,
+      "step": 4503
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.843985957162304e-06,
+      "loss": 3.6563,
+      "step": 4504
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.841986439054911e-06,
+      "loss": 3.7145,
+      "step": 4505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.839986946242446e-06,
+      "loss": 3.7379,
+      "step": 4506
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.837987479044998e-06,
+      "loss": 3.683,
+      "step": 4507
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.835988037782635e-06,
+      "loss": 3.6639,
+      "step": 4508
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.833988622775435e-06,
+      "loss": 3.6582,
+      "step": 4509
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.831989234343462e-06,
+      "loss": 3.6774,
+      "step": 4510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.829989872806784e-06,
+      "loss": 3.7365,
+      "step": 4511
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.827990538485456e-06,
+      "loss": 3.6686,
+      "step": 4512
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.825991231699534e-06,
+      "loss": 3.7196,
+      "step": 4513
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.82399195276907e-06,
+      "loss": 3.649,
+      "step": 4514
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.821992702014108e-06,
+      "loss": 3.7242,
+      "step": 4515
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.819993479754693e-06,
+      "loss": 3.6228,
+      "step": 4516
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.817994286310858e-06,
+      "loss": 3.6739,
+      "step": 4517
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8159951220026385e-06,
+      "loss": 3.6189,
+      "step": 4518
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8139959871500594e-06,
+      "loss": 3.6358,
+      "step": 4519
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.811996882073148e-06,
+      "loss": 3.5731,
+      "step": 4520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.809997807091918e-06,
+      "loss": 3.8832,
+      "step": 4521
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.807998762526387e-06,
+      "loss": 3.8198,
+      "step": 4522
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.805999748696559e-06,
+      "loss": 3.6967,
+      "step": 4523
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.804000765922441e-06,
+      "loss": 3.6164,
+      "step": 4524
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.802001814524031e-06,
+      "loss": 3.6704,
+      "step": 4525
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8000028948213224e-06,
+      "loss": 3.7062,
+      "step": 4526
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7980040071343025e-06,
+      "loss": 3.5825,
+      "step": 4527
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.796005151782958e-06,
+      "loss": 3.7975,
+      "step": 4528
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7940063290872645e-06,
+      "loss": 3.6729,
+      "step": 4529
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.792007539367198e-06,
+      "loss": 3.7703,
+      "step": 4530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.790008782942721e-06,
+      "loss": 3.5847,
+      "step": 4531
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.788010060133802e-06,
+      "loss": 3.6881,
+      "step": 4532
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7860113712603934e-06,
+      "loss": 3.6786,
+      "step": 4533
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.78401271664245e-06,
+      "loss": 3.6868,
+      "step": 4534
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.782014096599915e-06,
+      "loss": 3.7221,
+      "step": 4535
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.780015511452732e-06,
+      "loss": 3.7564,
+      "step": 4536
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.778016961520835e-06,
+      "loss": 3.7727,
+      "step": 4537
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.776018447124152e-06,
+      "loss": 3.7054,
+      "step": 4538
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.774019968582608e-06,
+      "loss": 3.5942,
+      "step": 4539
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.772021526216123e-06,
+      "loss": 3.7422,
+      "step": 4540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7700231203446044e-06,
+      "loss": 3.7566,
+      "step": 4541
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.768024751287963e-06,
+      "loss": 3.61,
+      "step": 4542
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.766026419366095e-06,
+      "loss": 3.6074,
+      "step": 4543
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.764028124898901e-06,
+      "loss": 3.7852,
+      "step": 4544
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7620298682062625e-06,
+      "loss": 3.6876,
+      "step": 4545
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.760031649608068e-06,
+      "loss": 3.7146,
+      "step": 4546
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.758033469424189e-06,
+      "loss": 3.6669,
+      "step": 4547
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.756035327974499e-06,
+      "loss": 3.6933,
+      "step": 4548
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.754037225578861e-06,
+      "loss": 3.6963,
+      "step": 4549
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7520391625571325e-06,
+      "loss": 3.8296,
+      "step": 4550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.750041139229164e-06,
+      "loss": 3.8088,
+      "step": 4551
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.748043155914804e-06,
+      "loss": 3.7754,
+      "step": 4552
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.746045212933887e-06,
+      "loss": 3.694,
+      "step": 4553
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.744047310606249e-06,
+      "loss": 3.6939,
+      "step": 4554
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7420494492517126e-06,
+      "loss": 3.7355,
+      "step": 4555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.740051629190099e-06,
+      "loss": 3.7462,
+      "step": 4556
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7380538507412196e-06,
+      "loss": 3.6595,
+      "step": 4557
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7360561142248814e-06,
+      "loss": 3.6766,
+      "step": 4558
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.734058419960883e-06,
+      "loss": 3.6204,
+      "step": 4559
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.732060768269016e-06,
+      "loss": 3.7077,
+      "step": 4560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.730063159469067e-06,
+      "loss": 3.7073,
+      "step": 4561
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.728065593880815e-06,
+      "loss": 3.7242,
+      "step": 4562
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.72606807182403e-06,
+      "loss": 3.7163,
+      "step": 4563
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.724070593618482e-06,
+      "loss": 3.5989,
+      "step": 4564
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7220731595839214e-06,
+      "loss": 3.6108,
+      "step": 4565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.720075770040105e-06,
+      "loss": 3.6034,
+      "step": 4566
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.718078425306772e-06,
+      "loss": 3.7113,
+      "step": 4567
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.716081125703665e-06,
+      "loss": 3.7087,
+      "step": 4568
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7140838715505044e-06,
+      "loss": 3.7236,
+      "step": 4569
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.712086663167021e-06,
+      "loss": 3.693,
+      "step": 4570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.710089500872922e-06,
+      "loss": 3.5909,
+      "step": 4571
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.708092384987921e-06,
+      "loss": 3.7752,
+      "step": 4572
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7060953158317124e-06,
+      "loss": 3.81,
+      "step": 4573
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.704098293723995e-06,
+      "loss": 3.7201,
+      "step": 4574
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7021013189844464e-06,
+      "loss": 3.7445,
+      "step": 4575
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.70010439193275e-06,
+      "loss": 3.7704,
+      "step": 4576
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.69810751288857e-06,
+      "loss": 3.7736,
+      "step": 4577
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.696110682171573e-06,
+      "loss": 3.688,
+      "step": 4578
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.69411390010141e-06,
+      "loss": 3.7593,
+      "step": 4579
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6921171669977304e-06,
+      "loss": 3.7887,
+      "step": 4580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.690120483180168e-06,
+      "loss": 3.5803,
+      "step": 4581
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6881238489683596e-06,
+      "loss": 3.6589,
+      "step": 4582
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.686127264681922e-06,
+      "loss": 3.6429,
+      "step": 4583
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.684130730640475e-06,
+      "loss": 3.7027,
+      "step": 4584
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.682134247163621e-06,
+      "loss": 3.7872,
+      "step": 4585
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.680137814570963e-06,
+      "loss": 3.7115,
+      "step": 4586
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6781414331820875e-06,
+      "loss": 3.6701,
+      "step": 4587
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.676145103316579e-06,
+      "loss": 3.7281,
+      "step": 4588
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6741488252940105e-06,
+      "loss": 3.781,
+      "step": 4589
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.67215259943395e-06,
+      "loss": 3.6605,
+      "step": 4590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.67015642605595e-06,
+      "loss": 3.7575,
+      "step": 4591
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6681603054795654e-06,
+      "loss": 3.7138,
+      "step": 4592
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.666164238024331e-06,
+      "loss": 3.6435,
+      "step": 4593
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.664168224009784e-06,
+      "loss": 3.7998,
+      "step": 4594
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.662172263755443e-06,
+      "loss": 3.7389,
+      "step": 4595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.660176357580827e-06,
+      "loss": 3.7815,
+      "step": 4596
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6581805058054416e-06,
+      "loss": 3.729,
+      "step": 4597
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.65618470874878e-06,
+      "loss": 3.7576,
+      "step": 4598
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6541889667303375e-06,
+      "loss": 3.7054,
+      "step": 4599
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.652193280069588e-06,
+      "loss": 3.6991,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.650197649086007e-06,
+      "loss": 3.7314,
+      "step": 4601
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.648202074099052e-06,
+      "loss": 3.7315,
+      "step": 4602
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.646206555428183e-06,
+      "loss": 3.6799,
+      "step": 4603
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.644211093392837e-06,
+      "loss": 3.6568,
+      "step": 4604
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.642215688312452e-06,
+      "loss": 3.8136,
+      "step": 4605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.640220340506456e-06,
+      "loss": 3.737,
+      "step": 4606
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.638225050294263e-06,
+      "loss": 3.7763,
+      "step": 4607
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.636229817995281e-06,
+      "loss": 3.6961,
+      "step": 4608
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.634234643928911e-06,
+      "loss": 3.6492,
+      "step": 4609
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6322395284145385e-06,
+      "loss": 3.6506,
+      "step": 4610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.630244471771546e-06,
+      "loss": 3.678,
+      "step": 4611
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6282494743193e-06,
+      "loss": 3.7012,
+      "step": 4612
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.626254536377165e-06,
+      "loss": 3.6671,
+      "step": 4613
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.624259658264489e-06,
+      "loss": 3.7586,
+      "step": 4614
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.622264840300617e-06,
+      "loss": 3.7893,
+      "step": 4615
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.620270082804879e-06,
+      "loss": 3.6732,
+      "step": 4616
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.618275386096597e-06,
+      "loss": 3.6707,
+      "step": 4617
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.616280750495085e-06,
+      "loss": 3.7876,
+      "step": 4618
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.614286176319644e-06,
+      "loss": 3.7105,
+      "step": 4619
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.612291663889567e-06,
+      "loss": 3.7241,
+      "step": 4620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.61029721352414e-06,
+      "loss": 3.7247,
+      "step": 4621
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.608302825542633e-06,
+      "loss": 3.6377,
+      "step": 4622
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.606308500264312e-06,
+      "loss": 3.7604,
+      "step": 4623
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.604314238008426e-06,
+      "loss": 3.6494,
+      "step": 4624
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.602320039094223e-06,
+      "loss": 3.6871,
+      "step": 4625
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.60032590384093e-06,
+      "loss": 3.721,
+      "step": 4626
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.598331832567776e-06,
+      "loss": 3.5914,
+      "step": 4627
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.596337825593969e-06,
+      "loss": 3.7071,
+      "step": 4628
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.594343883238714e-06,
+      "loss": 3.6888,
+      "step": 4629
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5923500058212e-06,
+      "loss": 3.6041,
+      "step": 4630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.59035619366061e-06,
+      "loss": 3.7111,
+      "step": 4631
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.588362447076115e-06,
+      "loss": 3.6476,
+      "step": 4632
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5863687663868775e-06,
+      "loss": 3.6805,
+      "step": 4633
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.584375151912043e-06,
+      "loss": 3.693,
+      "step": 4634
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.582381603970754e-06,
+      "loss": 3.6038,
+      "step": 4635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5803881228821375e-06,
+      "loss": 3.6941,
+      "step": 4636
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.578394708965314e-06,
+      "loss": 3.6805,
+      "step": 4637
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.576401362539387e-06,
+      "loss": 3.6441,
+      "step": 4638
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.574408083923457e-06,
+      "loss": 3.6564,
+      "step": 4639
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.572414873436606e-06,
+      "loss": 3.5873,
+      "step": 4640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.570421731397911e-06,
+      "loss": 3.7688,
+      "step": 4641
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.568428658126435e-06,
+      "loss": 3.6917,
+      "step": 4642
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.56643565394123e-06,
+      "loss": 3.7486,
+      "step": 4643
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.564442719161338e-06,
+      "loss": 3.7056,
+      "step": 4644
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.562449854105791e-06,
+      "loss": 3.8606,
+      "step": 4645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.560457059093606e-06,
+      "loss": 3.706,
+      "step": 4646
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.558464334443794e-06,
+      "loss": 3.8161,
+      "step": 4647
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.556471680475348e-06,
+      "loss": 3.7994,
+      "step": 4648
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5544790975072584e-06,
+      "loss": 3.8018,
+      "step": 4649
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.552486585858494e-06,
+      "loss": 3.7234,
+      "step": 4650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.550494145848023e-06,
+      "loss": 3.6832,
+      "step": 4651
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.548501777794792e-06,
+      "loss": 3.6408,
+      "step": 4652
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5465094820177435e-06,
+      "loss": 3.673,
+      "step": 4653
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5445172588358055e-06,
+      "loss": 3.6969,
+      "step": 4654
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.542525108567894e-06,
+      "loss": 3.6047,
+      "step": 4655
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.540533031532913e-06,
+      "loss": 3.656,
+      "step": 4656
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.53854102804976e-06,
+      "loss": 3.7314,
+      "step": 4657
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.53654909843731e-06,
+      "loss": 3.7263,
+      "step": 4658
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.534557243014439e-06,
+      "loss": 3.7952,
+      "step": 4659
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.532565462099999e-06,
+      "loss": 3.7239,
+      "step": 4660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.530573756012841e-06,
+      "loss": 3.6339,
+      "step": 4661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.528582125071795e-06,
+      "loss": 3.7266,
+      "step": 4662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.526590569595686e-06,
+      "loss": 3.6754,
+      "step": 4663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.524599089903319e-06,
+      "loss": 3.6725,
+      "step": 4664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.522607686313497e-06,
+      "loss": 3.7938,
+      "step": 4665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.520616359145001e-06,
+      "loss": 3.732,
+      "step": 4666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.518625108716608e-06,
+      "loss": 3.7159,
+      "step": 4667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.516633935347075e-06,
+      "loss": 3.7969,
+      "step": 4668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.514642839355155e-06,
+      "loss": 3.6865,
+      "step": 4669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.512651821059579e-06,
+      "loss": 3.7133,
+      "step": 4670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.510660880779077e-06,
+      "loss": 3.7124,
+      "step": 4671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.508670018832353e-06,
+      "loss": 3.7052,
+      "step": 4672
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.506679235538113e-06,
+      "loss": 3.778,
+      "step": 4673
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5046885312150365e-06,
+      "loss": 3.7122,
+      "step": 4674
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.502697906181803e-06,
+      "loss": 3.7866,
+      "step": 4675
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5007073607570674e-06,
+      "loss": 3.6741,
+      "step": 4676
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.498716895259482e-06,
+      "loss": 3.7473,
+      "step": 4677
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.496726510007681e-06,
+      "loss": 3.7671,
+      "step": 4678
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.494736205320286e-06,
+      "loss": 3.7839,
+      "step": 4679
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.492745981515907e-06,
+      "loss": 3.6632,
+      "step": 4680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.490755838913142e-06,
+      "loss": 3.7096,
+      "step": 4681
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.488765777830571e-06,
+      "loss": 3.6341,
+      "step": 4682
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.486775798586769e-06,
+      "loss": 3.7105,
+      "step": 4683
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.48478590150029e-06,
+      "loss": 3.6435,
+      "step": 4684
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.482796086889679e-06,
+      "loss": 3.7151,
+      "step": 4685
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.480806355073467e-06,
+      "loss": 3.7091,
+      "step": 4686
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4788167063701725e-06,
+      "loss": 3.7092,
+      "step": 4687
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4768271410983e-06,
+      "loss": 3.6838,
+      "step": 4688
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.474837659576338e-06,
+      "loss": 3.6593,
+      "step": 4689
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.47284826212277e-06,
+      "loss": 3.7272,
+      "step": 4690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.470858949056055e-06,
+      "loss": 3.7394,
+      "step": 4691
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.468869720694647e-06,
+      "loss": 3.7352,
+      "step": 4692
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.466880577356978e-06,
+      "loss": 3.7237,
+      "step": 4693
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.464891519361479e-06,
+      "loss": 3.7492,
+      "step": 4694
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.462902547026555e-06,
+      "loss": 3.6668,
+      "step": 4695
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.460913660670604e-06,
+      "loss": 3.6804,
+      "step": 4696
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4589248606120065e-06,
+      "loss": 3.7814,
+      "step": 4697
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.456936147169133e-06,
+      "loss": 3.7199,
+      "step": 4698
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.454947520660339e-06,
+      "loss": 3.6159,
+      "step": 4699
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.452958981403963e-06,
+      "loss": 3.766,
+      "step": 4700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4509705297183334e-06,
+      "loss": 3.722,
+      "step": 4701
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.448982165921764e-06,
+      "loss": 3.6952,
+      "step": 4702
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.446993890332551e-06,
+      "loss": 3.803,
+      "step": 4703
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.445005703268981e-06,
+      "loss": 3.6869,
+      "step": 4704
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.443017605049323e-06,
+      "loss": 3.7455,
+      "step": 4705
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.441029595991836e-06,
+      "loss": 3.7721,
+      "step": 4706
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.439041676414758e-06,
+      "loss": 3.7506,
+      "step": 4707
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4370538466363216e-06,
+      "loss": 3.626,
+      "step": 4708
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.435066106974736e-06,
+      "loss": 3.6769,
+      "step": 4709
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.433078457748201e-06,
+      "loss": 3.7786,
+      "step": 4710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.431090899274903e-06,
+      "loss": 3.6891,
+      "step": 4711
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.429103431873009e-06,
+      "loss": 3.7269,
+      "step": 4712
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.427116055860675e-06,
+      "loss": 3.7389,
+      "step": 4713
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4251287715560454e-06,
+      "loss": 3.6744,
+      "step": 4714
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.423141579277241e-06,
+      "loss": 3.7029,
+      "step": 4715
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.421154479342377e-06,
+      "loss": 3.7609,
+      "step": 4716
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.419167472069546e-06,
+      "loss": 3.6977,
+      "step": 4717
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.417180557776835e-06,
+      "loss": 3.6207,
+      "step": 4718
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.415193736782304e-06,
+      "loss": 3.7387,
+      "step": 4719
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.413207009404012e-06,
+      "loss": 3.6317,
+      "step": 4720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.411220375959988e-06,
+      "loss": 3.6143,
+      "step": 4721
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.409233836768261e-06,
+      "loss": 3.6001,
+      "step": 4722
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.407247392146835e-06,
+      "loss": 3.6582,
+      "step": 4723
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4052610424137e-06,
+      "loss": 3.723,
+      "step": 4724
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.403274787886833e-06,
+      "loss": 3.7075,
+      "step": 4725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.401288628884199e-06,
+      "loss": 3.6428,
+      "step": 4726
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.399302565723738e-06,
+      "loss": 3.6962,
+      "step": 4727
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.397316598723385e-06,
+      "loss": 3.7287,
+      "step": 4728
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.395330728201052e-06,
+      "loss": 3.6838,
+      "step": 4729
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.393344954474643e-06,
+      "loss": 3.6895,
+      "step": 4730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.391359277862036e-06,
+      "loss": 3.6852,
+      "step": 4731
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.389373698681105e-06,
+      "loss": 3.6465,
+      "step": 4732
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3873882172497e-06,
+      "loss": 3.6761,
+      "step": 4733
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3854028338856585e-06,
+      "loss": 3.6181,
+      "step": 4734
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.383417548906805e-06,
+      "loss": 3.6291,
+      "step": 4735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.381432362630942e-06,
+      "loss": 3.7434,
+      "step": 4736
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.37944727537586e-06,
+      "loss": 3.704,
+      "step": 4737
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.377462287459338e-06,
+      "loss": 3.7115,
+      "step": 4738
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.375477399199127e-06,
+      "loss": 3.7194,
+      "step": 4739
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.373492610912976e-06,
+      "loss": 3.6833,
+      "step": 4740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3715079229186065e-06,
+      "loss": 3.8059,
+      "step": 4741
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.369523335533733e-06,
+      "loss": 3.752,
+      "step": 4742
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3675388490760465e-06,
+      "loss": 3.7227,
+      "step": 4743
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.365554463863228e-06,
+      "loss": 3.6919,
+      "step": 4744
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.363570180212936e-06,
+      "loss": 3.6417,
+      "step": 4745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.361585998442819e-06,
+      "loss": 3.7213,
+      "step": 4746
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.359601918870504e-06,
+      "loss": 3.6771,
+      "step": 4747
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3576179418136075e-06,
+      "loss": 3.6626,
+      "step": 4748
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3556340675897216e-06,
+      "loss": 3.6702,
+      "step": 4749
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3536502965164305e-06,
+      "loss": 3.7458,
+      "step": 4750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.351666628911295e-06,
+      "loss": 3.7377,
+      "step": 4751
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.349683065091864e-06,
+      "loss": 3.6481,
+      "step": 4752
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.347699605375666e-06,
+      "loss": 3.7223,
+      "step": 4753
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.345716250080217e-06,
+      "loss": 3.6868,
+      "step": 4754
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.343732999523011e-06,
+      "loss": 3.7875,
+      "step": 4755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3417498540215325e-06,
+      "loss": 3.6737,
+      "step": 4756
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3397668138932404e-06,
+      "loss": 3.7448,
+      "step": 4757
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.337783879455584e-06,
+      "loss": 3.6909,
+      "step": 4758
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.335801051025993e-06,
+      "loss": 3.6978,
+      "step": 4759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.33381832892188e-06,
+      "loss": 3.7608,
+      "step": 4760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.331835713460638e-06,
+      "loss": 3.7322,
+      "step": 4761
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.32985320495965e-06,
+      "loss": 3.6152,
+      "step": 4762
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.327870803736273e-06,
+      "loss": 3.7234,
+      "step": 4763
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3258885101078565e-06,
+      "loss": 3.639,
+      "step": 4764
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.323906324391722e-06,
+      "loss": 3.6063,
+      "step": 4765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3219242469051844e-06,
+      "loss": 3.6232,
+      "step": 4766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.319942277965531e-06,
+      "loss": 3.6915,
+      "step": 4767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.317960417890043e-06,
+      "loss": 3.7402,
+      "step": 4768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.315978666995973e-06,
+      "loss": 3.6072,
+      "step": 4769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.313997025600562e-06,
+      "loss": 3.5952,
+      "step": 4770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.312015494021034e-06,
+      "loss": 3.6646,
+      "step": 4771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3100340725745934e-06,
+      "loss": 3.6253,
+      "step": 4772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.308052761578431e-06,
+      "loss": 3.6681,
+      "step": 4773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.306071561349709e-06,
+      "loss": 3.6162,
+      "step": 4774
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.304090472205588e-06,
+      "loss": 3.6532,
+      "step": 4775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3021094944631955e-06,
+      "loss": 3.6877,
+      "step": 4776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.300128628439653e-06,
+      "loss": 3.7193,
+      "step": 4777
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.298147874452055e-06,
+      "loss": 3.5733,
+      "step": 4778
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.296167232817484e-06,
+      "loss": 3.6688,
+      "step": 4779
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.294186703853004e-06,
+      "loss": 3.6975,
+      "step": 4780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.292206287875657e-06,
+      "loss": 3.7574,
+      "step": 4781
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.290225985202469e-06,
+      "loss": 3.7877,
+      "step": 4782
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.288245796150455e-06,
+      "loss": 3.7182,
+      "step": 4783
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.286265721036595e-06,
+      "loss": 3.6326,
+      "step": 4784
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.284285760177869e-06,
+      "loss": 3.7793,
+      "step": 4785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.282305913891226e-06,
+      "loss": 3.6441,
+      "step": 4786
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.280326182493606e-06,
+      "loss": 3.7265,
+      "step": 4787
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.27834656630192e-06,
+      "loss": 3.8141,
+      "step": 4788
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.276367065633071e-06,
+      "loss": 3.8222,
+      "step": 4789
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.274387680803936e-06,
+      "loss": 3.6916,
+      "step": 4790
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2724084121313794e-06,
+      "loss": 3.7215,
+      "step": 4791
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.270429259932243e-06,
+      "loss": 3.8602,
+      "step": 4792
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.268450224523349e-06,
+      "loss": 3.5768,
+      "step": 4793
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2664713062215055e-06,
+      "loss": 3.7376,
+      "step": 4794
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.264492505343499e-06,
+      "loss": 3.6187,
+      "step": 4795
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.262513822206095e-06,
+      "loss": 3.6398,
+      "step": 4796
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.260535257126046e-06,
+      "loss": 3.6532,
+      "step": 4797
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.258556810420078e-06,
+      "loss": 3.7564,
+      "step": 4798
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2565784824049086e-06,
+      "loss": 3.5909,
+      "step": 4799
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.254600273397223e-06,
+      "loss": 3.7013,
+      "step": 4800
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2526221837137e-06,
+      "loss": 3.5889,
+      "step": 4801
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.250644213670989e-06,
+      "loss": 3.647,
+      "step": 4802
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.248666363585729e-06,
+      "loss": 3.8203,
+      "step": 4803
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.246688633774534e-06,
+      "loss": 3.7244,
+      "step": 4804
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.244711024554001e-06,
+      "loss": 3.6152,
+      "step": 4805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.242733536240705e-06,
+      "loss": 3.7516,
+      "step": 4806
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.240756169151209e-06,
+      "loss": 3.6201,
+      "step": 4807
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.238778923602045e-06,
+      "loss": 3.737,
+      "step": 4808
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.236801799909736e-06,
+      "loss": 3.6894,
+      "step": 4809
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.234824798390781e-06,
+      "loss": 3.6835,
+      "step": 4810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2328479193616604e-06,
+      "loss": 3.7547,
+      "step": 4811
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.230871163138831e-06,
+      "loss": 3.691,
+      "step": 4812
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.22889453003874e-06,
+      "loss": 3.6523,
+      "step": 4813
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2269180203778e-06,
+      "loss": 3.6937,
+      "step": 4814
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.224941634472419e-06,
+      "loss": 3.7579,
+      "step": 4815
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2229653726389765e-06,
+      "loss": 3.6364,
+      "step": 4816
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.220989235193832e-06,
+      "loss": 3.7235,
+      "step": 4817
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.219013222453328e-06,
+      "loss": 3.6694,
+      "step": 4818
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.21703733473379e-06,
+      "loss": 3.8319,
+      "step": 4819
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.215061572351513e-06,
+      "loss": 3.6463,
+      "step": 4820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.213085935622785e-06,
+      "loss": 3.6025,
+      "step": 4821
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.211110424863861e-06,
+      "loss": 3.8144,
+      "step": 4822
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.209135040390989e-06,
+      "loss": 3.6935,
+      "step": 4823
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.207159782520383e-06,
+      "loss": 3.804,
+      "step": 4824
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.205184651568252e-06,
+      "loss": 3.6734,
+      "step": 4825
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.203209647850768e-06,
+      "loss": 3.5923,
+      "step": 4826
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2012347716840966e-06,
+      "loss": 3.6725,
+      "step": 4827
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.199260023384376e-06,
+      "loss": 3.643,
+      "step": 4828
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.197285403267725e-06,
+      "loss": 3.7284,
+      "step": 4829
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.195310911650242e-06,
+      "loss": 3.6278,
+      "step": 4830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.193336548848008e-06,
+      "loss": 3.7193,
+      "step": 4831
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1913623151770765e-06,
+      "loss": 3.7108,
+      "step": 4832
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1893882109534875e-06,
+      "loss": 3.6014,
+      "step": 4833
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.187414236493255e-06,
+      "loss": 3.5565,
+      "step": 4834
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.185440392112376e-06,
+      "loss": 3.687,
+      "step": 4835
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.183466678126822e-06,
+      "loss": 3.7007,
+      "step": 4836
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1814930948525525e-06,
+      "loss": 3.6436,
+      "step": 4837
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.179519642605493e-06,
+      "loss": 3.6242,
+      "step": 4838
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.17754632170156e-06,
+      "loss": 3.5549,
+      "step": 4839
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.175573132456644e-06,
+      "loss": 3.7871,
+      "step": 4840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.173600075186612e-06,
+      "loss": 3.6115,
+      "step": 4841
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.171627150207314e-06,
+      "loss": 3.7361,
+      "step": 4842
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.16965435783458e-06,
+      "loss": 3.6682,
+      "step": 4843
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.167681698384211e-06,
+      "loss": 3.7344,
+      "step": 4844
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.165709172171997e-06,
+      "loss": 3.688,
+      "step": 4845
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.163736779513697e-06,
+      "loss": 3.6512,
+      "step": 4846
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1617645207250566e-06,
+      "loss": 3.6025,
+      "step": 4847
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1597923961217935e-06,
+      "loss": 3.775,
+      "step": 4848
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1578204060196104e-06,
+      "loss": 3.6229,
+      "step": 4849
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.155848550734181e-06,
+      "loss": 3.6663,
+      "step": 4850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.153876830581165e-06,
+      "loss": 3.7489,
+      "step": 4851
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.151905245876194e-06,
+      "loss": 3.6428,
+      "step": 4852
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.149933796934884e-06,
+      "loss": 3.7611,
+      "step": 4853
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.147962484072824e-06,
+      "loss": 3.7771,
+      "step": 4854
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.145991307605582e-06,
+      "loss": 3.7067,
+      "step": 4855
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.144020267848707e-06,
+      "loss": 3.6726,
+      "step": 4856
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.142049365117724e-06,
+      "loss": 3.6292,
+      "step": 4857
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1400785997281395e-06,
+      "loss": 3.7035,
+      "step": 4858
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.13810797199543e-06,
+      "loss": 3.7236,
+      "step": 4859
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.13613748223506e-06,
+      "loss": 3.6661,
+      "step": 4860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.134167130762463e-06,
+      "loss": 3.6411,
+      "step": 4861
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.132196917893057e-06,
+      "loss": 3.7782,
+      "step": 4862
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1302268439422335e-06,
+      "loss": 3.6361,
+      "step": 4863
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.128256909225366e-06,
+      "loss": 3.672,
+      "step": 4864
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1262871140578e-06,
+      "loss": 3.7226,
+      "step": 4865
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.124317458754865e-06,
+      "loss": 3.7016,
+      "step": 4866
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1223479436318614e-06,
+      "loss": 3.7197,
+      "step": 4867
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.120378569004074e-06,
+      "loss": 3.6617,
+      "step": 4868
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.11840933518676e-06,
+      "loss": 3.7165,
+      "step": 4869
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.116440242495157e-06,
+      "loss": 3.7309,
+      "step": 4870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.114471291244476e-06,
+      "loss": 3.7599,
+      "step": 4871
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.112502481749911e-06,
+      "loss": 3.8034,
+      "step": 4872
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.11053381432663e-06,
+      "loss": 3.6691,
+      "step": 4873
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.108565289289779e-06,
+      "loss": 3.7422,
+      "step": 4874
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.10659690695448e-06,
+      "loss": 3.7846,
+      "step": 4875
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.104628667635835e-06,
+      "loss": 3.7519,
+      "step": 4876
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.102660571648919e-06,
+      "loss": 3.6763,
+      "step": 4877
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.100692619308788e-06,
+      "loss": 3.6663,
+      "step": 4878
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.098724810930472e-06,
+      "loss": 3.6668,
+      "step": 4879
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0967571468289815e-06,
+      "loss": 3.7755,
+      "step": 4880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0947896273192975e-06,
+      "loss": 3.6474,
+      "step": 4881
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.092822252716387e-06,
+      "loss": 3.7329,
+      "step": 4882
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0908550233351854e-06,
+      "loss": 3.7435,
+      "step": 4883
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0888879394906094e-06,
+      "loss": 3.5354,
+      "step": 4884
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.086921001497552e-06,
+      "loss": 3.803,
+      "step": 4885
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.084954209670881e-06,
+      "loss": 3.79,
+      "step": 4886
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.082987564325441e-06,
+      "loss": 3.7505,
+      "step": 4887
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.081021065776058e-06,
+      "loss": 3.6529,
+      "step": 4888
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.079054714337526e-06,
+      "loss": 3.7556,
+      "step": 4889
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.077088510324624e-06,
+      "loss": 3.6863,
+      "step": 4890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0751224540521e-06,
+      "loss": 3.7268,
+      "step": 4891
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.073156545834685e-06,
+      "loss": 3.6751,
+      "step": 4892
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.071190785987079e-06,
+      "loss": 3.6227,
+      "step": 4893
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.069225174823968e-06,
+      "loss": 3.6036,
+      "step": 4894
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0672597126600025e-06,
+      "loss": 3.7337,
+      "step": 4895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.065294399809819e-06,
+      "loss": 3.7716,
+      "step": 4896
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.063329236588026e-06,
+      "loss": 3.7078,
+      "step": 4897
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.061364223309207e-06,
+      "loss": 3.7062,
+      "step": 4898
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0593993602879224e-06,
+      "loss": 3.6009,
+      "step": 4899
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.057434647838713e-06,
+      "loss": 3.7174,
+      "step": 4900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.055470086276086e-06,
+      "loss": 3.6047,
+      "step": 4901
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0535056759145345e-06,
+      "loss": 3.664,
+      "step": 4902
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.051541417068518e-06,
+      "loss": 3.7886,
+      "step": 4903
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.049577310052482e-06,
+      "loss": 3.7325,
+      "step": 4904
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.047613355180838e-06,
+      "loss": 3.6679,
+      "step": 4905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.04564955276798e-06,
+      "loss": 3.7793,
+      "step": 4906
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.043685903128274e-06,
+      "loss": 3.6284,
+      "step": 4907
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.041722406576062e-06,
+      "loss": 3.6852,
+      "step": 4908
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.039759063425663e-06,
+      "loss": 3.6516,
+      "step": 4909
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.037795873991369e-06,
+      "loss": 3.5868,
+      "step": 4910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.035832838587451e-06,
+      "loss": 3.8219,
+      "step": 4911
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.033869957528153e-06,
+      "loss": 3.7088,
+      "step": 4912
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.031907231127691e-06,
+      "loss": 3.7344,
+      "step": 4913
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0299446597002655e-06,
+      "loss": 3.7498,
+      "step": 4914
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0279822435600415e-06,
+      "loss": 3.7428,
+      "step": 4915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.026019983021168e-06,
+      "loss": 3.7451,
+      "step": 4916
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.02405787839776e-06,
+      "loss": 3.6835,
+      "step": 4917
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.022095930003919e-06,
+      "loss": 3.8149,
+      "step": 4918
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.020134138153708e-06,
+      "loss": 3.7644,
+      "step": 4919
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.018172503161179e-06,
+      "loss": 3.6947,
+      "step": 4920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.016211025340349e-06,
+      "loss": 3.7066,
+      "step": 4921
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.014249705005213e-06,
+      "loss": 3.533,
+      "step": 4922
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.01228854246974e-06,
+      "loss": 3.7586,
+      "step": 4923
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.010327538047877e-06,
+      "loss": 3.695,
+      "step": 4924
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.008366692053539e-06,
+      "loss": 3.6794,
+      "step": 4925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.006406004800624e-06,
+      "loss": 3.7233,
+      "step": 4926
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.004445476602996e-06,
+      "loss": 3.6841,
+      "step": 4927
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.002485107774503e-06,
+      "loss": 3.6938,
+      "step": 4928
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.000524898628957e-06,
+      "loss": 3.6844,
+      "step": 4929
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.998564849480154e-06,
+      "loss": 3.7006,
+      "step": 4930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.996604960641856e-06,
+      "loss": 3.6605,
+      "step": 4931
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.994645232427809e-06,
+      "loss": 3.6318,
+      "step": 4932
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.992685665151723e-06,
+      "loss": 3.785,
+      "step": 4933
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.990726259127289e-06,
+      "loss": 3.6904,
+      "step": 4934
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.98876701466817e-06,
+      "loss": 3.6447,
+      "step": 4935
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.986807932088004e-06,
+      "loss": 3.7559,
+      "step": 4936
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.984849011700401e-06,
+      "loss": 3.7683,
+      "step": 4937
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9828902538189484e-06,
+      "loss": 3.7527,
+      "step": 4938
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.980931658757206e-06,
+      "loss": 3.6496,
+      "step": 4939
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.978973226828702e-06,
+      "loss": 3.7084,
+      "step": 4940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.977014958346949e-06,
+      "loss": 3.7053,
+      "step": 4941
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.975056853625427e-06,
+      "loss": 3.603,
+      "step": 4942
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.97309891297759e-06,
+      "loss": 3.7535,
+      "step": 4943
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.971141136716866e-06,
+      "loss": 3.5891,
+      "step": 4944
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.969183525156662e-06,
+      "loss": 3.7006,
+      "step": 4945
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.967226078610346e-06,
+      "loss": 3.6807,
+      "step": 4946
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.965268797391276e-06,
+      "loss": 3.6459,
+      "step": 4947
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.963311681812768e-06,
+      "loss": 3.7899,
+      "step": 4948
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.961354732188124e-06,
+      "loss": 3.6824,
+      "step": 4949
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.959397948830609e-06,
+      "loss": 3.6855,
+      "step": 4950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9574413320534715e-06,
+      "loss": 3.6568,
+      "step": 4951
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.955484882169923e-06,
+      "loss": 3.6439,
+      "step": 4952
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.953528599493158e-06,
+      "loss": 3.703,
+      "step": 4953
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.951572484336338e-06,
+      "loss": 3.7161,
+      "step": 4954
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.949616537012599e-06,
+      "loss": 3.6684,
+      "step": 4955
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.947660757835049e-06,
+      "loss": 3.6744,
+      "step": 4956
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9457051471167765e-06,
+      "loss": 3.599,
+      "step": 4957
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.94374970517083e-06,
+      "loss": 3.7075,
+      "step": 4958
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.941794432310243e-06,
+      "loss": 3.6899,
+      "step": 4959
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.939839328848014e-06,
+      "loss": 3.6935,
+      "step": 4960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.937884395097121e-06,
+      "loss": 3.6511,
+      "step": 4961
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.935929631370506e-06,
+      "loss": 3.7575,
+      "step": 4962
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.933975037981096e-06,
+      "loss": 3.7435,
+      "step": 4963
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.932020615241777e-06,
+      "loss": 3.6863,
+      "step": 4964
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.930066363465418e-06,
+      "loss": 3.7425,
+      "step": 4965
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.928112282964858e-06,
+      "loss": 3.7842,
+      "step": 4966
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9261583740529054e-06,
+      "loss": 3.6671,
+      "step": 4967
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9242046370423434e-06,
+      "loss": 3.7451,
+      "step": 4968
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.922251072245932e-06,
+      "loss": 3.638,
+      "step": 4969
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.920297679976393e-06,
+      "loss": 3.6322,
+      "step": 4970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.918344460546432e-06,
+      "loss": 3.6575,
+      "step": 4971
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9163914142687185e-06,
+      "loss": 3.6731,
+      "step": 4972
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9144385414559e-06,
+      "loss": 3.692,
+      "step": 4973
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.912485842420592e-06,
+      "loss": 3.6649,
+      "step": 4974
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.910533317475386e-06,
+      "loss": 3.7673,
+      "step": 4975
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9085809669328415e-06,
+      "loss": 3.6672,
+      "step": 4976
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.906628791105494e-06,
+      "loss": 3.5327,
+      "step": 4977
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.904676790305849e-06,
+      "loss": 3.7325,
+      "step": 4978
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9027249648463835e-06,
+      "loss": 3.6362,
+      "step": 4979
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.900773315039548e-06,
+      "loss": 3.7588,
+      "step": 4980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.898821841197765e-06,
+      "loss": 3.6917,
+      "step": 4981
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.896870543633424e-06,
+      "loss": 3.6667,
+      "step": 4982
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.894919422658896e-06,
+      "loss": 3.711,
+      "step": 4983
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.892968478586513e-06,
+      "loss": 3.7677,
+      "step": 4984
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8910177117285865e-06,
+      "loss": 3.7065,
+      "step": 4985
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8890671223973945e-06,
+      "loss": 3.6448,
+      "step": 4986
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.887116710905193e-06,
+      "loss": 3.5913,
+      "step": 4987
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.885166477564199e-06,
+      "loss": 3.7306,
+      "step": 4988
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.883216422686613e-06,
+      "loss": 3.7963,
+      "step": 4989
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.881266546584599e-06,
+      "loss": 3.7743,
+      "step": 4990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.879316849570294e-06,
+      "loss": 3.6673,
+      "step": 4991
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.877367331955808e-06,
+      "loss": 3.781,
+      "step": 4992
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.875417994053222e-06,
+      "loss": 3.6707,
+      "step": 4993
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.873468836174585e-06,
+      "loss": 3.8211,
+      "step": 4994
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.871519858631923e-06,
+      "loss": 3.7144,
+      "step": 4995
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.869571061737226e-06,
+      "loss": 3.6742,
+      "step": 4996
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.867622445802464e-06,
+      "loss": 3.6835,
+      "step": 4997
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.865674011139567e-06,
+      "loss": 3.6508,
+      "step": 4998
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.863725758060447e-06,
+      "loss": 3.7823,
+      "step": 4999
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.861777686876978e-06,
+      "loss": 3.7411,
+      "step": 5000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.859829797901011e-06,
+      "loss": 3.6638,
+      "step": 5001
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.857882091444366e-06,
+      "loss": 3.7219,
+      "step": 5002
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8559345678188305e-06,
+      "loss": 3.6483,
+      "step": 5003
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.853987227336168e-06,
+      "loss": 3.6622,
+      "step": 5004
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.852040070308112e-06,
+      "loss": 3.7079,
+      "step": 5005
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8500930970463594e-06,
+      "loss": 3.4637,
+      "step": 5006
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.848146307862589e-06,
+      "loss": 3.6702,
+      "step": 5007
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8461997030684386e-06,
+      "loss": 3.7413,
+      "step": 5008
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.844253282975528e-06,
+      "loss": 3.6377,
+      "step": 5009
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.842307047895437e-06,
+      "loss": 3.6888,
+      "step": 5010
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.840360998139724e-06,
+      "loss": 3.7441,
+      "step": 5011
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.838415134019911e-06,
+      "loss": 3.7208,
+      "step": 5012
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.836469455847495e-06,
+      "loss": 3.71,
+      "step": 5013
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.834523963933941e-06,
+      "loss": 3.6989,
+      "step": 5014
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.832578658590687e-06,
+      "loss": 3.7199,
+      "step": 5015
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.830633540129135e-06,
+      "loss": 3.5878,
+      "step": 5016
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8286886088606656e-06,
+      "loss": 3.7306,
+      "step": 5017
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8267438650966205e-06,
+      "loss": 3.6151,
+      "step": 5018
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.824799309148321e-06,
+      "loss": 3.7036,
+      "step": 5019
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.822854941327046e-06,
+      "loss": 3.7468,
+      "step": 5020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.820910761944059e-06,
+      "loss": 3.544,
+      "step": 5021
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.81896677131058e-06,
+      "loss": 3.6408,
+      "step": 5022
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.817022969737808e-06,
+      "loss": 3.6913,
+      "step": 5023
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.815079357536907e-06,
+      "loss": 3.7589,
+      "step": 5024
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8131359350190095e-06,
+      "loss": 3.6707,
+      "step": 5025
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.811192702495225e-06,
+      "loss": 3.7762,
+      "step": 5026
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.809249660276623e-06,
+      "loss": 3.814,
+      "step": 5027
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8073068086742514e-06,
+      "loss": 3.6299,
+      "step": 5028
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8053641479991187e-06,
+      "loss": 3.6675,
+      "step": 5029
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.803421678562213e-06,
+      "loss": 3.7353,
+      "step": 5030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8014794006744802e-06,
+      "loss": 3.624,
+      "step": 5031
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.799537314646848e-06,
+      "loss": 3.7705,
+      "step": 5032
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7975954207902006e-06,
+      "loss": 3.6437,
+      "step": 5033
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7956537194154026e-06,
+      "loss": 3.655,
+      "step": 5034
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.793712210833281e-06,
+      "loss": 3.6177,
+      "step": 5035
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.791770895354635e-06,
+      "loss": 3.5847,
+      "step": 5036
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7898297732902308e-06,
+      "loss": 3.6719,
+      "step": 5037
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7878888449508067e-06,
+      "loss": 3.7066,
+      "step": 5038
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.785948110647064e-06,
+      "loss": 3.7076,
+      "step": 5039
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7840075706896824e-06,
+      "loss": 3.7123,
+      "step": 5040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.782067225389299e-06,
+      "loss": 3.6325,
+      "step": 5041
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7801270750565323e-06,
+      "loss": 3.7913,
+      "step": 5042
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7781871200019566e-06,
+      "loss": 3.643,
+      "step": 5043
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.776247360536127e-06,
+      "loss": 3.6878,
+      "step": 5044
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.774307796969557e-06,
+      "loss": 3.7229,
+      "step": 5045
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7723684296127362e-06,
+      "loss": 3.6306,
+      "step": 5046
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7704292587761193e-06,
+      "loss": 3.6522,
+      "step": 5047
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.768490284770131e-06,
+      "loss": 3.744,
+      "step": 5048
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.766551507905161e-06,
+      "loss": 3.6814,
+      "step": 5049
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7646129284915754e-06,
+      "loss": 3.6819,
+      "step": 5050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7626745468396974e-06,
+      "loss": 3.6481,
+      "step": 5051
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7607363632598305e-06,
+      "loss": 3.6899,
+      "step": 5052
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.758798378062235e-06,
+      "loss": 3.6919,
+      "step": 5053
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7568605915571498e-06,
+      "loss": 3.6992,
+      "step": 5054
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.754923004054773e-06,
+      "loss": 3.7307,
+      "step": 5055
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7529856158652792e-06,
+      "loss": 3.7784,
+      "step": 5056
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7510484272988033e-06,
+      "loss": 3.6907,
+      "step": 5057
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.749111438665455e-06,
+      "loss": 3.6505,
+      "step": 5058
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.747174650275307e-06,
+      "loss": 3.6555,
+      "step": 5059
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7452380624384026e-06,
+      "loss": 3.6665,
+      "step": 5060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.743301675464751e-06,
+      "loss": 3.6635,
+      "step": 5061
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7413654896643347e-06,
+      "loss": 3.738,
+      "step": 5062
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7394295053470937e-06,
+      "loss": 3.7435,
+      "step": 5063
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7374937228229472e-06,
+      "loss": 3.6852,
+      "step": 5064
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.735558142401773e-06,
+      "loss": 3.7645,
+      "step": 5065
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.733622764393423e-06,
+      "loss": 3.7346,
+      "step": 5066
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.731687589107711e-06,
+      "loss": 3.6911,
+      "step": 5067
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7297526168544253e-06,
+      "loss": 3.711,
+      "step": 5068
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7278178479433137e-06,
+      "loss": 3.7248,
+      "step": 5069
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.725883282684098e-06,
+      "loss": 3.8699,
+      "step": 5070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7239489213864642e-06,
+      "loss": 3.6563,
+      "step": 5071
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.722014764360067e-06,
+      "loss": 3.6725,
+      "step": 5072
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7200808119145247e-06,
+      "loss": 3.6334,
+      "step": 5073
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.718147064359432e-06,
+      "loss": 3.6074,
+      "step": 5074
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7162135220043383e-06,
+      "loss": 3.7214,
+      "step": 5075
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.714280185158771e-06,
+      "loss": 3.6675,
+      "step": 5076
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.712347054132217e-06,
+      "loss": 3.7996,
+      "step": 5077
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.710414129234137e-06,
+      "loss": 3.634,
+      "step": 5078
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.70848141077395e-06,
+      "loss": 3.6488,
+      "step": 5079
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.706548899061052e-06,
+      "loss": 3.7501,
+      "step": 5080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7046165944047974e-06,
+      "loss": 3.8098,
+      "step": 5081
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.702684497114513e-06,
+      "loss": 3.6779,
+      "step": 5082
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7007526074994906e-06,
+      "loss": 3.6584,
+      "step": 5083
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6988209258689877e-06,
+      "loss": 3.7172,
+      "step": 5084
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6968894525322285e-06,
+      "loss": 3.7289,
+      "step": 5085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6949581877984076e-06,
+      "loss": 3.6007,
+      "step": 5086
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6930271319766797e-06,
+      "loss": 3.6155,
+      "step": 5087
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6910962853761738e-06,
+      "loss": 3.7124,
+      "step": 5088
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6891656483059768e-06,
+      "loss": 3.7271,
+      "step": 5089
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6872352210751515e-06,
+      "loss": 3.674,
+      "step": 5090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.685305003992717e-06,
+      "loss": 3.689,
+      "step": 5091
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.683374997367668e-06,
+      "loss": 3.6622,
+      "step": 5092
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6814452015089584e-06,
+      "loss": 3.8173,
+      "step": 5093
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.679515616725514e-06,
+      "loss": 3.6945,
+      "step": 5094
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6775862433262232e-06,
+      "loss": 3.7558,
+      "step": 5095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.675657081619941e-06,
+      "loss": 3.6914,
+      "step": 5096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.673728131915488e-06,
+      "loss": 3.7668,
+      "step": 5097
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.671799394521657e-06,
+      "loss": 3.6441,
+      "step": 5098
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.669870869747195e-06,
+      "loss": 3.7918,
+      "step": 5099
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6679425579008278e-06,
+      "loss": 3.6468,
+      "step": 5100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.666014459291236e-06,
+      "loss": 3.6666,
+      "step": 5101
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6640865742270756e-06,
+      "loss": 3.6839,
+      "step": 5102
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6621589030169592e-06,
+      "loss": 3.8074,
+      "step": 5103
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6602314459694743e-06,
+      "loss": 3.6869,
+      "step": 5104
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.658304203393166e-06,
+      "loss": 3.7075,
+      "step": 5105
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6563771755965515e-06,
+      "loss": 3.6332,
+      "step": 5106
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.65445036288811e-06,
+      "loss": 3.5632,
+      "step": 5107
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.652523765576287e-06,
+      "loss": 3.5724,
+      "step": 5108
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6505973839694946e-06,
+      "loss": 3.8275,
+      "step": 5109
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6486712183761062e-06,
+      "loss": 3.6643,
+      "step": 5110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.646745269104469e-06,
+      "loss": 3.8174,
+      "step": 5111
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6448195364628857e-06,
+      "loss": 3.6928,
+      "step": 5112
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.642894020759633e-06,
+      "loss": 3.6121,
+      "step": 5113
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6409687223029445e-06,
+      "loss": 3.6454,
+      "step": 5114
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6390436414010278e-06,
+      "loss": 3.7804,
+      "step": 5115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6371187783620486e-06,
+      "loss": 3.6785,
+      "step": 5116
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6351941334941414e-06,
+      "loss": 3.6222,
+      "step": 5117
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.633269707105404e-06,
+      "loss": 3.6734,
+      "step": 5118
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.631345499503903e-06,
+      "loss": 3.5914,
+      "step": 5119
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6294215109976628e-06,
+      "loss": 3.7282,
+      "step": 5120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6274977418946804e-06,
+      "loss": 3.686,
+      "step": 5121
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.625574192502911e-06,
+      "loss": 3.6554,
+      "step": 5122
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6236508631302824e-06,
+      "loss": 3.7276,
+      "step": 5123
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6217277540846775e-06,
+      "loss": 3.7371,
+      "step": 5124
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.619804865673953e-06,
+      "loss": 3.717,
+      "step": 5125
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.617882198205923e-06,
+      "loss": 3.6976,
+      "step": 5126
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.615959751988373e-06,
+      "loss": 3.6728,
+      "step": 5127
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.614037527329048e-06,
+      "loss": 3.6765,
+      "step": 5128
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6121155245356587e-06,
+      "loss": 3.5714,
+      "step": 5129
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.610193743915881e-06,
+      "loss": 3.6473,
+      "step": 5130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6082721857773573e-06,
+      "loss": 3.6857,
+      "step": 5131
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.606350850427688e-06,
+      "loss": 3.7059,
+      "step": 5132
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6044297381744463e-06,
+      "loss": 3.6912,
+      "step": 5133
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6025088493251616e-06,
+      "loss": 3.6943,
+      "step": 5134
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6005881841873343e-06,
+      "loss": 3.6668,
+      "step": 5135
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5986677430684224e-06,
+      "loss": 3.7259,
+      "step": 5136
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5967475262758554e-06,
+      "loss": 3.6514,
+      "step": 5137
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.59482753411702e-06,
+      "loss": 3.6946,
+      "step": 5138
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.592907766899272e-06,
+      "loss": 3.6397,
+      "step": 5139
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5909882249299287e-06,
+      "loss": 3.6946,
+      "step": 5140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.589068908516271e-06,
+      "loss": 3.6719,
+      "step": 5141
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5871498179655453e-06,
+      "loss": 3.6725,
+      "step": 5142
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5852309535849626e-06,
+      "loss": 3.6912,
+      "step": 5143
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.583312315681693e-06,
+      "loss": 3.6617,
+      "step": 5144
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5813939045628763e-06,
+      "loss": 3.6324,
+      "step": 5145
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5794757205356113e-06,
+      "loss": 3.5842,
+      "step": 5146
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5775577639069644e-06,
+      "loss": 3.582,
+      "step": 5147
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5756400349839603e-06,
+      "loss": 3.7581,
+      "step": 5148
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.573722534073595e-06,
+      "loss": 3.7187,
+      "step": 5149
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.571805261482818e-06,
+      "loss": 3.7093,
+      "step": 5150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.569888217518553e-06,
+      "loss": 3.7299,
+      "step": 5151
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.567971402487679e-06,
+      "loss": 3.5634,
+      "step": 5152
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.566054816697042e-06,
+      "loss": 3.566,
+      "step": 5153
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5641384604534503e-06,
+      "loss": 3.6576,
+      "step": 5154
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5622223340636764e-06,
+      "loss": 3.7453,
+      "step": 5155
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5603064378344536e-06,
+      "loss": 3.6508,
+      "step": 5156
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5583907720724832e-06,
+      "loss": 3.7821,
+      "step": 5157
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.556475337084422e-06,
+      "loss": 3.6825,
+      "step": 5158
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5545601331768985e-06,
+      "loss": 3.6959,
+      "step": 5159
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.552645160656497e-06,
+      "loss": 3.8159,
+      "step": 5160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5507304198297705e-06,
+      "loss": 3.6372,
+      "step": 5161
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5488159110032285e-06,
+      "loss": 3.746,
+      "step": 5162
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5469016344833505e-06,
+      "loss": 3.6421,
+      "step": 5163
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.544987590576574e-06,
+      "loss": 3.6135,
+      "step": 5164
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5430737795893006e-06,
+      "loss": 3.5699,
+      "step": 5165
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5411602018278934e-06,
+      "loss": 3.6812,
+      "step": 5166
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5392468575986833e-06,
+      "loss": 3.7188,
+      "step": 5167
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.537333747207955e-06,
+      "loss": 3.6686,
+      "step": 5168
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5354208709619653e-06,
+      "loss": 3.7535,
+      "step": 5169
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.533508229166924e-06,
+      "loss": 3.6904,
+      "step": 5170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.531595822129014e-06,
+      "loss": 3.7009,
+      "step": 5171
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.529683650154368e-06,
+      "loss": 3.6949,
+      "step": 5172
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5277717135490947e-06,
+      "loss": 3.636,
+      "step": 5173
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5258600126192533e-06,
+      "loss": 3.6968,
+      "step": 5174
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5239485476708737e-06,
+      "loss": 3.673,
+      "step": 5175
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5220373190099428e-06,
+      "loss": 3.6873,
+      "step": 5176
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5201263269424122e-06,
+      "loss": 3.7315,
+      "step": 5177
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.518215571774194e-06,
+      "loss": 3.6883,
+      "step": 5178
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5163050538111663e-06,
+      "loss": 3.6294,
+      "step": 5179
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5143947733591633e-06,
+      "loss": 3.8186,
+      "step": 5180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5124847307239863e-06,
+      "loss": 3.6697,
+      "step": 5181
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.510574926211393e-06,
+      "loss": 3.6149,
+      "step": 5182
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5086653601271105e-06,
+      "loss": 3.7109,
+      "step": 5183
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.50675603277682e-06,
+      "loss": 3.6997,
+      "step": 5184
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5048469444661713e-06,
+      "loss": 3.7001,
+      "step": 5185
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.50293809550077e-06,
+      "loss": 3.6453,
+      "step": 5186
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.501029486186188e-06,
+      "loss": 3.7383,
+      "step": 5187
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.499121116827956e-06,
+      "loss": 3.6262,
+      "step": 5188
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.497212987731568e-06,
+      "loss": 3.6662,
+      "step": 5189
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4953050992024775e-06,
+      "loss": 3.6571,
+      "step": 5190
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.493397451546102e-06,
+      "loss": 3.7669,
+      "step": 5191
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.491490045067818e-06,
+      "loss": 3.7437,
+      "step": 5192
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.489582880072967e-06,
+      "loss": 3.6938,
+      "step": 5193
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.487675956866846e-06,
+      "loss": 3.724,
+      "step": 5194
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.48576927575472e-06,
+      "loss": 3.7304,
+      "step": 5195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.48386283704181e-06,
+      "loss": 3.6128,
+      "step": 5196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4819566410333007e-06,
+      "loss": 3.7233,
+      "step": 5197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4800506880343364e-06,
+      "loss": 3.6812,
+      "step": 5198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.478144978350024e-06,
+      "loss": 3.6714,
+      "step": 5199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4762395122854336e-06,
+      "loss": 3.6637,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.474334290145589e-06,
+      "loss": 3.7737,
+      "step": 5201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4724293122354845e-06,
+      "loss": 3.6946,
+      "step": 5202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4705245788600663e-06,
+      "loss": 3.6911,
+      "step": 5203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4686200903242497e-06,
+      "loss": 3.7474,
+      "step": 5204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4667158469329008e-06,
+      "loss": 3.7927,
+      "step": 5205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4648118489908588e-06,
+      "loss": 3.7066,
+      "step": 5206
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4629080968029128e-06,
+      "loss": 3.6979,
+      "step": 5207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.461004590673819e-06,
+      "loss": 3.6345,
+      "step": 5208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4591013309082904e-06,
+      "loss": 3.6974,
+      "step": 5209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4571983178110047e-06,
+      "loss": 3.6604,
+      "step": 5210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4552955516865946e-06,
+      "loss": 3.5852,
+      "step": 5211
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.45339303283966e-06,
+      "loss": 3.6814,
+      "step": 5212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.451490761574754e-06,
+      "loss": 3.6415,
+      "step": 5213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4495887381963964e-06,
+      "loss": 3.7279,
+      "step": 5214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4476869630090627e-06,
+      "loss": 3.6178,
+      "step": 5215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.445785436317193e-06,
+      "loss": 3.6448,
+      "step": 5216
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4438841584251814e-06,
+      "loss": 3.8284,
+      "step": 5217
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4419831296373897e-06,
+      "loss": 3.6627,
+      "step": 5218
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4400823502581325e-06,
+      "loss": 3.6949,
+      "step": 5219
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4381818205916907e-06,
+      "loss": 3.6893,
+      "step": 5220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.436281540942302e-06,
+      "loss": 3.7377,
+      "step": 5221
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.434381511614164e-06,
+      "loss": 3.6999,
+      "step": 5222
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4324817329114345e-06,
+      "loss": 3.6068,
+      "step": 5223
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4305822051382345e-06,
+      "loss": 3.5954,
+      "step": 5224
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4286829285986365e-06,
+      "loss": 3.7858,
+      "step": 5225
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4267839035966844e-06,
+      "loss": 3.6527,
+      "step": 5226
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4248851304363694e-06,
+      "loss": 3.5926,
+      "step": 5227
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.422986609421655e-06,
+      "loss": 3.6973,
+      "step": 5228
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.421088340856451e-06,
+      "loss": 3.7511,
+      "step": 5229
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4191903250446397e-06,
+      "loss": 3.6364,
+      "step": 5230
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4172925622900522e-06,
+      "loss": 3.7021,
+      "step": 5231
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4153950528964867e-06,
+      "loss": 3.6807,
+      "step": 5232
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.413497797167698e-06,
+      "loss": 3.654,
+      "step": 5233
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4116007954073992e-06,
+      "loss": 3.773,
+      "step": 5234
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4097040479192634e-06,
+      "loss": 3.7027,
+      "step": 5235
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4078075550069255e-06,
+      "loss": 3.6738,
+      "step": 5236
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4059113169739742e-06,
+      "loss": 3.7331,
+      "step": 5237
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4040153341239653e-06,
+      "loss": 3.726,
+      "step": 5238
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.402119606760404e-06,
+      "loss": 3.7633,
+      "step": 5239
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.400224135186765e-06,
+      "loss": 3.7492,
+      "step": 5240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3983289197064718e-06,
+      "loss": 3.6674,
+      "step": 5241
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.396433960622916e-06,
+      "loss": 3.7128,
+      "step": 5242
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.394539258239441e-06,
+      "loss": 3.684,
+      "step": 5243
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.392644812859354e-06,
+      "loss": 3.6154,
+      "step": 5244
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3907506247859193e-06,
+      "loss": 3.576,
+      "step": 5245
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.38885669432236e-06,
+      "loss": 3.7015,
+      "step": 5246
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.386963021771856e-06,
+      "loss": 3.6845,
+      "step": 5247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.385069607437552e-06,
+      "loss": 3.5842,
+      "step": 5248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3831764516225424e-06,
+      "loss": 3.7405,
+      "step": 5249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3812835546298893e-06,
+      "loss": 3.7198,
+      "step": 5250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.379390916762605e-06,
+      "loss": 3.7194,
+      "step": 5251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3774985383236685e-06,
+      "loss": 3.6004,
+      "step": 5252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.375606419616009e-06,
+      "loss": 3.6731,
+      "step": 5253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3737145609425226e-06,
+      "loss": 3.6929,
+      "step": 5254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.371822962606056e-06,
+      "loss": 3.7226,
+      "step": 5255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3699316249094195e-06,
+      "loss": 3.7358,
+      "step": 5256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3680405481553796e-06,
+      "loss": 3.7144,
+      "step": 5257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.366149732646661e-06,
+      "loss": 3.7719,
+      "step": 5258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.364259178685947e-06,
+      "loss": 3.6997,
+      "step": 5259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.36236888657588e-06,
+      "loss": 3.6164,
+      "step": 5260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3604788566190572e-06,
+      "loss": 3.7861,
+      "step": 5261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3585890891180383e-06,
+      "loss": 3.6672,
+      "step": 5262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.356699584375337e-06,
+      "loss": 3.5846,
+      "step": 5263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3548103426934287e-06,
+      "loss": 3.6793,
+      "step": 5264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3529213643747415e-06,
+      "loss": 3.7294,
+      "step": 5265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.351032649721668e-06,
+      "loss": 3.6881,
+      "step": 5266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3491441990365526e-06,
+      "loss": 3.7598,
+      "step": 5267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3472560126217004e-06,
+      "loss": 3.7531,
+      "step": 5268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.345368090779375e-06,
+      "loss": 3.6399,
+      "step": 5269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3434804338117954e-06,
+      "loss": 3.6968,
+      "step": 5270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.341593042021138e-06,
+      "loss": 3.7277,
+      "step": 5271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3397059157095412e-06,
+      "loss": 3.7028,
+      "step": 5272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3378190551790935e-06,
+      "loss": 3.6087,
+      "step": 5273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3359324607318493e-06,
+      "loss": 3.5875,
+      "step": 5274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.334046132669812e-06,
+      "loss": 3.7507,
+      "step": 5275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.33216007129495e-06,
+      "loss": 3.7423,
+      "step": 5276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3302742769091823e-06,
+      "loss": 3.7856,
+      "step": 5277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.328388749814392e-06,
+      "loss": 3.7608,
+      "step": 5278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3265034903124103e-06,
+      "loss": 3.6582,
+      "step": 5279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3246184987050366e-06,
+      "loss": 3.7304,
+      "step": 5280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3227337752940197e-06,
+      "loss": 3.6589,
+      "step": 5281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3208493203810655e-06,
+      "loss": 3.7668,
+      "step": 5282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.318965134267843e-06,
+      "loss": 3.6777,
+      "step": 5283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3170812172559695e-06,
+      "loss": 3.647,
+      "step": 5284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3151975696470286e-06,
+      "loss": 3.7044,
+      "step": 5285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3133141917425515e-06,
+      "loss": 3.5932,
+      "step": 5286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3114310838440357e-06,
+      "loss": 3.7492,
+      "step": 5287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.309548246252925e-06,
+      "loss": 3.6835,
+      "step": 5288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3076656792706306e-06,
+      "loss": 3.6854,
+      "step": 5289
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3057833831985116e-06,
+      "loss": 3.6773,
+      "step": 5290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3039013583378894e-06,
+      "loss": 3.5893,
+      "step": 5291
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3020196049900386e-06,
+      "loss": 3.6396,
+      "step": 5292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.300138123456195e-06,
+      "loss": 3.6155,
+      "step": 5293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.298256914037542e-06,
+      "loss": 3.6999,
+      "step": 5294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.296375977035231e-06,
+      "loss": 3.709,
+      "step": 5295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2944953127503593e-06,
+      "loss": 3.6662,
+      "step": 5296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2926149214839885e-06,
+      "loss": 3.653,
+      "step": 5297
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2907348035371297e-06,
+      "loss": 3.6465,
+      "step": 5298
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2888549592107568e-06,
+      "loss": 3.7222,
+      "step": 5299
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2869753888057936e-06,
+      "loss": 3.8101,
+      "step": 5300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.285096092623127e-06,
+      "loss": 3.6864,
+      "step": 5301
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.283217070963592e-06,
+      "loss": 3.669,
+      "step": 5302
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2813383241279866e-06,
+      "loss": 3.7264,
+      "step": 5303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2794598524170606e-06,
+      "loss": 3.6846,
+      "step": 5304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.277581656131523e-06,
+      "loss": 3.6562,
+      "step": 5305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.275703735572035e-06,
+      "loss": 3.7207,
+      "step": 5306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.273826091039218e-06,
+      "loss": 3.7046,
+      "step": 5307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.271948722833643e-06,
+      "loss": 3.6975,
+      "step": 5308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2700716312558445e-06,
+      "loss": 3.7076,
+      "step": 5309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.268194816606305e-06,
+      "loss": 3.5205,
+      "step": 5310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2663182791854705e-06,
+      "loss": 3.5991,
+      "step": 5311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.264442019293734e-06,
+      "loss": 3.7074,
+      "step": 5312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2625660372314537e-06,
+      "loss": 3.6026,
+      "step": 5313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2606903332989327e-06,
+      "loss": 3.6702,
+      "step": 5314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2588149077964397e-06,
+      "loss": 3.6643,
+      "step": 5315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2569397610241915e-06,
+      "loss": 3.759,
+      "step": 5316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.255064893282366e-06,
+      "loss": 3.546,
+      "step": 5317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2531903048710887e-06,
+      "loss": 3.7028,
+      "step": 5318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.251315996090452e-06,
+      "loss": 3.6596,
+      "step": 5319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.249441967240489e-06,
+      "loss": 3.6645,
+      "step": 5320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2475682186212023e-06,
+      "loss": 3.5692,
+      "step": 5321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.245694750532537e-06,
+      "loss": 3.6543,
+      "step": 5322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2438215632744054e-06,
+      "loss": 3.6669,
+      "step": 5323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.241948657146663e-06,
+      "loss": 3.7805,
+      "step": 5324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2400760324491316e-06,
+      "loss": 3.7273,
+      "step": 5325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.238203689481577e-06,
+      "loss": 3.7029,
+      "step": 5326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2363316285437297e-06,
+      "loss": 3.7113,
+      "step": 5327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2344598499352663e-06,
+      "loss": 3.797,
+      "step": 5328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.232588353955829e-06,
+      "loss": 3.673,
+      "step": 5329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.230717140905001e-06,
+      "loss": 3.7188,
+      "step": 5330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2288462110823336e-06,
+      "loss": 3.579,
+      "step": 5331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.226975564787322e-06,
+      "loss": 3.6276,
+      "step": 5332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2251052023194245e-06,
+      "loss": 3.6842,
+      "step": 5333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.223235123978047e-06,
+      "loss": 3.7021,
+      "step": 5334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.221365330062556e-06,
+      "loss": 3.5576,
+      "step": 5335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2194958208722656e-06,
+      "loss": 3.785,
+      "step": 5336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.217626596706452e-06,
+      "loss": 3.6309,
+      "step": 5337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.21575765786434e-06,
+      "loss": 3.612,
+      "step": 5338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2138890046451128e-06,
+      "loss": 3.682,
+      "step": 5339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2120206373479024e-06,
+      "loss": 3.811,
+      "step": 5340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.210152556271803e-06,
+      "loss": 3.7014,
+      "step": 5341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.208284761715854e-06,
+      "loss": 3.7323,
+      "step": 5342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2064172539790584e-06,
+      "loss": 3.631,
+      "step": 5343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.204550033360362e-06,
+      "loss": 3.8025,
+      "step": 5344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2026831001586768e-06,
+      "loss": 3.7714,
+      "step": 5345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.200816454672858e-06,
+      "loss": 3.7477,
+      "step": 5346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.198950097201724e-06,
+      "loss": 3.6706,
+      "step": 5347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1970840280440384e-06,
+      "loss": 3.7096,
+      "step": 5348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1952182474985273e-06,
+      "loss": 3.7506,
+      "step": 5349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.193352755863863e-06,
+      "loss": 3.6969,
+      "step": 5350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.191487553438677e-06,
+      "loss": 3.7714,
+      "step": 5351
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1896226405215503e-06,
+      "loss": 3.626,
+      "step": 5352
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.187758017411022e-06,
+      "loss": 3.7279,
+      "step": 5353
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.18589368440558e-06,
+      "loss": 3.6787,
+      "step": 5354
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1840296418036706e-06,
+      "loss": 3.6872,
+      "step": 5355
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1821658899036876e-06,
+      "loss": 3.6617,
+      "step": 5356
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.180302429003987e-06,
+      "loss": 3.7523,
+      "step": 5357
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1784392594028672e-06,
+      "loss": 3.7981,
+      "step": 5358
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1765763813985917e-06,
+      "loss": 3.7323,
+      "step": 5359
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.174713795289366e-06,
+      "loss": 3.5895,
+      "step": 5360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1728515013733576e-06,
+      "loss": 3.6617,
+      "step": 5361
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1709894999486828e-06,
+      "loss": 3.8607,
+      "step": 5362
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.169127791313413e-06,
+      "loss": 3.725,
+      "step": 5363
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1672663757655707e-06,
+      "loss": 3.769,
+      "step": 5364
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1654052536031354e-06,
+      "loss": 3.6589,
+      "step": 5365
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.163544425124035e-06,
+      "loss": 3.6937,
+      "step": 5366
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1616838906261504e-06,
+      "loss": 3.6479,
+      "step": 5367
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1598236504073215e-06,
+      "loss": 3.7132,
+      "step": 5368
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.157963704765333e-06,
+      "loss": 3.7533,
+      "step": 5369
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1561040539979305e-06,
+      "loss": 3.7001,
+      "step": 5370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.154244698402803e-06,
+      "loss": 3.6674,
+      "step": 5371
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.152385638277603e-06,
+      "loss": 3.6668,
+      "step": 5372
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.150526873919927e-06,
+      "loss": 3.5746,
+      "step": 5373
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1486684056273296e-06,
+      "loss": 3.6284,
+      "step": 5374
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.146810233697313e-06,
+      "loss": 3.7117,
+      "step": 5375
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.144952358427339e-06,
+      "loss": 3.7487,
+      "step": 5376
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1430947801148125e-06,
+      "loss": 3.6972,
+      "step": 5377
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1412374990571005e-06,
+      "loss": 3.7712,
+      "step": 5378
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1393805155515143e-06,
+      "loss": 3.7252,
+      "step": 5379
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.137523829895326e-06,
+      "loss": 3.6461,
+      "step": 5380
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1356674423857496e-06,
+      "loss": 3.6545,
+      "step": 5381
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.133811353319962e-06,
+      "loss": 3.7195,
+      "step": 5382
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.131955562995083e-06,
+      "loss": 3.6133,
+      "step": 5383
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1301000717081926e-06,
+      "loss": 3.6623,
+      "step": 5384
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1282448797563164e-06,
+      "loss": 3.7387,
+      "step": 5385
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.126389987436439e-06,
+      "loss": 3.6559,
+      "step": 5386
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.124535395045487e-06,
+      "loss": 3.6944,
+      "step": 5387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1226811028803514e-06,
+      "loss": 3.7096,
+      "step": 5388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1208271112378634e-06,
+      "loss": 3.7151,
+      "step": 5389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.118973420414816e-06,
+      "loss": 3.691,
+      "step": 5390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.117120030707944e-06,
+      "loss": 3.7737,
+      "step": 5391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.115266942413946e-06,
+      "loss": 3.6788,
+      "step": 5392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1134141558294595e-06,
+      "loss": 3.6012,
+      "step": 5393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1115616712510843e-06,
+      "loss": 3.5432,
+      "step": 5394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.109709488975365e-06,
+      "loss": 3.7846,
+      "step": 5395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.107857609298802e-06,
+      "loss": 3.6214,
+      "step": 5396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1060060325178445e-06,
+      "loss": 3.6157,
+      "step": 5397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1041547589288956e-06,
+      "loss": 3.7653,
+      "step": 5398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.102303788828307e-06,
+      "loss": 3.6184,
+      "step": 5399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1004531225123857e-06,
+      "loss": 3.7655,
+      "step": 5400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0986027602773848e-06,
+      "loss": 3.6329,
+      "step": 5401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0967527024195154e-06,
+      "loss": 3.6762,
+      "step": 5402
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0949029492349324e-06,
+      "loss": 3.6342,
+      "step": 5403
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.093053501019748e-06,
+      "loss": 3.7161,
+      "step": 5404
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0912043580700226e-06,
+      "loss": 3.6423,
+      "step": 5405
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0893555206817692e-06,
+      "loss": 3.7372,
+      "step": 5406
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0875069891509484e-06,
+      "loss": 3.6716,
+      "step": 5407
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0856587637734783e-06,
+      "loss": 3.6779,
+      "step": 5408
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0838108448452203e-06,
+      "loss": 3.7679,
+      "step": 5409
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.081963232661996e-06,
+      "loss": 3.555,
+      "step": 5410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.080115927519567e-06,
+      "loss": 3.6591,
+      "step": 5411
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0782689297136554e-06,
+      "loss": 3.7889,
+      "step": 5412
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.076422239539927e-06,
+      "loss": 3.7545,
+      "step": 5413
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0745758572940044e-06,
+      "loss": 3.768,
+      "step": 5414
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.072729783271454e-06,
+      "loss": 3.6421,
+      "step": 5415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.070884017767801e-06,
+      "loss": 3.7543,
+      "step": 5416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0690385610785134e-06,
+      "loss": 3.7876,
+      "step": 5417
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0671934134990174e-06,
+      "loss": 3.7796,
+      "step": 5418
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0653485753246803e-06,
+      "loss": 3.7064,
+      "step": 5419
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0635040468508294e-06,
+      "loss": 3.6661,
+      "step": 5420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0616598283727354e-06,
+      "loss": 3.7617,
+      "step": 5421
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.059815920185626e-06,
+      "loss": 3.6669,
+      "step": 5422
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0579723225846712e-06,
+      "loss": 3.6482,
+      "step": 5423
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0561290358649996e-06,
+      "loss": 3.7959,
+      "step": 5424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.054286060321682e-06,
+      "loss": 3.6583,
+      "step": 5425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0524433962497456e-06,
+      "loss": 3.7739,
+      "step": 5426
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0506010439441645e-06,
+      "loss": 3.6272,
+      "step": 5427
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.048759003699866e-06,
+      "loss": 3.6116,
+      "step": 5428
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0469172758117215e-06,
+      "loss": 3.6864,
+      "step": 5429
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0450758605745606e-06,
+      "loss": 3.7495,
+      "step": 5430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0432347582831547e-06,
+      "loss": 3.7761,
+      "step": 5431
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0413939692322304e-06,
+      "loss": 3.7936,
+      "step": 5432
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0395534937164622e-06,
+      "loss": 3.7068,
+      "step": 5433
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0377133320304773e-06,
+      "loss": 3.6771,
+      "step": 5434
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0358734844688463e-06,
+      "loss": 3.6577,
+      "step": 5435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0340339513260976e-06,
+      "loss": 3.678,
+      "step": 5436
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0321947328967003e-06,
+      "loss": 3.7385,
+      "step": 5437
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.030355829475082e-06,
+      "loss": 3.7279,
+      "step": 5438
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.028517241355613e-06,
+      "loss": 3.7255,
+      "step": 5439
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0266789688326187e-06,
+      "loss": 3.6194,
+      "step": 5440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0248410122003673e-06,
+      "loss": 3.6565,
+      "step": 5441
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.023003371753085e-06,
+      "loss": 3.7086,
+      "step": 5442
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.021166047784938e-06,
+      "loss": 3.6022,
+      "step": 5443
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0193290405900494e-06,
+      "loss": 3.6542,
+      "step": 5444
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0174923504624875e-06,
+      "loss": 3.7353,
+      "step": 5445
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0156559776962734e-06,
+      "loss": 3.7875,
+      "step": 5446
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0138199225853717e-06,
+      "loss": 3.6676,
+      "step": 5447
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0119841854237027e-06,
+      "loss": 3.692,
+      "step": 5448
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.010148766505129e-06,
+      "loss": 3.7329,
+      "step": 5449
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.00831366612347e-06,
+      "loss": 3.7404,
+      "step": 5450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.006478884572488e-06,
+      "loss": 3.575,
+      "step": 5451
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.004644422145894e-06,
+      "loss": 3.5995,
+      "step": 5452
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0028102791373536e-06,
+      "loss": 3.8362,
+      "step": 5453
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0009764558404756e-06,
+      "loss": 3.7891,
+      "step": 5454
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.999142952548821e-06,
+      "loss": 3.6776,
+      "step": 5455
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9973097695558982e-06,
+      "loss": 3.695,
+      "step": 5456
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.995476907155166e-06,
+      "loss": 3.7181,
+      "step": 5457
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.993644365640026e-06,
+      "loss": 3.6978,
+      "step": 5458
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9918121453038386e-06,
+      "loss": 3.6735,
+      "step": 5459
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9899802464399018e-06,
+      "loss": 3.6406,
+      "step": 5460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9881486693414717e-06,
+      "loss": 3.6699,
+      "step": 5461
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.986317414301745e-06,
+      "loss": 3.6566,
+      "step": 5462
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.984486481613874e-06,
+      "loss": 3.6621,
+      "step": 5463
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.982655871570952e-06,
+      "loss": 3.7303,
+      "step": 5464
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9808255844660273e-06,
+      "loss": 3.6545,
+      "step": 5465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.978995620592092e-06,
+      "loss": 3.6139,
+      "step": 5466
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9771659802420895e-06,
+      "loss": 3.7925,
+      "step": 5467
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9753366637089075e-06,
+      "loss": 3.5976,
+      "step": 5468
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9735076712853895e-06,
+      "loss": 3.7948,
+      "step": 5469
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9716790032643162e-06,
+      "loss": 3.583,
+      "step": 5470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9698506599384265e-06,
+      "loss": 3.7897,
+      "step": 5471
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9680226416003986e-06,
+      "loss": 3.6077,
+      "step": 5472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.966194948542869e-06,
+      "loss": 3.7325,
+      "step": 5473
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9643675810584105e-06,
+      "loss": 3.8091,
+      "step": 5474
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9625405394395536e-06,
+      "loss": 3.7134,
+      "step": 5475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.960713823978769e-06,
+      "loss": 3.6183,
+      "step": 5476
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9588874349684815e-06,
+      "loss": 3.6504,
+      "step": 5477
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.95706137270106e-06,
+      "loss": 3.7206,
+      "step": 5478
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9552356374688214e-06,
+      "loss": 3.7584,
+      "step": 5479
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9534102295640305e-06,
+      "loss": 3.809,
+      "step": 5480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9515851492789026e-06,
+      "loss": 3.6963,
+      "step": 5481
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9497603969055943e-06,
+      "loss": 3.7054,
+      "step": 5482
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.947935972736217e-06,
+      "loss": 3.7219,
+      "step": 5483
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9461118770628226e-06,
+      "loss": 3.633,
+      "step": 5484
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9442881101774168e-06,
+      "loss": 3.6282,
+      "step": 5485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9424646723719457e-06,
+      "loss": 3.8,
+      "step": 5486
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9406415639383113e-06,
+      "loss": 3.6979,
+      "step": 5487
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9388187851683537e-06,
+      "loss": 3.7638,
+      "step": 5488
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9369963363538678e-06,
+      "loss": 3.6736,
+      "step": 5489
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9351742177865906e-06,
+      "loss": 3.7182,
+      "step": 5490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9333524297582117e-06,
+      "loss": 3.7354,
+      "step": 5491
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9315309725603596e-06,
+      "loss": 3.6269,
+      "step": 5492
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.929709846484619e-06,
+      "loss": 3.6568,
+      "step": 5493
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.927889051822513e-06,
+      "loss": 3.6725,
+      "step": 5494
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.92606858886552e-06,
+      "loss": 3.604,
+      "step": 5495
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9242484579050566e-06,
+      "loss": 3.7237,
+      "step": 5496
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9224286592324958e-06,
+      "loss": 3.5456,
+      "step": 5497
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9206091931391466e-06,
+      "loss": 3.7463,
+      "step": 5498
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9187900599162753e-06,
+      "loss": 3.6949,
+      "step": 5499
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9169712598550885e-06,
+      "loss": 3.6547,
+      "step": 5500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.915152793246741e-06,
+      "loss": 3.6781,
+      "step": 5501
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9133346603823342e-06,
+      "loss": 3.7417,
+      "step": 5502
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9115168615529165e-06,
+      "loss": 3.745,
+      "step": 5503
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9096993970494825e-06,
+      "loss": 3.7006,
+      "step": 5504
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9078822671629737e-06,
+      "loss": 3.7092,
+      "step": 5505
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9060654721842764e-06,
+      "loss": 3.8004,
+      "step": 5506
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9042490124042257e-06,
+      "loss": 3.713,
+      "step": 5507
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9024328881136e-06,
+      "loss": 3.6098,
+      "step": 5508
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9006170996031303e-06,
+      "loss": 3.7899,
+      "step": 5509
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8988016471634835e-06,
+      "loss": 3.6996,
+      "step": 5510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.896986531085285e-06,
+      "loss": 3.7614,
+      "step": 5511
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.895171751659093e-06,
+      "loss": 3.5956,
+      "step": 5512
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.893357309175426e-06,
+      "loss": 3.8646,
+      "step": 5513
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8915432039247337e-06,
+      "loss": 3.6602,
+      "step": 5514
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.889729436197428e-06,
+      "loss": 3.7405,
+      "step": 5515
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.887916006283849e-06,
+      "loss": 3.6221,
+      "step": 5516
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.886102914474301e-06,
+      "loss": 3.7533,
+      "step": 5517
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.884290161059017e-06,
+      "loss": 3.6836,
+      "step": 5518
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8824777463281896e-06,
+      "loss": 3.6939,
+      "step": 5519
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8806656705719492e-06,
+      "loss": 3.8242,
+      "step": 5520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8788539340803754e-06,
+      "loss": 3.746,
+      "step": 5521
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.877042537143492e-06,
+      "loss": 3.7003,
+      "step": 5522
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.875231480051268e-06,
+      "loss": 3.6967,
+      "step": 5523
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8734207630936195e-06,
+      "loss": 3.6957,
+      "step": 5524
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.871610386560407e-06,
+      "loss": 3.7648,
+      "step": 5525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8698003507414374e-06,
+      "loss": 3.7259,
+      "step": 5526
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8679906559264626e-06,
+      "loss": 3.7209,
+      "step": 5527
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.86618130240518e-06,
+      "loss": 3.6204,
+      "step": 5528
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8643722904672324e-06,
+      "loss": 3.7635,
+      "step": 5529
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.862563620402207e-06,
+      "loss": 3.705,
+      "step": 5530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8607552924996374e-06,
+      "loss": 3.6802,
+      "step": 5531
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.858947307049001e-06,
+      "loss": 3.6548,
+      "step": 5532
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8571396643397263e-06,
+      "loss": 3.6177,
+      "step": 5533
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8553323646611754e-06,
+      "loss": 3.6852,
+      "step": 5534
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.853525408302669e-06,
+      "loss": 3.6882,
+      "step": 5535
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.851718795553461e-06,
+      "loss": 3.7358,
+      "step": 5536
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8499125267027565e-06,
+      "loss": 3.6398,
+      "step": 5537
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8481066020397054e-06,
+      "loss": 3.7305,
+      "step": 5538
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8463010218534005e-06,
+      "loss": 3.6532,
+      "step": 5539
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.84449578643288e-06,
+      "loss": 3.6745,
+      "step": 5540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8426908960671302e-06,
+      "loss": 3.6069,
+      "step": 5541
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8408863510450765e-06,
+      "loss": 3.6232,
+      "step": 5542
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8390821516555906e-06,
+      "loss": 3.6729,
+      "step": 5543
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8372782981874964e-06,
+      "loss": 3.6711,
+      "step": 5544
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8354747909295475e-06,
+      "loss": 3.7207,
+      "step": 5545
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.83367163017046e-06,
+      "loss": 3.7175,
+      "step": 5546
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8318688161988765e-06,
+      "loss": 3.6917,
+      "step": 5547
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.830066349303401e-06,
+      "loss": 3.7056,
+      "step": 5548
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.828264229772565e-06,
+      "loss": 3.6065,
+      "step": 5549
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8264624578948624e-06,
+      "loss": 3.6912,
+      "step": 5550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8246610339587132e-06,
+      "loss": 3.6664,
+      "step": 5551
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8228599582524975e-06,
+      "loss": 3.6686,
+      "step": 5552
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8210592310645304e-06,
+      "loss": 3.74,
+      "step": 5553
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8192588526830746e-06,
+      "loss": 3.6838,
+      "step": 5554
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.817458823396334e-06,
+      "loss": 3.7778,
+      "step": 5555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.815659143492461e-06,
+      "loss": 3.6999,
+      "step": 5556
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.813859813259548e-06,
+      "loss": 3.7449,
+      "step": 5557
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.812060832985634e-06,
+      "loss": 3.6419,
+      "step": 5558
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.810262202958701e-06,
+      "loss": 3.7069,
+      "step": 5559
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8084639234666753e-06,
+      "loss": 3.6461,
+      "step": 5560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.806665994797427e-06,
+      "loss": 3.7383,
+      "step": 5561
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.804868417238769e-06,
+      "loss": 3.6827,
+      "step": 5562
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.80307119107846e-06,
+      "loss": 3.6823,
+      "step": 5563
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8012743166042002e-06,
+      "loss": 3.5102,
+      "step": 5564
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7994777941036355e-06,
+      "loss": 3.7831,
+      "step": 5565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7976816238643546e-06,
+      "loss": 3.6426,
+      "step": 5566
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.795885806173887e-06,
+      "loss": 3.6299,
+      "step": 5567
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.794090341319715e-06,
+      "loss": 3.6262,
+      "step": 5568
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7922952295892506e-06,
+      "loss": 3.779,
+      "step": 5569
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7905004712698646e-06,
+      "loss": 3.6806,
+      "step": 5570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.788706066648854e-06,
+      "loss": 3.771,
+      "step": 5571
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.786912016013478e-06,
+      "loss": 3.6894,
+      "step": 5572
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.78511831965092e-06,
+      "loss": 3.6221,
+      "step": 5573
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7833249778483263e-06,
+      "loss": 3.5924,
+      "step": 5574
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7815319908927674e-06,
+      "loss": 3.6667,
+      "step": 5575
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7797393590712706e-06,
+      "loss": 3.7107,
+      "step": 5576
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7779470826708027e-06,
+      "loss": 3.6532,
+      "step": 5577
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.77615516197827e-06,
+      "loss": 3.6178,
+      "step": 5578
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7743635972805268e-06,
+      "loss": 3.5787,
+      "step": 5579
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7725723888643663e-06,
+      "loss": 3.6377,
+      "step": 5580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7707815370165274e-06,
+      "loss": 3.6513,
+      "step": 5581
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.768991042023691e-06,
+      "loss": 3.7739,
+      "step": 5582
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7672009041724807e-06,
+      "loss": 3.7077,
+      "step": 5583
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.765411123749463e-06,
+      "loss": 3.5504,
+      "step": 5584
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.763621701041148e-06,
+      "loss": 3.8017,
+      "step": 5585
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7618326363339874e-06,
+      "loss": 3.6003,
+      "step": 5586
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.760043929914376e-06,
+      "loss": 3.5569,
+      "step": 5587
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.758255582068651e-06,
+      "loss": 3.7345,
+      "step": 5588
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.756467593083094e-06,
+      "loss": 3.7769,
+      "step": 5589
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.754679963243926e-06,
+      "loss": 3.6488,
+      "step": 5590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.75289269283731e-06,
+      "loss": 3.6405,
+      "step": 5591
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.75110578214936e-06,
+      "loss": 3.6273,
+      "step": 5592
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7493192314661198e-06,
+      "loss": 3.643,
+      "step": 5593
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.747533041073588e-06,
+      "loss": 3.5581,
+      "step": 5594
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7457472112576912e-06,
+      "loss": 3.7137,
+      "step": 5595
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7439617423043146e-06,
+      "loss": 3.7042,
+      "step": 5596
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7421766344992708e-06,
+      "loss": 3.6954,
+      "step": 5597
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.740391888128327e-06,
+      "loss": 3.7098,
+      "step": 5598
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7386075034771804e-06,
+      "loss": 3.6042,
+      "step": 5599
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.736823480831482e-06,
+      "loss": 3.6896,
+      "step": 5600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7350398204768182e-06,
+      "loss": 3.6648,
+      "step": 5601
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7332565226987173e-06,
+      "loss": 3.8054,
+      "step": 5602
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7314735877826525e-06,
+      "loss": 3.7451,
+      "step": 5603
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7296910160140365e-06,
+      "loss": 3.6964,
+      "step": 5604
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7279088076782257e-06,
+      "loss": 3.5998,
+      "step": 5605
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.726126963060516e-06,
+      "loss": 3.6403,
+      "step": 5606
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.724345482446148e-06,
+      "loss": 3.6664,
+      "step": 5607
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7225643661203004e-06,
+      "loss": 3.7405,
+      "step": 5608
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7207836143680976e-06,
+      "loss": 3.7109,
+      "step": 5609
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7190032274746037e-06,
+      "loss": 3.6831,
+      "step": 5610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.717223205724823e-06,
+      "loss": 3.691,
+      "step": 5611
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7154435494037033e-06,
+      "loss": 3.6775,
+      "step": 5612
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7136642587961336e-06,
+      "loss": 3.5855,
+      "step": 5613
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7118853341869446e-06,
+      "loss": 3.6973,
+      "step": 5614
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7101067758609046e-06,
+      "loss": 3.7554,
+      "step": 5615
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.708328584102734e-06,
+      "loss": 3.6783,
+      "step": 5616
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.706550759197078e-06,
+      "loss": 3.772,
+      "step": 5617
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7047733014285406e-06,
+      "loss": 3.7432,
+      "step": 5618
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.702996211081651e-06,
+      "loss": 3.8034,
+      "step": 5619
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7012194884408942e-06,
+      "loss": 3.7776,
+      "step": 5620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6994431337906845e-06,
+      "loss": 3.6949,
+      "step": 5621
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.697667147415383e-06,
+      "loss": 3.7144,
+      "step": 5622
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6958915295992915e-06,
+      "loss": 3.6692,
+      "step": 5623
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6941162806266506e-06,
+      "loss": 3.7905,
+      "step": 5624
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6923414007816495e-06,
+      "loss": 3.7735,
+      "step": 5625
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6905668903484034e-06,
+      "loss": 3.8409,
+      "step": 5626
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6887927496109863e-06,
+      "loss": 3.7206,
+      "step": 5627
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6870189788533953e-06,
+      "loss": 3.6466,
+      "step": 5628
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6852455783595855e-06,
+      "loss": 3.6022,
+      "step": 5629
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.683472548413436e-06,
+      "loss": 3.6865,
+      "step": 5630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6816998892987824e-06,
+      "loss": 3.8019,
+      "step": 5631
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.679927601299386e-06,
+      "loss": 3.7763,
+      "step": 5632
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.678155684698962e-06,
+      "loss": 3.7096,
+      "step": 5633
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6763841397811576e-06,
+      "loss": 3.7595,
+      "step": 5634
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6746129668295637e-06,
+      "loss": 3.598,
+      "step": 5635
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6728421661277105e-06,
+      "loss": 3.6694,
+      "step": 5636
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6710717379590687e-06,
+      "loss": 3.6945,
+      "step": 5637
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6693016826070515e-06,
+      "loss": 3.7222,
+      "step": 5638
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6675320003550096e-06,
+      "loss": 3.7982,
+      "step": 5639
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.665762691486235e-06,
+      "loss": 3.6605,
+      "step": 5640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6639937562839602e-06,
+      "loss": 3.6703,
+      "step": 5641
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6622251950313573e-06,
+      "loss": 3.6985,
+      "step": 5642
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.66045700801154e-06,
+      "loss": 3.8349,
+      "step": 5643
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.65868919550756e-06,
+      "loss": 3.6182,
+      "step": 5644
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.65692175780241e-06,
+      "loss": 3.7305,
+      "step": 5645
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6551546951790243e-06,
+      "loss": 3.6831,
+      "step": 5646
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6533880079202734e-06,
+      "loss": 3.7479,
+      "step": 5647
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6516216963089698e-06,
+      "loss": 3.6634,
+      "step": 5648
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6498557606278706e-06,
+      "loss": 3.7405,
+      "step": 5649
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.648090201159662e-06,
+      "loss": 3.6376,
+      "step": 5650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.646325018186983e-06,
+      "loss": 3.6486,
+      "step": 5651
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6445602119923963e-06,
+      "loss": 3.7559,
+      "step": 5652
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6427957828584235e-06,
+      "loss": 3.6652,
+      "step": 5653
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6410317310675066e-06,
+      "loss": 3.7248,
+      "step": 5654
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6392680569020444e-06,
+      "loss": 3.647,
+      "step": 5655
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.637504760644359e-06,
+      "loss": 3.6101,
+      "step": 5656
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6357418425767278e-06,
+      "loss": 3.6171,
+      "step": 5657
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6339793029813565e-06,
+      "loss": 3.7341,
+      "step": 5658
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.632217142140394e-06,
+      "loss": 3.7311,
+      "step": 5659
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.630455360335929e-06,
+      "loss": 3.6217,
+      "step": 5660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.628693957849988e-06,
+      "loss": 3.6625,
+      "step": 5661
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6269329349645385e-06,
+      "loss": 3.7974,
+      "step": 5662
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.625172291961485e-06,
+      "loss": 3.6882,
+      "step": 5663
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.623412029122675e-06,
+      "loss": 3.7337,
+      "step": 5664
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6216521467298905e-06,
+      "loss": 3.7362,
+      "step": 5665
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.619892645064856e-06,
+      "loss": 3.7547,
+      "step": 5666
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.618133524409233e-06,
+      "loss": 3.8204,
+      "step": 5667
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.616374785044624e-06,
+      "loss": 3.6731,
+      "step": 5668
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6146164272525686e-06,
+      "loss": 3.6464,
+      "step": 5669
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6128584513145463e-06,
+      "loss": 3.6407,
+      "step": 5670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6111008575119746e-06,
+      "loss": 3.6028,
+      "step": 5671
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.60934364612621e-06,
+      "loss": 3.712,
+      "step": 5672
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6075868174385523e-06,
+      "loss": 3.7711,
+      "step": 5673
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.605830371730229e-06,
+      "loss": 3.7087,
+      "step": 5674
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6040743092824216e-06,
+      "loss": 3.6858,
+      "step": 5675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.602318630376234e-06,
+      "loss": 3.7441,
+      "step": 5676
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6005633352927234e-06,
+      "loss": 3.6733,
+      "step": 5677
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.598808424312872e-06,
+      "loss": 3.6434,
+      "step": 5678
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.597053897717615e-06,
+      "loss": 3.7533,
+      "step": 5679
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.59529975578781e-06,
+      "loss": 3.7349,
+      "step": 5680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.593545998804267e-06,
+      "loss": 3.6265,
+      "step": 5681
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5917926270477277e-06,
+      "loss": 3.5866,
+      "step": 5682
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5900396407988736e-06,
+      "loss": 3.6235,
+      "step": 5683
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.588287040338323e-06,
+      "loss": 3.6757,
+      "step": 5684
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.586534825946634e-06,
+      "loss": 3.6931,
+      "step": 5685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5847829979043015e-06,
+      "loss": 3.7264,
+      "step": 5686
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.583031556491761e-06,
+      "loss": 3.6325,
+      "step": 5687
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5812805019893837e-06,
+      "loss": 3.7605,
+      "step": 5688
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.57952983467748e-06,
+      "loss": 3.5533,
+      "step": 5689
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5777795548362984e-06,
+      "loss": 3.6564,
+      "step": 5690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5760296627460236e-06,
+      "loss": 3.6582,
+      "step": 5691
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.574280158686782e-06,
+      "loss": 3.6446,
+      "step": 5692
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.572531042938633e-06,
+      "loss": 3.6749,
+      "step": 5693
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.570782315781578e-06,
+      "loss": 3.7554,
+      "step": 5694
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5690339774955543e-06,
+      "loss": 3.5986,
+      "step": 5695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.567286028360435e-06,
+      "loss": 3.5051,
+      "step": 5696
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.56553846865604e-06,
+      "loss": 3.7175,
+      "step": 5697
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5637912986621103e-06,
+      "loss": 3.6848,
+      "step": 5698
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.562044518658344e-06,
+      "loss": 3.7144,
+      "step": 5699
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.560298128924358e-06,
+      "loss": 3.604,
+      "step": 5700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5585521297397243e-06,
+      "loss": 3.7501,
+      "step": 5701
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5568065213839343e-06,
+      "loss": 3.7449,
+      "step": 5702
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5550613041364364e-06,
+      "loss": 3.7296,
+      "step": 5703
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5533164782765974e-06,
+      "loss": 3.7195,
+      "step": 5704
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5515720440837356e-06,
+      "loss": 3.696,
+      "step": 5705
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5498280018371e-06,
+      "loss": 3.741,
+      "step": 5706
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.548084351815879e-06,
+      "loss": 3.6094,
+      "step": 5707
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5463410942991986e-06,
+      "loss": 3.5865,
+      "step": 5708
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.544598229566114e-06,
+      "loss": 3.6816,
+      "step": 5709
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5428557578956337e-06,
+      "loss": 3.6327,
+      "step": 5710
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5411136795666857e-06,
+      "loss": 3.7417,
+      "step": 5711
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5393719948581507e-06,
+      "loss": 3.7063,
+      "step": 5712
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5376307040488324e-06,
+      "loss": 3.6601,
+      "step": 5713
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5358898074174815e-06,
+      "loss": 3.8229,
+      "step": 5714
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5341493052427813e-06,
+      "loss": 3.7621,
+      "step": 5715
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.532409197803353e-06,
+      "loss": 3.6361,
+      "step": 5716
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5306694853777546e-06,
+      "loss": 3.7361,
+      "step": 5717
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.52893016824448e-06,
+      "loss": 3.6515,
+      "step": 5718
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5271912466819597e-06,
+      "loss": 3.6072,
+      "step": 5719
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.525452720968563e-06,
+      "loss": 3.7564,
+      "step": 5720
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.523714591382594e-06,
+      "loss": 3.6226,
+      "step": 5721
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5219768582022934e-06,
+      "loss": 3.6743,
+      "step": 5722
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5202395217058395e-06,
+      "loss": 3.6728,
+      "step": 5723
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5185025821713454e-06,
+      "loss": 3.6964,
+      "step": 5724
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.516766039876863e-06,
+      "loss": 3.7085,
+      "step": 5725
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5150298951003783e-06,
+      "loss": 3.6953,
+      "step": 5726
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.513294148119816e-06,
+      "loss": 3.6373,
+      "step": 5727
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5115587992130342e-06,
+      "loss": 3.6646,
+      "step": 5728
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5098238486578276e-06,
+      "loss": 3.7848,
+      "step": 5729
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.508089296731934e-06,
+      "loss": 3.6306,
+      "step": 5730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5063551437130153e-06,
+      "loss": 3.6374,
+      "step": 5731
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.504621389878682e-06,
+      "loss": 3.7131,
+      "step": 5732
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5028880355064682e-06,
+      "loss": 3.5839,
+      "step": 5733
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5011550808738585e-06,
+      "loss": 3.7428,
+      "step": 5734
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.499422526258256e-06,
+      "loss": 3.7424,
+      "step": 5735
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4976903719370193e-06,
+      "loss": 3.7122,
+      "step": 5736
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4959586181874246e-06,
+      "loss": 3.7368,
+      "step": 5737
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4942272652866977e-06,
+      "loss": 3.7254,
+      "step": 5738
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4924963135119932e-06,
+      "loss": 3.5485,
+      "step": 5739
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4907657631404037e-06,
+      "loss": 3.7643,
+      "step": 5740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4890356144489563e-06,
+      "loss": 3.6699,
+      "step": 5741
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.487305867714615e-06,
+      "loss": 3.7219,
+      "step": 5742
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4855765232142786e-06,
+      "loss": 3.7423,
+      "step": 5743
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.483847581224782e-06,
+      "loss": 3.5755,
+      "step": 5744
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.482119042022896e-06,
+      "loss": 3.7251,
+      "step": 5745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.480390905885326e-06,
+      "loss": 3.6668,
+      "step": 5746
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4786631730887134e-06,
+      "loss": 3.6014,
+      "step": 5747
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4769358439096347e-06,
+      "loss": 3.7149,
+      "step": 5748
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.475208918624603e-06,
+      "loss": 3.654,
+      "step": 5749
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.473482397510065e-06,
+      "loss": 3.6596,
+      "step": 5750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.471756280842404e-06,
+      "loss": 3.6507,
+      "step": 5751
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.470030568897938e-06,
+      "loss": 3.6491,
+      "step": 5752
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.468305261952918e-06,
+      "loss": 3.7104,
+      "step": 5753
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4665803602835385e-06,
+      "loss": 3.607,
+      "step": 5754
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.464855864165916e-06,
+      "loss": 3.6617,
+      "step": 5755
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4631317738761155e-06,
+      "loss": 3.6689,
+      "step": 5756
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4614080896901238e-06,
+      "loss": 3.6779,
+      "step": 5757
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.459684811883877e-06,
+      "loss": 3.8016,
+      "step": 5758
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4579619407332324e-06,
+      "loss": 3.6069,
+      "step": 5759
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.456239476513994e-06,
+      "loss": 3.7124,
+      "step": 5760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4545174195018888e-06,
+      "loss": 3.6407,
+      "step": 5761
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4527957699725914e-06,
+      "loss": 3.6464,
+      "step": 5762
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.451074528201701e-06,
+      "loss": 3.7034,
+      "step": 5763
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4493536944647566e-06,
+      "loss": 3.6327,
+      "step": 5764
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4476332690372304e-06,
+      "loss": 3.6975,
+      "step": 5765
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4459132521945295e-06,
+      "loss": 3.593,
+      "step": 5766
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.444193644211995e-06,
+      "loss": 3.5821,
+      "step": 5767
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.442474445364904e-06,
+      "loss": 3.7434,
+      "step": 5768
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.440755655928466e-06,
+      "loss": 3.6996,
+      "step": 5769
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.439037276177827e-06,
+      "loss": 3.6935,
+      "step": 5770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4373193063880663e-06,
+      "loss": 3.6289,
+      "step": 5771
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4356017468341977e-06,
+      "loss": 3.6148,
+      "step": 5772
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4338845977911695e-06,
+      "loss": 3.5666,
+      "step": 5773
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4321678595338643e-06,
+      "loss": 3.6768,
+      "step": 5774
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.430451532337099e-06,
+      "loss": 3.7633,
+      "step": 5775
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4287356164756244e-06,
+      "loss": 3.6386,
+      "step": 5776
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.427020112224124e-06,
+      "loss": 3.5691,
+      "step": 5777
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.425305019857222e-06,
+      "loss": 3.6587,
+      "step": 5778
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.423590339649465e-06,
+      "loss": 3.6798,
+      "step": 5779
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.421876071875347e-06,
+      "loss": 3.6619,
+      "step": 5780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4201622168092827e-06,
+      "loss": 3.6467,
+      "step": 5781
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4184487747256346e-06,
+      "loss": 3.7139,
+      "step": 5782
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4167357458986834e-06,
+      "loss": 3.6096,
+      "step": 5783
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.415023130602661e-06,
+      "loss": 3.7094,
+      "step": 5784
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4133109291117156e-06,
+      "loss": 3.5979,
+      "step": 5785
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.411599141699944e-06,
+      "loss": 3.7848,
+      "step": 5786
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4098877686413684e-06,
+      "loss": 3.7051,
+      "step": 5787
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.408176810209946e-06,
+      "loss": 3.7099,
+      "step": 5788
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4064662666795694e-06,
+      "loss": 3.759,
+      "step": 5789
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4047561383240635e-06,
+      "loss": 3.7707,
+      "step": 5790
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4030464254171865e-06,
+      "loss": 3.6327,
+      "step": 5791
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.401337128232631e-06,
+      "loss": 3.6992,
+      "step": 5792
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3996282470440236e-06,
+      "loss": 3.689,
+      "step": 5793
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.397919782124919e-06,
+      "loss": 3.7256,
+      "step": 5794
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3962117337488137e-06,
+      "loss": 3.8035,
+      "step": 5795
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3945041021891335e-06,
+      "loss": 3.6165,
+      "step": 5796
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.392796887719235e-06,
+      "loss": 3.6597,
+      "step": 5797
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.391090090612413e-06,
+      "loss": 3.7629,
+      "step": 5798
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.389383711141891e-06,
+      "loss": 3.8035,
+      "step": 5799
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.387677749580828e-06,
+      "loss": 3.6149,
+      "step": 5800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3859722062023154e-06,
+      "loss": 3.6507,
+      "step": 5801
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.384267081279379e-06,
+      "loss": 3.6083,
+      "step": 5802
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3825623750849746e-06,
+      "loss": 3.6454,
+      "step": 5803
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3808580878919948e-06,
+      "loss": 3.6631,
+      "step": 5804
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.379154219973262e-06,
+      "loss": 3.6455,
+      "step": 5805
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.377450771601533e-06,
+      "loss": 3.6118,
+      "step": 5806
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.375747743049498e-06,
+      "loss": 3.6675,
+      "step": 5807
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3740451345897773e-06,
+      "loss": 3.8233,
+      "step": 5808
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3723429464949276e-06,
+      "loss": 3.554,
+      "step": 5809
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3706411790374336e-06,
+      "loss": 3.6759,
+      "step": 5810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3689398324897205e-06,
+      "loss": 3.6794,
+      "step": 5811
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3672389071241354e-06,
+      "loss": 3.7492,
+      "step": 5812
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.36553840321297e-06,
+      "loss": 3.6726,
+      "step": 5813
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3638383210284354e-06,
+      "loss": 3.6283,
+      "step": 5814
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.362138660842689e-06,
+      "loss": 3.729,
+      "step": 5815
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3604394229278064e-06,
+      "loss": 3.7925,
+      "step": 5816
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.358740607555811e-06,
+      "loss": 3.7338,
+      "step": 5817
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.357042214998642e-06,
+      "loss": 3.5455,
+      "step": 5818
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3553442455281856e-06,
+      "loss": 3.77,
+      "step": 5819
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3536466994162522e-06,
+      "loss": 3.6853,
+      "step": 5820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.351949576934586e-06,
+      "loss": 3.6758,
+      "step": 5821
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3502528783548637e-06,
+      "loss": 3.6486,
+      "step": 5822
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3485566039486948e-06,
+      "loss": 3.7774,
+      "step": 5823
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3468607539876186e-06,
+      "loss": 3.6509,
+      "step": 5824
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3451653287431096e-06,
+      "loss": 3.5566,
+      "step": 5825
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.343470328486572e-06,
+      "loss": 3.701,
+      "step": 5826
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.341775753489344e-06,
+      "loss": 3.7321,
+      "step": 5827
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3400816040226925e-06,
+      "loss": 3.7973,
+      "step": 5828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3383878803578193e-06,
+      "loss": 3.707,
+      "step": 5829
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.336694582765857e-06,
+      "loss": 3.7303,
+      "step": 5830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.33500171151787e-06,
+      "loss": 3.6706,
+      "step": 5831
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3333092668848544e-06,
+      "loss": 3.6925,
+      "step": 5832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.331617249137738e-06,
+      "loss": 3.6516,
+      "step": 5833
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3299256585473777e-06,
+      "loss": 3.6258,
+      "step": 5834
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3282344953845715e-06,
+      "loss": 3.7107,
+      "step": 5835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.326543759920034e-06,
+      "loss": 3.6978,
+      "step": 5836
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.324853452424427e-06,
+      "loss": 3.7346,
+      "step": 5837
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3231635731683283e-06,
+      "loss": 3.7035,
+      "step": 5838
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.321474122422263e-06,
+      "loss": 3.752,
+      "step": 5839
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3197851004566723e-06,
+      "loss": 3.679,
+      "step": 5840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3180965075419436e-06,
+      "loss": 3.5956,
+      "step": 5841
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3164083439483803e-06,
+      "loss": 3.686,
+      "step": 5842
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3147206099462313e-06,
+      "loss": 3.6563,
+      "step": 5843
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.313033305805667e-06,
+      "loss": 3.6374,
+      "step": 5844
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.311346431796793e-06,
+      "loss": 3.6865,
+      "step": 5845
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3096599881896462e-06,
+      "loss": 3.6959,
+      "step": 5846
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3079739752541936e-06,
+      "loss": 3.6091,
+      "step": 5847
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3062883932603326e-06,
+      "loss": 3.7391,
+      "step": 5848
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.304603242477893e-06,
+      "loss": 3.7677,
+      "step": 5849
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3029185231766356e-06,
+      "loss": 3.8101,
+      "step": 5850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3012342356262503e-06,
+      "loss": 3.7592,
+      "step": 5851
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2995503800963593e-06,
+      "loss": 3.5999,
+      "step": 5852
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2978669568565164e-06,
+      "loss": 3.6084,
+      "step": 5853
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2961839661762047e-06,
+      "loss": 3.7328,
+      "step": 5854
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.294501408324839e-06,
+      "loss": 3.5856,
+      "step": 5855
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2928192835717642e-06,
+      "loss": 3.6637,
+      "step": 5856
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2911375921862566e-06,
+      "loss": 3.6759,
+      "step": 5857
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2894563344375197e-06,
+      "loss": 3.6985,
+      "step": 5858
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.287775510594697e-06,
+      "loss": 3.638,
+      "step": 5859
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2860951209268485e-06,
+      "loss": 3.7305,
+      "step": 5860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.284415165702979e-06,
+      "loss": 3.6104,
+      "step": 5861
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2827356451920107e-06,
+      "loss": 3.6862,
+      "step": 5862
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2810565596628093e-06,
+      "loss": 3.7126,
+      "step": 5863
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2793779093841565e-06,
+      "loss": 3.6055,
+      "step": 5864
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2776996946247797e-06,
+      "loss": 3.6807,
+      "step": 5865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.276021915653322e-06,
+      "loss": 3.7039,
+      "step": 5866
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.274344572738368e-06,
+      "loss": 3.6667,
+      "step": 5867
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2726676661484265e-06,
+      "loss": 3.6036,
+      "step": 5868
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.270991196151938e-06,
+      "loss": 3.6734,
+      "step": 5869
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.269315163017274e-06,
+      "loss": 3.7197,
+      "step": 5870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2676395670127336e-06,
+      "loss": 3.6679,
+      "step": 5871
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2659644084065483e-06,
+      "loss": 3.8214,
+      "step": 5872
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2642896874668795e-06,
+      "loss": 3.6924,
+      "step": 5873
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2626154044618165e-06,
+      "loss": 3.6568,
+      "step": 5874
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2609415596593808e-06,
+      "loss": 3.7311,
+      "step": 5875
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2592681533275223e-06,
+      "loss": 3.6362,
+      "step": 5876
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.257595185734121e-06,
+      "loss": 3.6274,
+      "step": 5877
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.255922657146987e-06,
+      "loss": 3.7327,
+      "step": 5878
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.25425056783386e-06,
+      "loss": 3.7251,
+      "step": 5879
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2525789180624096e-06,
+      "loss": 3.7383,
+      "step": 5880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2509077081002344e-06,
+      "loss": 3.7136,
+      "step": 5881
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2492369382148634e-06,
+      "loss": 3.8025,
+      "step": 5882
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.247566608673754e-06,
+      "loss": 3.5739,
+      "step": 5883
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2458967197442942e-06,
+      "loss": 3.6942,
+      "step": 5884
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2442272716938017e-06,
+      "loss": 3.5977,
+      "step": 5885
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.242558264789522e-06,
+      "loss": 3.8039,
+      "step": 5886
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2408896992986325e-06,
+      "loss": 3.6947,
+      "step": 5887
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2392215754882363e-06,
+      "loss": 3.6461,
+      "step": 5888
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.23755389362537e-06,
+      "loss": 3.6137,
+      "step": 5889
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2358866539769956e-06,
+      "loss": 3.6455,
+      "step": 5890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2342198568100055e-06,
+      "loss": 3.7197,
+      "step": 5891
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.232553502391227e-06,
+      "loss": 3.7265,
+      "step": 5892
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2308875909874043e-06,
+      "loss": 3.7153,
+      "step": 5893
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.229222122865225e-06,
+      "loss": 3.7581,
+      "step": 5894
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.227557098291291e-06,
+      "loss": 3.7003,
+      "step": 5895
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2258925175321476e-06,
+      "loss": 3.7332,
+      "step": 5896
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.224228380854256e-06,
+      "loss": 3.7933,
+      "step": 5897
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2225646885240186e-06,
+      "loss": 3.6581,
+      "step": 5898
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.220901440807754e-06,
+      "loss": 3.6396,
+      "step": 5899
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2192386379717214e-06,
+      "loss": 3.7073,
+      "step": 5900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2175762802821015e-06,
+      "loss": 3.6404,
+      "step": 5901
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2159143680050067e-06,
+      "loss": 3.6431,
+      "step": 5902
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.214252901406476e-06,
+      "loss": 3.7251,
+      "step": 5903
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2125918807524795e-06,
+      "loss": 3.6492,
+      "step": 5904
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2109313063089135e-06,
+      "loss": 3.6903,
+      "step": 5905
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.209271178341604e-06,
+      "loss": 3.749,
+      "step": 5906
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.207611497116307e-06,
+      "loss": 3.6142,
+      "step": 5907
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.205952262898704e-06,
+      "loss": 3.6259,
+      "step": 5908
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.204293475954408e-06,
+      "loss": 3.7107,
+      "step": 5909
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.202635136548957e-06,
+      "loss": 3.7064,
+      "step": 5910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.20097724494782e-06,
+      "loss": 3.7399,
+      "step": 5911
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.199319801416394e-06,
+      "loss": 3.7117,
+      "step": 5912
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.197662806220004e-06,
+      "loss": 3.6059,
+      "step": 5913
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.196006259623903e-06,
+      "loss": 3.6845,
+      "step": 5914
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1943501618932696e-06,
+      "loss": 3.7544,
+      "step": 5915
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1926945132932188e-06,
+      "loss": 3.6748,
+      "step": 5916
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1910393140887813e-06,
+      "loss": 3.7334,
+      "step": 5917
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1893845645449302e-06,
+      "loss": 3.7016,
+      "step": 5918
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1877302649265515e-06,
+      "loss": 3.6212,
+      "step": 5919
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1860764154984736e-06,
+      "loss": 3.6478,
+      "step": 5920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1844230165254394e-06,
+      "loss": 3.6465,
+      "step": 5921
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1827700682721326e-06,
+      "loss": 3.701,
+      "step": 5922
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1811175710031523e-06,
+      "loss": 3.6702,
+      "step": 5923
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.179465524983036e-06,
+      "loss": 3.6253,
+      "step": 5924
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1778139304762435e-06,
+      "loss": 3.6437,
+      "step": 5925
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1761627877471625e-06,
+      "loss": 3.5598,
+      "step": 5926
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1745120970601104e-06,
+      "loss": 3.6596,
+      "step": 5927
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1728618586793292e-06,
+      "loss": 3.6233,
+      "step": 5928
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.171212072868992e-06,
+      "loss": 3.6897,
+      "step": 5929
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1695627398931975e-06,
+      "loss": 3.626,
+      "step": 5930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1679138600159716e-06,
+      "loss": 3.7148,
+      "step": 5931
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1662654335012684e-06,
+      "loss": 3.6773,
+      "step": 5932
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1646174606129696e-06,
+      "loss": 3.68,
+      "step": 5933
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1629699416148832e-06,
+      "loss": 3.7097,
+      "step": 5934
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.161322876770746e-06,
+      "loss": 3.6698,
+      "step": 5935
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.159676266344222e-06,
+      "loss": 3.6619,
+      "step": 5936
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1580301105989002e-06,
+      "loss": 3.7099,
+      "step": 5937
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1563844097983e-06,
+      "loss": 3.7717,
+      "step": 5938
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1547391642058642e-06,
+      "loss": 3.7083,
+      "step": 5939
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1530943740849696e-06,
+      "loss": 3.6916,
+      "step": 5940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.151450039698909e-06,
+      "loss": 3.6758,
+      "step": 5941
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.149806161310916e-06,
+      "loss": 3.6125,
+      "step": 5942
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1481627391841366e-06,
+      "loss": 3.7907,
+      "step": 5943
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1465197735816585e-06,
+      "loss": 3.7142,
+      "step": 5944
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1448772647664806e-06,
+      "loss": 3.6234,
+      "step": 5945
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1432352130015456e-06,
+      "loss": 3.78,
+      "step": 5946
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.141593618549706e-06,
+      "loss": 3.732,
+      "step": 5947
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.139952481673756e-06,
+      "loss": 3.6213,
+      "step": 5948
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1383118026364074e-06,
+      "loss": 3.659,
+      "step": 5949
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.136671581700302e-06,
+      "loss": 3.7174,
+      "step": 5950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.135031819128007e-06,
+      "loss": 3.5874,
+      "step": 5951
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1333925151820175e-06,
+      "loss": 3.6684,
+      "step": 5952
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.131753670124753e-06,
+      "loss": 3.7258,
+      "step": 5953
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1301152842185634e-06,
+      "loss": 3.753,
+      "step": 5954
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1284773577257217e-06,
+      "loss": 3.7254,
+      "step": 5955
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.126839890908428e-06,
+      "loss": 3.7877,
+      "step": 5956
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1252028840288096e-06,
+      "loss": 3.6691,
+      "step": 5957
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.12356633734892e-06,
+      "loss": 3.7431,
+      "step": 5958
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1219302511307384e-06,
+      "loss": 3.7788,
+      "step": 5959
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.120294625636171e-06,
+      "loss": 3.6192,
+      "step": 5960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1186594611270507e-06,
+      "loss": 3.5989,
+      "step": 5961
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.117024757865134e-06,
+      "loss": 3.5291,
+      "step": 5962
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1153905161121078e-06,
+      "loss": 3.6845,
+      "step": 5963
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.113756736129581e-06,
+      "loss": 3.6765,
+      "step": 5964
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1121234181790912e-06,
+      "loss": 3.8205,
+      "step": 5965
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1104905625221008e-06,
+      "loss": 3.6125,
+      "step": 5966
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1088581694199988e-06,
+      "loss": 3.6794,
+      "step": 5967
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1072262391340996e-06,
+      "loss": 3.7186,
+      "step": 5968
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.105594771925645e-06,
+      "loss": 3.7153,
+      "step": 5969
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1039637680557995e-06,
+      "loss": 3.6304,
+      "step": 5970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1023332277856575e-06,
+      "loss": 3.7231,
+      "step": 5971
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.100703151376234e-06,
+      "loss": 3.6601,
+      "step": 5972
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0990735390884787e-06,
+      "loss": 3.6785,
+      "step": 5973
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.097444391183254e-06,
+      "loss": 3.7261,
+      "step": 5974
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0958157079213624e-06,
+      "loss": 3.6858,
+      "step": 5975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0941874895635184e-06,
+      "loss": 3.687,
+      "step": 5976
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0925597363703738e-06,
+      "loss": 3.7681,
+      "step": 5977
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0909324486024947e-06,
+      "loss": 3.7659,
+      "step": 5978
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.089305626520385e-06,
+      "loss": 3.6877,
+      "step": 5979
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.087679270384461e-06,
+      "loss": 3.6493,
+      "step": 5980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.086053380455077e-06,
+      "loss": 3.7126,
+      "step": 5981
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0844279569925037e-06,
+      "loss": 3.7597,
+      "step": 5982
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0828030002569403e-06,
+      "loss": 3.6428,
+      "step": 5983
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.081178510508512e-06,
+      "loss": 3.7164,
+      "step": 5984
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0795544880072676e-06,
+      "loss": 3.7676,
+      "step": 5985
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.077930933013182e-06,
+      "loss": 3.8914,
+      "step": 5986
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0763078457861553e-06,
+      "loss": 3.676,
+      "step": 5987
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0746852265860123e-06,
+      "loss": 3.6417,
+      "step": 5988
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.073063075672504e-06,
+      "loss": 3.7372,
+      "step": 5989
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.071441393305304e-06,
+      "loss": 3.7022,
+      "step": 5990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0698201797440135e-06,
+      "loss": 3.7063,
+      "step": 5991
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0681994352481575e-06,
+      "loss": 3.6502,
+      "step": 5992
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0665791600771868e-06,
+      "loss": 3.5974,
+      "step": 5993
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0649593544904743e-06,
+      "loss": 3.7294,
+      "step": 5994
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0633400187473214e-06,
+      "loss": 3.7488,
+      "step": 5995
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0617211531069503e-06,
+      "loss": 3.6909,
+      "step": 5996
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0601027578285152e-06,
+      "loss": 3.658,
+      "step": 5997
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.058484833171084e-06,
+      "loss": 3.655,
+      "step": 5998
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.056867379393661e-06,
+      "loss": 3.6591,
+      "step": 5999
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0552503967551634e-06,
+      "loss": 3.6845,
+      "step": 6000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.053633885514446e-06,
+      "loss": 3.785,
+      "step": 6001
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.052017845930274e-06,
+      "loss": 3.6705,
+      "step": 6002
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.050402278261351e-06,
+      "loss": 3.7067,
+      "step": 6003
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0487871827662913e-06,
+      "loss": 3.7192,
+      "step": 6004
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.047172559703646e-06,
+      "loss": 3.674,
+      "step": 6005
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.045558409331884e-06,
+      "loss": 3.7633,
+      "step": 6006
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0439447319093993e-06,
+      "loss": 3.6649,
+      "step": 6007
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0423315276945103e-06,
+      "loss": 3.6962,
+      "step": 6008
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0407187969454605e-06,
+      "loss": 3.6279,
+      "step": 6009
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0391065399204163e-06,
+      "loss": 3.823,
+      "step": 6010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.03749475687747e-06,
+      "loss": 3.6922,
+      "step": 6011
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0358834480746363e-06,
+      "loss": 3.6241,
+      "step": 6012
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0342726137698553e-06,
+      "loss": 3.6802,
+      "step": 6013
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0326622542209906e-06,
+      "loss": 3.6722,
+      "step": 6014
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0310523696858297e-06,
+      "loss": 3.6343,
+      "step": 6015
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0294429604220833e-06,
+      "loss": 3.6283,
+      "step": 6016
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.027834026687388e-06,
+      "loss": 3.8184,
+      "step": 6017
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.026225568739303e-06,
+      "loss": 3.7067,
+      "step": 6018
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0246175868353106e-06,
+      "loss": 3.7735,
+      "step": 6019
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0230100812328167e-06,
+      "loss": 3.7621,
+      "step": 6020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.021403052189158e-06,
+      "loss": 3.7523,
+      "step": 6021
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0197964999615804e-06,
+      "loss": 3.7626,
+      "step": 6022
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0181904248072703e-06,
+      "loss": 3.6241,
+      "step": 6023
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0165848269833215e-06,
+      "loss": 3.665,
+      "step": 6024
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0149797067467667e-06,
+      "loss": 3.6051,
+      "step": 6025
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0133750643545488e-06,
+      "loss": 3.6437,
+      "step": 6026
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.011770900063545e-06,
+      "loss": 3.7059,
+      "step": 6027
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.010167214130546e-06,
+      "loss": 3.72,
+      "step": 6028
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.008564006812275e-06,
+      "loss": 3.6251,
+      "step": 6029
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.006961278365375e-06,
+      "loss": 3.6369,
+      "step": 6030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.005359029046409e-06,
+      "loss": 3.6391,
+      "step": 6031
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.003757259111869e-06,
+      "loss": 3.7135,
+      "step": 6032
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0021559688181665e-06,
+      "loss": 3.7375,
+      "step": 6033
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0005551584216366e-06,
+      "loss": 3.6149,
+      "step": 6034
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.998954828178539e-06,
+      "loss": 3.7346,
+      "step": 6035
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9973549783450563e-06,
+      "loss": 3.6068,
+      "step": 6036
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9957556091772923e-06,
+      "loss": 3.7269,
+      "step": 6037
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9941567209312768e-06,
+      "loss": 3.6515,
+      "step": 6038
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9925583138629594e-06,
+      "loss": 3.746,
+      "step": 6039
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9909603882282156e-06,
+      "loss": 3.5233,
+      "step": 6040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9893629442828427e-06,
+      "loss": 3.6507,
+      "step": 6041
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.98776598228256e-06,
+      "loss": 3.6949,
+      "step": 6042
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9861695024830102e-06,
+      "loss": 3.7019,
+      "step": 6043
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.984573505139758e-06,
+      "loss": 3.7683,
+      "step": 6044
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9829779905082967e-06,
+      "loss": 3.7427,
+      "step": 6045
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.981382958844031e-06,
+      "loss": 3.6504,
+      "step": 6046
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9797884104023014e-06,
+      "loss": 3.6254,
+      "step": 6047
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9781943454383583e-06,
+      "loss": 3.5901,
+      "step": 6048
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9766007642073842e-06,
+      "loss": 3.7949,
+      "step": 6049
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.97500766696448e-06,
+      "loss": 3.6947,
+      "step": 6050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.973415053964671e-06,
+      "loss": 3.6343,
+      "step": 6051
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9718229254629018e-06,
+      "loss": 3.7476,
+      "step": 6052
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9702312817140412e-06,
+      "loss": 3.6267,
+      "step": 6053
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9686401229728857e-06,
+      "loss": 3.6687,
+      "step": 6054
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.967049449494142e-06,
+      "loss": 3.6491,
+      "step": 6055
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9654592615324542e-06,
+      "loss": 3.6498,
+      "step": 6056
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.963869559342373e-06,
+      "loss": 3.6604,
+      "step": 6057
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9622803431783866e-06,
+      "loss": 3.7277,
+      "step": 6058
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.96069161329489e-06,
+      "loss": 3.6922,
+      "step": 6059
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9591033699462165e-06,
+      "loss": 3.5695,
+      "step": 6060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.957515613386606e-06,
+      "loss": 3.7217,
+      "step": 6061
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9559283438702325e-06,
+      "loss": 3.6717,
+      "step": 6062
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.954341561651186e-06,
+      "loss": 3.639,
+      "step": 6063
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9527552669834797e-06,
+      "loss": 3.7835,
+      "step": 6064
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.951169460121049e-06,
+      "loss": 3.6965,
+      "step": 6065
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.949584141317751e-06,
+      "loss": 3.7653,
+      "step": 6066
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.947999310827365e-06,
+      "loss": 3.6577,
+      "step": 6067
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9464149689035912e-06,
+      "loss": 3.7186,
+      "step": 6068
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9448311158000526e-06,
+      "loss": 3.6649,
+      "step": 6069
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9432477517702943e-06,
+      "loss": 3.7355,
+      "step": 6070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9416648770677816e-06,
+      "loss": 3.7139,
+      "step": 6071
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.940082491945902e-06,
+      "loss": 3.7585,
+      "step": 6072
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.938500596657966e-06,
+      "loss": 3.6334,
+      "step": 6073
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9369191914572037e-06,
+      "loss": 3.8038,
+      "step": 6074
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.935338276596769e-06,
+      "loss": 3.6359,
+      "step": 6075
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.933757852329734e-06,
+      "loss": 3.678,
+      "step": 6076
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.932177918909094e-06,
+      "loss": 3.7612,
+      "step": 6077
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.93059847658777e-06,
+      "loss": 3.753,
+      "step": 6078
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9290195256185946e-06,
+      "loss": 3.628,
+      "step": 6079
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.927441066254334e-06,
+      "loss": 3.7419,
+      "step": 6080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.925863098747662e-06,
+      "loss": 3.6679,
+      "step": 6081
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9242856233511875e-06,
+      "loss": 3.7166,
+      "step": 6082
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.922708640317428e-06,
+      "loss": 3.6339,
+      "step": 6083
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.921132149898834e-06,
+      "loss": 3.6801,
+      "step": 6084
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.919556152347765e-06,
+      "loss": 3.5947,
+      "step": 6085
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9179806479165137e-06,
+      "loss": 3.7679,
+      "step": 6086
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9164056368572847e-06,
+      "loss": 3.6992,
+      "step": 6087
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9148311194222084e-06,
+      "loss": 3.6426,
+      "step": 6088
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.913257095863334e-06,
+      "loss": 3.7525,
+      "step": 6089
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.911683566432633e-06,
+      "loss": 3.8098,
+      "step": 6090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.910110531381997e-06,
+      "loss": 3.663,
+      "step": 6091
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.908537990963238e-06,
+      "loss": 3.6319,
+      "step": 6092
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9069659454280898e-06,
+      "loss": 3.7167,
+      "step": 6093
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9053943950282071e-06,
+      "loss": 3.6617,
+      "step": 6094
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9038233400151651e-06,
+      "loss": 3.7019,
+      "step": 6095
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9022527806404583e-06,
+      "loss": 3.7445,
+      "step": 6096
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9006827171555042e-06,
+      "loss": 3.6761,
+      "step": 6097
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8991131498116394e-06,
+      "loss": 3.5913,
+      "step": 6098
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8975440788601214e-06,
+      "loss": 3.726,
+      "step": 6099
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8959755045521283e-06,
+      "loss": 3.776,
+      "step": 6100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8944074271387565e-06,
+      "loss": 3.679,
+      "step": 6101
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8928398468710313e-06,
+      "loss": 3.6855,
+      "step": 6102
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.891272763999884e-06,
+      "loss": 3.6311,
+      "step": 6103
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8897061787761823e-06,
+      "loss": 3.7052,
+      "step": 6104
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8881400914506986e-06,
+      "loss": 3.6598,
+      "step": 6105
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.886574502274141e-06,
+      "loss": 3.5905,
+      "step": 6106
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8850094114971235e-06,
+      "loss": 3.7154,
+      "step": 6107
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.883444819370193e-06,
+      "loss": 3.7171,
+      "step": 6108
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8818807261438043e-06,
+      "loss": 3.7571,
+      "step": 6109
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8803171320683438e-06,
+      "loss": 3.6919,
+      "step": 6110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.878754037394111e-06,
+      "loss": 3.6976,
+      "step": 6111
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8771914423713277e-06,
+      "loss": 3.7211,
+      "step": 6112
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8756293472501347e-06,
+      "loss": 3.765,
+      "step": 6113
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8740677522805933e-06,
+      "loss": 3.7227,
+      "step": 6114
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8725066577126854e-06,
+      "loss": 3.6734,
+      "step": 6115
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8709460637963123e-06,
+      "loss": 3.6979,
+      "step": 6116
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8693859707812945e-06,
+      "loss": 3.7603,
+      "step": 6117
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8678263789173728e-06,
+      "loss": 3.6099,
+      "step": 6118
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8662672884542082e-06,
+      "loss": 3.6726,
+      "step": 6119
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8647086996413816e-06,
+      "loss": 3.6046,
+      "step": 6120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8631506127283922e-06,
+      "loss": 3.6964,
+      "step": 6121
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8615930279646605e-06,
+      "loss": 3.7394,
+      "step": 6122
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8600359455995249e-06,
+      "loss": 3.657,
+      "step": 6123
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8584793658822452e-06,
+      "loss": 3.6699,
+      "step": 6124
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8569232890619982e-06,
+      "loss": 3.7069,
+      "step": 6125
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8553677153878864e-06,
+      "loss": 3.751,
+      "step": 6126
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8538126451089206e-06,
+      "loss": 3.6052,
+      "step": 6127
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8522580784740452e-06,
+      "loss": 3.6134,
+      "step": 6128
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8507040157321081e-06,
+      "loss": 3.6916,
+      "step": 6129
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.849150457131893e-06,
+      "loss": 3.7493,
+      "step": 6130
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8475974029220873e-06,
+      "loss": 3.6851,
+      "step": 6131
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8460448533513126e-06,
+      "loss": 3.7229,
+      "step": 6132
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8444928086680958e-06,
+      "loss": 3.6326,
+      "step": 6133
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.842941269120891e-06,
+      "loss": 3.6316,
+      "step": 6134
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8413902349580737e-06,
+      "loss": 3.663,
+      "step": 6135
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8398397064279282e-06,
+      "loss": 3.7845,
+      "step": 6136
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8382896837786719e-06,
+      "loss": 3.6875,
+      "step": 6137
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8367401672584256e-06,
+      "loss": 3.6681,
+      "step": 6138
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8351911571152442e-06,
+      "loss": 3.6978,
+      "step": 6139
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.833642653597088e-06,
+      "loss": 3.6324,
+      "step": 6140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8320946569518493e-06,
+      "loss": 3.6931,
+      "step": 6141
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8305471674273262e-06,
+      "loss": 3.7266,
+      "step": 6142
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8290001852712463e-06,
+      "loss": 3.6512,
+      "step": 6143
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8274537107312502e-06,
+      "loss": 3.7908,
+      "step": 6144
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8259077440548988e-06,
+      "loss": 3.6614,
+      "step": 6145
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8243622854896724e-06,
+      "loss": 3.6566,
+      "step": 6146
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8228173352829681e-06,
+      "loss": 3.6756,
+      "step": 6147
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8212728936821034e-06,
+      "loss": 3.6266,
+      "step": 6148
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8197289609343126e-06,
+      "loss": 3.8275,
+      "step": 6149
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8181855372867513e-06,
+      "loss": 3.7039,
+      "step": 6150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.816642622986491e-06,
+      "loss": 3.8312,
+      "step": 6151
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8151002182805226e-06,
+      "loss": 3.7943,
+      "step": 6152
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8135583234157555e-06,
+      "loss": 3.6973,
+      "step": 6153
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.812016938639018e-06,
+      "loss": 3.6143,
+      "step": 6154
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8104760641970549e-06,
+      "loss": 3.6347,
+      "step": 6155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8089357003365316e-06,
+      "loss": 3.7509,
+      "step": 6156
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8073958473040303e-06,
+      "loss": 3.77,
+      "step": 6157
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8058565053460503e-06,
+      "loss": 3.5921,
+      "step": 6158
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8043176747090157e-06,
+      "loss": 3.755,
+      "step": 6159
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.802779355639257e-06,
+      "loss": 3.6565,
+      "step": 6160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8012415483830365e-06,
+      "loss": 3.6209,
+      "step": 6161
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.79970425318652e-06,
+      "loss": 3.7121,
+      "step": 6162
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7981674702958063e-06,
+      "loss": 3.4601,
+      "step": 6163
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7966311999568974e-06,
+      "loss": 3.6504,
+      "step": 6164
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7950954424157286e-06,
+      "loss": 3.8168,
+      "step": 6165
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7935601979181371e-06,
+      "loss": 3.6411,
+      "step": 6166
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.792025466709892e-06,
+      "loss": 3.667,
+      "step": 6167
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7904912490366723e-06,
+      "loss": 3.7808,
+      "step": 6168
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.788957545144076e-06,
+      "loss": 3.6612,
+      "step": 6169
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.787424355277621e-06,
+      "loss": 3.6526,
+      "step": 6170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7858916796827402e-06,
+      "loss": 3.6708,
+      "step": 6171
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7843595186047863e-06,
+      "loss": 3.8158,
+      "step": 6172
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.782827872289028e-06,
+      "loss": 3.7265,
+      "step": 6173
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7812967409806537e-06,
+      "loss": 3.7202,
+      "step": 6174
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.779766124924767e-06,
+      "loss": 3.6782,
+      "step": 6175
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7782360243663905e-06,
+      "loss": 3.7308,
+      "step": 6176
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7767064395504635e-06,
+      "loss": 3.6528,
+      "step": 6177
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7751773707218438e-06,
+      "loss": 3.6621,
+      "step": 6178
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7736488181253053e-06,
+      "loss": 3.8511,
+      "step": 6179
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7721207820055402e-06,
+      "loss": 3.7933,
+      "step": 6180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.770593262607158e-06,
+      "loss": 3.6582,
+      "step": 6181
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7690662601746827e-06,
+      "loss": 3.7587,
+      "step": 6182
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7675397749525636e-06,
+      "loss": 3.6673,
+      "step": 6183
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7660138071851545e-06,
+      "loss": 3.6231,
+      "step": 6184
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7644883571167404e-06,
+      "loss": 3.7052,
+      "step": 6185
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76296342499151e-06,
+      "loss": 3.6831,
+      "step": 6186
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7614390110535822e-06,
+      "loss": 3.7284,
+      "step": 6187
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7599151155469802e-06,
+      "loss": 3.6751,
+      "step": 6188
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7583917387156557e-06,
+      "loss": 3.6675,
+      "step": 6189
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7568688808034667e-06,
+      "loss": 3.7125,
+      "step": 6190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7553465420541976e-06,
+      "loss": 3.5739,
+      "step": 6191
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.753824722711544e-06,
+      "loss": 3.6277,
+      "step": 6192
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7523034230191205e-06,
+      "loss": 3.6803,
+      "step": 6193
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7507826432204571e-06,
+      "loss": 3.756,
+      "step": 6194
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7492623835590023e-06,
+      "loss": 3.7541,
+      "step": 6195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7477426442781198e-06,
+      "loss": 3.6231,
+      "step": 6196
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7462234256210908e-06,
+      "loss": 3.6963,
+      "step": 6197
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.744704727831113e-06,
+      "loss": 3.653,
+      "step": 6198
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7431865511513007e-06,
+      "loss": 3.6415,
+      "step": 6199
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7416688958246847e-06,
+      "loss": 3.5756,
+      "step": 6200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.740151762094212e-06,
+      "loss": 3.6445,
+      "step": 6201
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7386351502027481e-06,
+      "loss": 3.6518,
+      "step": 6202
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7371190603930715e-06,
+      "loss": 3.7138,
+      "step": 6203
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7356034929078803e-06,
+      "loss": 3.6668,
+      "step": 6204
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7340884479897868e-06,
+      "loss": 3.7533,
+      "step": 6205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7325739258813201e-06,
+      "loss": 3.6116,
+      "step": 6206
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7310599268249296e-06,
+      "loss": 3.6387,
+      "step": 6207
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7295464510629718e-06,
+      "loss": 3.6906,
+      "step": 6208
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7280334988377312e-06,
+      "loss": 3.5987,
+      "step": 6209
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.726521070391396e-06,
+      "loss": 3.6599,
+      "step": 6210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7250091659660833e-06,
+      "loss": 3.5859,
+      "step": 6211
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7234977858038137e-06,
+      "loss": 3.7385,
+      "step": 6212
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7219869301465365e-06,
+      "loss": 3.6077,
+      "step": 6213
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7204765992361039e-06,
+      "loss": 3.6659,
+      "step": 6214
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7189667933142957e-06,
+      "loss": 3.5978,
+      "step": 6215
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7174575126228016e-06,
+      "loss": 3.8556,
+      "step": 6216
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.715948757403228e-06,
+      "loss": 3.5707,
+      "step": 6217
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.714440527897097e-06,
+      "loss": 3.65,
+      "step": 6218
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7129328243458486e-06,
+      "loss": 3.6225,
+      "step": 6219
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.711425646990838e-06,
+      "loss": 3.7941,
+      "step": 6220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7099189960733304e-06,
+      "loss": 3.5985,
+      "step": 6221
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7084128718345183e-06,
+      "loss": 3.6068,
+      "step": 6222
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7069072745154962e-06,
+      "loss": 3.6687,
+      "step": 6223
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7054022043572882e-06,
+      "loss": 3.6417,
+      "step": 6224
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7038976616008208e-06,
+      "loss": 3.6845,
+      "step": 6225
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7023936464869468e-06,
+      "loss": 3.5669,
+      "step": 6226
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7008901592564287e-06,
+      "loss": 3.8131,
+      "step": 6227
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6993872001499461e-06,
+      "loss": 3.6421,
+      "step": 6228
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.697884769408093e-06,
+      "loss": 3.6791,
+      "step": 6229
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.696382867271381e-06,
+      "loss": 3.7407,
+      "step": 6230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6948814939802343e-06,
+      "loss": 3.7095,
+      "step": 6231
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6933806497749955e-06,
+      "loss": 3.5937,
+      "step": 6232
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.691880334895919e-06,
+      "loss": 3.6218,
+      "step": 6233
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6903805495831782e-06,
+      "loss": 3.7487,
+      "step": 6234
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6888812940768584e-06,
+      "loss": 3.711,
+      "step": 6235
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6873825686169625e-06,
+      "loss": 3.6291,
+      "step": 6236
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6858843734434071e-06,
+      "loss": 3.6385,
+      "step": 6237
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6843867087960252e-06,
+      "loss": 3.6877,
+      "step": 6238
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6828895749145613e-06,
+      "loss": 3.6799,
+      "step": 6239
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.681392972038684e-06,
+      "loss": 3.7539,
+      "step": 6240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6798969004079624e-06,
+      "loss": 3.6566,
+      "step": 6241
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.678401360261897e-06,
+      "loss": 3.685,
+      "step": 6242
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6769063518398876e-06,
+      "loss": 3.6896,
+      "step": 6243
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6754118753812632e-06,
+      "loss": 3.7571,
+      "step": 6244
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.673917931125254e-06,
+      "loss": 3.71,
+      "step": 6245
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6724245193110178e-06,
+      "loss": 3.7645,
+      "step": 6246
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6709316401776154e-06,
+      "loss": 3.7763,
+      "step": 6247
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.669439293964033e-06,
+      "loss": 3.7866,
+      "step": 6248
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6679474809091639e-06,
+      "loss": 3.6461,
+      "step": 6249
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6664562012518198e-06,
+      "loss": 3.7593,
+      "step": 6250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6649654552307249e-06,
+      "loss": 3.7356,
+      "step": 6251
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6634752430845196e-06,
+      "loss": 3.7462,
+      "step": 6252
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6619855650517586e-06,
+      "loss": 3.751,
+      "step": 6253
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6604964213709097e-06,
+      "loss": 3.7491,
+      "step": 6254
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.659007812280356e-06,
+      "loss": 3.6718,
+      "step": 6255
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6575197380183965e-06,
+      "loss": 3.6819,
+      "step": 6256
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6560321988232426e-06,
+      "loss": 3.679,
+      "step": 6257
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6545451949330204e-06,
+      "loss": 3.6707,
+      "step": 6258
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6530587265857712e-06,
+      "loss": 3.7016,
+      "step": 6259
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6515727940194493e-06,
+      "loss": 3.6248,
+      "step": 6260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6500873974719244e-06,
+      "loss": 3.7221,
+      "step": 6261
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6486025371809804e-06,
+      "loss": 3.6461,
+      "step": 6262
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6471182133843122e-06,
+      "loss": 3.6575,
+      "step": 6263
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6456344263195374e-06,
+      "loss": 3.8206,
+      "step": 6264
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.644151176224174e-06,
+      "loss": 3.6586,
+      "step": 6265
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6426684633356692e-06,
+      "loss": 3.6441,
+      "step": 6266
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6411862878913704e-06,
+      "loss": 3.6057,
+      "step": 6267
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.639704650128552e-06,
+      "loss": 3.6492,
+      "step": 6268
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.638223550284388e-06,
+      "loss": 3.7305,
+      "step": 6269
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6367429885959817e-06,
+      "loss": 3.7133,
+      "step": 6270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.635262965300336e-06,
+      "loss": 3.6747,
+      "step": 6271
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6337834806343783e-06,
+      "loss": 3.6738,
+      "step": 6272
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6323045348349447e-06,
+      "loss": 3.7423,
+      "step": 6273
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.630826128138786e-06,
+      "loss": 3.6142,
+      "step": 6274
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.629348260782566e-06,
+      "loss": 3.6141,
+      "step": 6275
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6278709330028636e-06,
+      "loss": 3.6272,
+      "step": 6276
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6263941450361703e-06,
+      "loss": 3.7229,
+      "step": 6277
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6249178971188911e-06,
+      "loss": 3.6418,
+      "step": 6278
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6234421894873453e-06,
+      "loss": 3.6342,
+      "step": 6279
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.621967022377765e-06,
+      "loss": 3.6788,
+      "step": 6280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6204923960262963e-06,
+      "loss": 3.6732,
+      "step": 6281
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.619018310668999e-06,
+      "loss": 3.5373,
+      "step": 6282
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6175447665418448e-06,
+      "loss": 3.6685,
+      "step": 6283
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6160717638807205e-06,
+      "loss": 3.6893,
+      "step": 6284
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6145993029214253e-06,
+      "loss": 3.6355,
+      "step": 6285
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6131273838996714e-06,
+      "loss": 3.6557,
+      "step": 6286
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6116560070510839e-06,
+      "loss": 3.7417,
+      "step": 6287
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6101851726112067e-06,
+      "loss": 3.7226,
+      "step": 6288
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6087148808154851e-06,
+      "loss": 3.7749,
+      "step": 6289
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6072451318992915e-06,
+      "loss": 3.7141,
+      "step": 6290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6057759260978978e-06,
+      "loss": 3.6731,
+      "step": 6291
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6043072636465017e-06,
+      "loss": 3.6539,
+      "step": 6292
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6028391447802017e-06,
+      "loss": 3.7676,
+      "step": 6293
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6013715697340222e-06,
+      "loss": 3.7815,
+      "step": 6294
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5999045387428863e-06,
+      "loss": 3.5488,
+      "step": 6295
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.598438052041643e-06,
+      "loss": 3.6189,
+      "step": 6296
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5969721098650475e-06,
+      "loss": 3.7156,
+      "step": 6297
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.595506712447768e-06,
+      "loss": 3.7012,
+      "step": 6298
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5940418600243868e-06,
+      "loss": 3.6401,
+      "step": 6299
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5925775528293985e-06,
+      "loss": 3.6061,
+      "step": 6300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5911137910972108e-06,
+      "loss": 3.7838,
+      "step": 6301
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5896505750621438e-06,
+      "loss": 3.6387,
+      "step": 6302
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5881879049584297e-06,
+      "loss": 3.6063,
+      "step": 6303
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.586725781020214e-06,
+      "loss": 3.5845,
+      "step": 6304
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5852642034815574e-06,
+      "loss": 3.6183,
+      "step": 6305
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5838031725764235e-06,
+      "loss": 3.5786,
+      "step": 6306
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5823426885387017e-06,
+      "loss": 3.6843,
+      "step": 6307
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5808827516021851e-06,
+      "loss": 3.6667,
+      "step": 6308
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5794233620005816e-06,
+      "loss": 3.7029,
+      "step": 6309
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.577964519967512e-06,
+      "loss": 3.7048,
+      "step": 6310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5765062257365077e-06,
+      "loss": 3.685,
+      "step": 6311
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5750484795410143e-06,
+      "loss": 3.7525,
+      "step": 6312
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5735912816143889e-06,
+      "loss": 3.7398,
+      "step": 6313
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5721346321899012e-06,
+      "loss": 3.6318,
+      "step": 6314
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5706785315007318e-06,
+      "loss": 3.6341,
+      "step": 6315
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5692229797799752e-06,
+      "loss": 3.708,
+      "step": 6316
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.567767977260637e-06,
+      "loss": 3.7217,
+      "step": 6317
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5663135241756345e-06,
+      "loss": 3.7135,
+      "step": 6318
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5648596207577988e-06,
+      "loss": 3.7249,
+      "step": 6319
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5634062672398697e-06,
+      "loss": 3.6783,
+      "step": 6320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5619534638545064e-06,
+      "loss": 3.6843,
+      "step": 6321
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5605012108342676e-06,
+      "loss": 3.6107,
+      "step": 6322
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5590495084116375e-06,
+      "loss": 3.7491,
+      "step": 6323
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.557598356819e-06,
+      "loss": 3.7315,
+      "step": 6324
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.556147756288663e-06,
+      "loss": 3.7327,
+      "step": 6325
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.554697707052833e-06,
+      "loss": 3.6965,
+      "step": 6326
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5532482093436407e-06,
+      "loss": 3.7118,
+      "step": 6327
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5517992633931178e-06,
+      "loss": 3.7895,
+      "step": 6328
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5503508694332182e-06,
+      "loss": 3.6165,
+      "step": 6329
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5489030276957956e-06,
+      "loss": 3.6514,
+      "step": 6330
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5474557384126266e-06,
+      "loss": 3.7263,
+      "step": 6331
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5460090018153923e-06,
+      "loss": 3.806,
+      "step": 6332
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5445628181356881e-06,
+      "loss": 3.686,
+      "step": 6333
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5431171876050198e-06,
+      "loss": 3.7041,
+      "step": 6334
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5416721104548055e-06,
+      "loss": 3.7343,
+      "step": 6335
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5402275869163736e-06,
+      "loss": 3.7218,
+      "step": 6336
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5387836172209652e-06,
+      "loss": 3.7076,
+      "step": 6337
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5373402015997312e-06,
+      "loss": 3.6612,
+      "step": 6338
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5358973402837358e-06,
+      "loss": 3.7589,
+      "step": 6339
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5344550335039521e-06,
+      "loss": 3.7108,
+      "step": 6340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5330132814912662e-06,
+      "loss": 3.6629,
+      "step": 6341
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.531572084476476e-06,
+      "loss": 3.6022,
+      "step": 6342
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5301314426902875e-06,
+      "loss": 3.6032,
+      "step": 6343
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5286913563633194e-06,
+      "loss": 3.7778,
+      "step": 6344
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5272518257261059e-06,
+      "loss": 3.6796,
+      "step": 6345
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5258128510090825e-06,
+      "loss": 3.7679,
+      "step": 6346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5243744324426075e-06,
+      "loss": 3.7189,
+      "step": 6347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.522936570256937e-06,
+      "loss": 3.6633,
+      "step": 6348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5214992646822529e-06,
+      "loss": 3.6936,
+      "step": 6349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5200625159486322e-06,
+      "loss": 3.7127,
+      "step": 6350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5186263242860783e-06,
+      "loss": 3.7859,
+      "step": 6351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.517190689924491e-06,
+      "loss": 3.7955,
+      "step": 6352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5157556130936941e-06,
+      "loss": 3.6731,
+      "step": 6353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5143210940234094e-06,
+      "loss": 3.6806,
+      "step": 6354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.512887132943281e-06,
+      "loss": 3.663,
+      "step": 6355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5114537300828558e-06,
+      "loss": 3.6418,
+      "step": 6356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.510020885671596e-06,
+      "loss": 3.6816,
+      "step": 6357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5085885999388705e-06,
+      "loss": 3.7502,
+      "step": 6358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5071568731139614e-06,
+      "loss": 3.6137,
+      "step": 6359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.505725705426061e-06,
+      "loss": 3.6105,
+      "step": 6360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5042950971042713e-06,
+      "loss": 3.7696,
+      "step": 6361
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5028650483776053e-06,
+      "loss": 3.5156,
+      "step": 6362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5014355594749859e-06,
+      "loss": 3.766,
+      "step": 6363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5000066306252475e-06,
+      "loss": 3.7658,
+      "step": 6364
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4985782620571337e-06,
+      "loss": 3.8033,
+      "step": 6365
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4971504539992992e-06,
+      "loss": 3.7147,
+      "step": 6366
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4957232066803079e-06,
+      "loss": 3.6228,
+      "step": 6367
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4942965203286337e-06,
+      "loss": 3.7039,
+      "step": 6368
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4928703951726664e-06,
+      "loss": 3.6843,
+      "step": 6369
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4914448314406942e-06,
+      "loss": 3.6254,
+      "step": 6370
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4900198293609298e-06,
+      "loss": 3.6472,
+      "step": 6371
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4885953891614813e-06,
+      "loss": 3.577,
+      "step": 6372
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.487171511070381e-06,
+      "loss": 3.6251,
+      "step": 6373
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4857481953155584e-06,
+      "loss": 3.7252,
+      "step": 6374
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4843254421248644e-06,
+      "loss": 3.7725,
+      "step": 6375
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.482903251726049e-06,
+      "loss": 3.7342,
+      "step": 6376
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4814816243467833e-06,
+      "loss": 3.6993,
+      "step": 6377
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4800605602146361e-06,
+      "loss": 3.6891,
+      "step": 6378
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4786400595570977e-06,
+      "loss": 3.6637,
+      "step": 6379
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4772201226015615e-06,
+      "loss": 3.6766,
+      "step": 6380
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4758007495753308e-06,
+      "loss": 3.6179,
+      "step": 6381
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4743819407056214e-06,
+      "loss": 3.5823,
+      "step": 6382
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4729636962195564e-06,
+      "loss": 3.6302,
+      "step": 6383
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4715460163441703e-06,
+      "loss": 3.5744,
+      "step": 6384
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.470128901306405e-06,
+      "loss": 3.5302,
+      "step": 6385
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4687123513331148e-06,
+      "loss": 3.6454,
+      "step": 6386
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4672963666510614e-06,
+      "loss": 3.7447,
+      "step": 6387
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4658809474869174e-06,
+      "loss": 3.7029,
+      "step": 6388
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4644660940672628e-06,
+      "loss": 3.7275,
+      "step": 6389
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4630518066185895e-06,
+      "loss": 3.7405,
+      "step": 6390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.461638085367298e-06,
+      "loss": 3.6717,
+      "step": 6391
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4602249305396966e-06,
+      "loss": 3.6696,
+      "step": 6392
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4588123423620054e-06,
+      "loss": 3.6981,
+      "step": 6393
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4574003210603522e-06,
+      "loss": 3.7076,
+      "step": 6394
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4559888668607741e-06,
+      "loss": 3.6579,
+      "step": 6395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4545779799892179e-06,
+      "loss": 3.664,
+      "step": 6396
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4531676606715401e-06,
+      "loss": 3.7902,
+      "step": 6397
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4517579091335054e-06,
+      "loss": 3.7408,
+      "step": 6398
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4503487256007875e-06,
+      "loss": 3.708,
+      "step": 6399
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4489401102989698e-06,
+      "loss": 3.7009,
+      "step": 6400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4475320634535428e-06,
+      "loss": 3.6454,
+      "step": 6401
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.446124585289913e-06,
+      "loss": 3.6619,
+      "step": 6402
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4447176760333835e-06,
+      "loss": 3.7312,
+      "step": 6403
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4433113359091805e-06,
+      "loss": 3.6959,
+      "step": 6404
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4419055651424253e-06,
+      "loss": 3.6407,
+      "step": 6405
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.440500363958161e-06,
+      "loss": 3.6423,
+      "step": 6406
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.439095732581327e-06,
+      "loss": 3.6615,
+      "step": 6407
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4376916712367838e-06,
+      "loss": 3.6301,
+      "step": 6408
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4362881801492896e-06,
+      "loss": 3.7253,
+      "step": 6409
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.434885259543521e-06,
+      "loss": 3.7206,
+      "step": 6410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4334829096440528e-06,
+      "loss": 3.7915,
+      "step": 6411
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4320811306753796e-06,
+      "loss": 3.6759,
+      "step": 6412
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4306799228618972e-06,
+      "loss": 3.6581,
+      "step": 6413
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.429279286427912e-06,
+      "loss": 3.7659,
+      "step": 6414
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.427879221597639e-06,
+      "loss": 3.6084,
+      "step": 6415
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.426479728595202e-06,
+      "loss": 3.6539,
+      "step": 6416
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4250808076446322e-06,
+      "loss": 3.595,
+      "step": 6417
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4236824589698706e-06,
+      "loss": 3.6775,
+      "step": 6418
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4222846827947651e-06,
+      "loss": 3.8088,
+      "step": 6419
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4208874793430738e-06,
+      "loss": 3.6865,
+      "step": 6420
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4194908488384618e-06,
+      "loss": 3.5665,
+      "step": 6421
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4180947915045024e-06,
+      "loss": 3.6673,
+      "step": 6422
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.416699307564678e-06,
+      "loss": 3.6098,
+      "step": 6423
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.415304397242378e-06,
+      "loss": 3.6927,
+      "step": 6424
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4139100607608996e-06,
+      "loss": 3.7139,
+      "step": 6425
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4125162983434544e-06,
+      "loss": 3.6427,
+      "step": 6426
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4111231102131507e-06,
+      "loss": 3.6119,
+      "step": 6427
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.409730496593016e-06,
+      "loss": 3.6889,
+      "step": 6428
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4083384577059761e-06,
+      "loss": 3.6296,
+      "step": 6429
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4069469937748758e-06,
+      "loss": 3.8432,
+      "step": 6430
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4055561050224547e-06,
+      "loss": 3.669,
+      "step": 6431
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4041657916713741e-06,
+      "loss": 3.608,
+      "step": 6432
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4027760539441904e-06,
+      "loss": 3.7724,
+      "step": 6433
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.40138689206338e-06,
+      "loss": 3.7341,
+      "step": 6434
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.399998306251314e-06,
+      "loss": 3.6904,
+      "step": 6435
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.398610296730284e-06,
+      "loss": 3.6454,
+      "step": 6436
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3972228637224816e-06,
+      "loss": 3.5879,
+      "step": 6437
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3958360074500078e-06,
+      "loss": 3.6729,
+      "step": 6438
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3944497281348723e-06,
+      "loss": 3.6921,
+      "step": 6439
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3930640259989914e-06,
+      "loss": 3.5941,
+      "step": 6440
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3916789012641895e-06,
+      "loss": 3.7675,
+      "step": 6441
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3902943541521986e-06,
+      "loss": 3.6886,
+      "step": 6442
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3889103848846568e-06,
+      "loss": 3.7291,
+      "step": 6443
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3875269936831133e-06,
+      "loss": 3.741,
+      "step": 6444
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.38614418076902e-06,
+      "loss": 3.7252,
+      "step": 6445
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3847619463637402e-06,
+      "loss": 3.7055,
+      "step": 6446
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3833802906885423e-06,
+      "loss": 3.631,
+      "step": 6447
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3819992139646037e-06,
+      "loss": 3.773,
+      "step": 6448
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3806187164130057e-06,
+      "loss": 3.6291,
+      "step": 6449
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3792387982547445e-06,
+      "loss": 3.7648,
+      "step": 6450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.377859459710712e-06,
+      "loss": 3.6987,
+      "step": 6451
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.37648070100172e-06,
+      "loss": 3.7618,
+      "step": 6452
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3751025223484755e-06,
+      "loss": 3.7854,
+      "step": 6453
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3737249239716043e-06,
+      "loss": 3.7347,
+      "step": 6454
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3723479060916273e-06,
+      "loss": 3.7847,
+      "step": 6455
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3709714689289844e-06,
+      "loss": 3.6582,
+      "step": 6456
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3695956127040112e-06,
+      "loss": 3.7193,
+      "step": 6457
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3682203376369612e-06,
+      "loss": 3.5666,
+      "step": 6458
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3668456439479839e-06,
+      "loss": 3.5572,
+      "step": 6459
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3654715318571455e-06,
+      "loss": 3.7735,
+      "step": 6460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.364098001584414e-06,
+      "loss": 3.6928,
+      "step": 6461
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3627250533496643e-06,
+      "loss": 3.6968,
+      "step": 6462
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.36135268737268e-06,
+      "loss": 3.7554,
+      "step": 6463
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3599809038731503e-06,
+      "loss": 3.6991,
+      "step": 6464
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3586097030706707e-06,
+      "loss": 3.6206,
+      "step": 6465
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3572390851847444e-06,
+      "loss": 3.6877,
+      "step": 6466
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3558690504347816e-06,
+      "loss": 3.7289,
+      "step": 6467
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.354499599040097e-06,
+      "loss": 3.7517,
+      "step": 6468
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.353130731219915e-06,
+      "loss": 3.7144,
+      "step": 6469
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3517624471933645e-06,
+      "loss": 3.6338,
+      "step": 6470
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.350394747179481e-06,
+      "loss": 3.6106,
+      "step": 6471
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3490276313972073e-06,
+      "loss": 3.6784,
+      "step": 6472
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3476611000653911e-06,
+      "loss": 3.7084,
+      "step": 6473
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3462951534027917e-06,
+      "loss": 3.7041,
+      "step": 6474
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3449297916280673e-06,
+      "loss": 3.6724,
+      "step": 6475
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3435650149597862e-06,
+      "loss": 3.7289,
+      "step": 6476
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3422008236164242e-06,
+      "loss": 3.6517,
+      "step": 6477
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3408372178163614e-06,
+      "loss": 3.6552,
+      "step": 6478
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3394741977778851e-06,
+      "loss": 3.6178,
+      "step": 6479
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3381117637191887e-06,
+      "loss": 3.5926,
+      "step": 6480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3367499158583713e-06,
+      "loss": 3.6775,
+      "step": 6481
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3353886544134392e-06,
+      "loss": 3.5829,
+      "step": 6482
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3340279796023031e-06,
+      "loss": 3.7119,
+      "step": 6483
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3326678916427803e-06,
+      "loss": 3.751,
+      "step": 6484
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3313083907525992e-06,
+      "loss": 3.7007,
+      "step": 6485
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3299494771493831e-06,
+      "loss": 3.7107,
+      "step": 6486
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3285911510506749e-06,
+      "loss": 3.7746,
+      "step": 6487
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3272334126739094e-06,
+      "loss": 3.7488,
+      "step": 6488
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3258762622364414e-06,
+      "loss": 3.5993,
+      "step": 6489
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3245196999555177e-06,
+      "loss": 3.7707,
+      "step": 6490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3231637260483044e-06,
+      "loss": 3.7003,
+      "step": 6491
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3218083407318606e-06,
+      "loss": 3.5804,
+      "step": 6492
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3204535442231614e-06,
+      "loss": 3.8098,
+      "step": 6493
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3190993367390836e-06,
+      "loss": 3.7121,
+      "step": 6494
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3177457184964087e-06,
+      "loss": 3.7012,
+      "step": 6495
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3163926897118252e-06,
+      "loss": 3.6395,
+      "step": 6496
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3150402506019266e-06,
+      "loss": 3.5381,
+      "step": 6497
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.313688401383213e-06,
+      "loss": 3.7723,
+      "step": 6498
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3123371422720894e-06,
+      "loss": 3.6761,
+      "step": 6499
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.310986473484866e-06,
+      "loss": 3.689,
+      "step": 6500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3096363952377589e-06,
+      "loss": 3.6873,
+      "step": 6501
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3082869077468902e-06,
+      "loss": 3.5954,
+      "step": 6502
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3069380112282865e-06,
+      "loss": 3.6597,
+      "step": 6503
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3055897058978806e-06,
+      "loss": 3.7128,
+      "step": 6504
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3042419919715094e-06,
+      "loss": 3.682,
+      "step": 6505
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3028948696649164e-06,
+      "loss": 3.6239,
+      "step": 6506
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3015483391937506e-06,
+      "loss": 3.5647,
+      "step": 6507
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.300202400773563e-06,
+      "loss": 3.6379,
+      "step": 6508
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.298857054619818e-06,
+      "loss": 3.7283,
+      "step": 6509
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2975123009478729e-06,
+      "loss": 3.7131,
+      "step": 6510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2961681399730036e-06,
+      "loss": 3.7034,
+      "step": 6511
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2948245719103774e-06,
+      "loss": 3.7089,
+      "step": 6512
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2934815969750807e-06,
+      "loss": 3.7373,
+      "step": 6513
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2921392153820916e-06,
+      "loss": 3.5478,
+      "step": 6514
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2907974273463053e-06,
+      "loss": 3.694,
+      "step": 6515
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2894562330825106e-06,
+      "loss": 3.6685,
+      "step": 6516
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2881156328054112e-06,
+      "loss": 3.6544,
+      "step": 6517
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2867756267296105e-06,
+      "loss": 3.8266,
+      "step": 6518
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.285436215069617e-06,
+      "loss": 3.6203,
+      "step": 6519
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2840973980398446e-06,
+      "loss": 3.7529,
+      "step": 6520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.282759175854613e-06,
+      "loss": 3.7195,
+      "step": 6521
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.281421548728145e-06,
+      "loss": 3.736,
+      "step": 6522
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2800845168745695e-06,
+      "loss": 3.7167,
+      "step": 6523
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.27874808050792e-06,
+      "loss": 3.76,
+      "step": 6524
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.277412239842133e-06,
+      "loss": 3.6637,
+      "step": 6525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2760769950910518e-06,
+      "loss": 3.6889,
+      "step": 6526
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2747423464684233e-06,
+      "loss": 3.6359,
+      "step": 6527
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.273408294187899e-06,
+      "loss": 3.6599,
+      "step": 6528
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2720748384630349e-06,
+      "loss": 3.6741,
+      "step": 6529
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2707419795072917e-06,
+      "loss": 3.574,
+      "step": 6530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2694097175340342e-06,
+      "loss": 3.5942,
+      "step": 6531
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2680780527565313e-06,
+      "loss": 3.69,
+      "step": 6532
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2667469853879605e-06,
+      "loss": 3.6546,
+      "step": 6533
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2654165156413944e-06,
+      "loss": 3.6265,
+      "step": 6534
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2640866437298216e-06,
+      "loss": 3.5823,
+      "step": 6535
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2627573698661228e-06,
+      "loss": 3.5794,
+      "step": 6536
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.261428694263096e-06,
+      "loss": 3.6486,
+      "step": 6537
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2601006171334295e-06,
+      "loss": 3.7772,
+      "step": 6538
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2587731386897301e-06,
+      "loss": 3.6653,
+      "step": 6539
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.257446259144494e-06,
+      "loss": 3.7059,
+      "step": 6540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2561199787101352e-06,
+      "loss": 3.7115,
+      "step": 6541
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2547942975989635e-06,
+      "loss": 3.6522,
+      "step": 6542
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2534692160231948e-06,
+      "loss": 3.6737,
+      "step": 6543
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2521447341949494e-06,
+      "loss": 3.7395,
+      "step": 6544
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.250820852326251e-06,
+      "loss": 3.7495,
+      "step": 6545
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.249497570629028e-06,
+      "loss": 3.715,
+      "step": 6546
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2481748893151124e-06,
+      "loss": 3.7549,
+      "step": 6547
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.24685280859624e-06,
+      "loss": 3.6909,
+      "step": 6548
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2455313286840504e-06,
+      "loss": 3.676,
+      "step": 6549
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.244210449790087e-06,
+      "loss": 3.6222,
+      "step": 6550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2428901721257974e-06,
+      "loss": 3.6486,
+      "step": 6551
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2415704959025321e-06,
+      "loss": 3.6778,
+      "step": 6552
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2402514213315463e-06,
+      "loss": 3.6526,
+      "step": 6553
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2389329486239975e-06,
+      "loss": 3.6698,
+      "step": 6554
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2376150779909512e-06,
+      "loss": 3.6624,
+      "step": 6555
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.236297809643368e-06,
+      "loss": 3.7557,
+      "step": 6556
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2349811437921227e-06,
+      "loss": 3.6865,
+      "step": 6557
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2336650806479826e-06,
+      "loss": 3.6122,
+      "step": 6558
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2323496204216302e-06,
+      "loss": 3.731,
+      "step": 6559
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2310347633236402e-06,
+      "loss": 3.6854,
+      "step": 6560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2297205095644977e-06,
+      "loss": 3.7345,
+      "step": 6561
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.22840685935459e-06,
+      "loss": 3.6857,
+      "step": 6562
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2270938129042065e-06,
+      "loss": 3.6411,
+      "step": 6563
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.225781370423541e-06,
+      "loss": 3.6654,
+      "step": 6564
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.224469532122689e-06,
+      "loss": 3.6747,
+      "step": 6565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2231582982116546e-06,
+      "loss": 3.6393,
+      "step": 6566
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2218476689003355e-06,
+      "loss": 3.7286,
+      "step": 6567
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2205376443985444e-06,
+      "loss": 3.6846,
+      "step": 6568
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2192282249159848e-06,
+      "loss": 3.5416,
+      "step": 6569
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2179194106622755e-06,
+      "loss": 3.7576,
+      "step": 6570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2166112018469262e-06,
+      "loss": 3.7263,
+      "step": 6571
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2153035986793627e-06,
+      "loss": 3.7151,
+      "step": 6572
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2139966013689002e-06,
+      "loss": 3.7159,
+      "step": 6573
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2126902101247696e-06,
+      "loss": 3.6777,
+      "step": 6574
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2113844251560964e-06,
+      "loss": 3.6328,
+      "step": 6575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2100792466719118e-06,
+      "loss": 3.6973,
+      "step": 6576
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2087746748811508e-06,
+      "loss": 3.6463,
+      "step": 6577
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2074707099926485e-06,
+      "loss": 3.7596,
+      "step": 6578
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2061673522151463e-06,
+      "loss": 3.5533,
+      "step": 6579
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2048646017572857e-06,
+      "loss": 3.5956,
+      "step": 6580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2035624588276124e-06,
+      "loss": 3.645,
+      "step": 6581
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2022609236345746e-06,
+      "loss": 3.7526,
+      "step": 6582
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.200959996386522e-06,
+      "loss": 3.688,
+      "step": 6583
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1996596772917091e-06,
+      "loss": 3.5683,
+      "step": 6584
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1983599665582918e-06,
+      "loss": 3.7581,
+      "step": 6585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1970608643943288e-06,
+      "loss": 3.7321,
+      "step": 6586
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1957623710077809e-06,
+      "loss": 3.6972,
+      "step": 6587
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1944644866065125e-06,
+      "loss": 3.7508,
+      "step": 6588
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1931672113982885e-06,
+      "loss": 3.6206,
+      "step": 6589
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1918705455907815e-06,
+      "loss": 3.6515,
+      "step": 6590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1905744893915573e-06,
+      "loss": 3.6605,
+      "step": 6591
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1892790430080952e-06,
+      "loss": 3.7593,
+      "step": 6592
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1879842066477655e-06,
+      "loss": 3.7679,
+      "step": 6593
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1866899805178527e-06,
+      "loss": 3.6969,
+      "step": 6594
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1853963648255317e-06,
+      "loss": 3.7533,
+      "step": 6595
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1841033597778905e-06,
+      "loss": 3.7371,
+      "step": 6596
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1828109655819097e-06,
+      "loss": 3.6946,
+      "step": 6597
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1815191824444805e-06,
+      "loss": 3.7057,
+      "step": 6598
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1802280105723907e-06,
+      "loss": 3.6891,
+      "step": 6599
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1789374501723328e-06,
+      "loss": 3.656,
+      "step": 6600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1776475014509004e-06,
+      "loss": 3.7103,
+      "step": 6601
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1763581646145895e-06,
+      "loss": 3.7006,
+      "step": 6602
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1750694398697976e-06,
+      "loss": 3.7218,
+      "step": 6603
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.173781327422826e-06,
+      "loss": 3.5897,
+      "step": 6604
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1724938274798752e-06,
+      "loss": 3.7262,
+      "step": 6605
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1712069402470506e-06,
+      "loss": 3.6558,
+      "step": 6606
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1699206659303574e-06,
+      "loss": 3.7542,
+      "step": 6607
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1686350047357036e-06,
+      "loss": 3.6797,
+      "step": 6608
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1673499568688984e-06,
+      "loss": 3.6622,
+      "step": 6609
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1660655225356533e-06,
+      "loss": 3.6711,
+      "step": 6610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.164781701941582e-06,
+      "loss": 3.6386,
+      "step": 6611
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1634984952921996e-06,
+      "loss": 3.6997,
+      "step": 6612
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1622159027929202e-06,
+      "loss": 3.6819,
+      "step": 6613
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.160933924649068e-06,
+      "loss": 3.711,
+      "step": 6614
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1596525610658565e-06,
+      "loss": 3.7042,
+      "step": 6615
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1583718122484128e-06,
+      "loss": 3.7284,
+      "step": 6616
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1570916784017556e-06,
+      "loss": 3.6459,
+      "step": 6617
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1558121597308136e-06,
+      "loss": 3.6834,
+      "step": 6618
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1545332564404093e-06,
+      "loss": 3.7043,
+      "step": 6619
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1532549687352752e-06,
+      "loss": 3.8435,
+      "step": 6620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1519772968200354e-06,
+      "loss": 3.649,
+      "step": 6621
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.150700240899224e-06,
+      "loss": 3.6585,
+      "step": 6622
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1494238011772724e-06,
+      "loss": 3.7301,
+      "step": 6623
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1481479778585142e-06,
+      "loss": 3.619,
+      "step": 6624
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1468727711471838e-06,
+      "loss": 3.6561,
+      "step": 6625
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.145598181247417e-06,
+      "loss": 3.6402,
+      "step": 6626
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1443242083632517e-06,
+      "loss": 3.6284,
+      "step": 6627
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1430508526986261e-06,
+      "loss": 3.6681,
+      "step": 6628
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1417781144573803e-06,
+      "loss": 3.6869,
+      "step": 6629
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1405059938432545e-06,
+      "loss": 3.6945,
+      "step": 6630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.139234491059892e-06,
+      "loss": 3.5906,
+      "step": 6631
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.137963606310834e-06,
+      "loss": 3.7127,
+      "step": 6632
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1366933397995266e-06,
+      "loss": 3.744,
+      "step": 6633
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1354236917293142e-06,
+      "loss": 3.6265,
+      "step": 6634
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1341546623034434e-06,
+      "loss": 3.6121,
+      "step": 6635
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.132886251725061e-06,
+      "loss": 3.8599,
+      "step": 6636
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1316184601972142e-06,
+      "loss": 3.8875,
+      "step": 6637
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1303512879228555e-06,
+      "loss": 3.6604,
+      "step": 6638
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1290847351048296e-06,
+      "loss": 3.7326,
+      "step": 6639
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.127818801945893e-06,
+      "loss": 3.6958,
+      "step": 6640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1265534886486917e-06,
+      "loss": 3.628,
+      "step": 6641
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1252887954157832e-06,
+      "loss": 3.5959,
+      "step": 6642
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1240247224496153e-06,
+      "loss": 3.7992,
+      "step": 6643
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1227612699525475e-06,
+      "loss": 3.5965,
+      "step": 6644
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1214984381268284e-06,
+      "loss": 3.6631,
+      "step": 6645
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.120236227174617e-06,
+      "loss": 3.7292,
+      "step": 6646
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1189746372979699e-06,
+      "loss": 3.7206,
+      "step": 6647
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1177136686988383e-06,
+      "loss": 3.6856,
+      "step": 6648
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1164533215790847e-06,
+      "loss": 3.768,
+      "step": 6649
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1151935961404613e-06,
+      "loss": 3.8069,
+      "step": 6650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1139344925846313e-06,
+      "loss": 3.771,
+      "step": 6651
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1126760111131474e-06,
+      "loss": 3.7013,
+      "step": 6652
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1114181519274735e-06,
+      "loss": 3.6618,
+      "step": 6653
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1101609152289633e-06,
+      "loss": 3.782,
+      "step": 6654
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.108904301218881e-06,
+      "loss": 3.6725,
+      "step": 6655
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1076483100983843e-06,
+      "loss": 3.6722,
+      "step": 6656
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1063929420685331e-06,
+      "loss": 3.7695,
+      "step": 6657
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1051381973302882e-06,
+      "loss": 3.7428,
+      "step": 6658
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.10388407608451e-06,
+      "loss": 3.68,
+      "step": 6659
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1026305785319585e-06,
+      "loss": 3.743,
+      "step": 6660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1013777048732954e-06,
+      "loss": 3.6232,
+      "step": 6661
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1001254553090812e-06,
+      "loss": 3.6372,
+      "step": 6662
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0988738300397772e-06,
+      "loss": 3.6741,
+      "step": 6663
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0976228292657447e-06,
+      "loss": 3.7107,
+      "step": 6664
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0963724531872444e-06,
+      "loss": 3.772,
+      "step": 6665
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0951227020044375e-06,
+      "loss": 3.7249,
+      "step": 6666
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0938735759173858e-06,
+      "loss": 3.6044,
+      "step": 6667
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0926250751260492e-06,
+      "loss": 3.7226,
+      "step": 6668
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0913771998302897e-06,
+      "loss": 3.6341,
+      "step": 6669
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0901299502298667e-06,
+      "loss": 3.7412,
+      "step": 6670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0888833265244442e-06,
+      "loss": 3.6377,
+      "step": 6671
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0876373289135778e-06,
+      "loss": 3.729,
+      "step": 6672
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0863919575967335e-06,
+      "loss": 3.7342,
+      "step": 6673
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0851472127732648e-06,
+      "loss": 3.6496,
+      "step": 6674
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0839030946424384e-06,
+      "loss": 3.6467,
+      "step": 6675
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0826596034034066e-06,
+      "loss": 3.7527,
+      "step": 6676
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.081416739255235e-06,
+      "loss": 3.6174,
+      "step": 6677
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0801745023968763e-06,
+      "loss": 3.5562,
+      "step": 6678
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0789328930271925e-06,
+      "loss": 3.6327,
+      "step": 6679
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0776919113449407e-06,
+      "loss": 3.7723,
+      "step": 6680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0764515575487784e-06,
+      "loss": 3.7626,
+      "step": 6681
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0752118318372617e-06,
+      "loss": 3.5669,
+      "step": 6682
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.073972734408847e-06,
+      "loss": 3.8083,
+      "step": 6683
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0727342654618905e-06,
+      "loss": 3.6543,
+      "step": 6684
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0714964251946464e-06,
+      "loss": 3.6292,
+      "step": 6685
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0702592138052693e-06,
+      "loss": 3.6915,
+      "step": 6686
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0690226314918135e-06,
+      "loss": 3.7748,
+      "step": 6687
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0677866784522317e-06,
+      "loss": 3.6455,
+      "step": 6688
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0665513548843765e-06,
+      "loss": 3.7126,
+      "step": 6689
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.065316660985999e-06,
+      "loss": 3.6705,
+      "step": 6690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0640825969547498e-06,
+      "loss": 3.6324,
+      "step": 6691
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0628491629881794e-06,
+      "loss": 3.8036,
+      "step": 6692
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0616163592837364e-06,
+      "loss": 3.637,
+      "step": 6693
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0603841860387681e-06,
+      "loss": 3.7062,
+      "step": 6694
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0591526434505256e-06,
+      "loss": 3.7672,
+      "step": 6695
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0579217317161494e-06,
+      "loss": 3.6892,
+      "step": 6696
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0566914510326908e-06,
+      "loss": 3.558,
+      "step": 6697
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.055461801597088e-06,
+      "loss": 3.6825,
+      "step": 6698
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0542327836061906e-06,
+      "loss": 3.6995,
+      "step": 6699
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0530043972567339e-06,
+      "loss": 3.6082,
+      "step": 6700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0517766427453657e-06,
+      "loss": 3.7206,
+      "step": 6701
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.05054952026862e-06,
+      "loss": 3.7883,
+      "step": 6702
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.049323030022939e-06,
+      "loss": 3.6973,
+      "step": 6703
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0480971722046602e-06,
+      "loss": 3.639,
+      "step": 6704
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.046871947010018e-06,
+      "loss": 3.6898,
+      "step": 6705
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0456473546351492e-06,
+      "loss": 3.7014,
+      "step": 6706
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0444233952760863e-06,
+      "loss": 3.743,
+      "step": 6707
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0432000691287619e-06,
+      "loss": 3.7213,
+      "step": 6708
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0419773763890073e-06,
+      "loss": 3.6774,
+      "step": 6709
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0407553172525525e-06,
+      "loss": 3.7527,
+      "step": 6710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0395338919150245e-06,
+      "loss": 3.6164,
+      "step": 6711
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0383131005719505e-06,
+      "loss": 3.6545,
+      "step": 6712
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.037092943418756e-06,
+      "loss": 3.7577,
+      "step": 6713
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0358734206507642e-06,
+      "loss": 3.6689,
+      "step": 6714
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.034654532463198e-06,
+      "loss": 3.7132,
+      "step": 6715
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0334362790511776e-06,
+      "loss": 3.6217,
+      "step": 6716
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0322186606097212e-06,
+      "loss": 3.61,
+      "step": 6717
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.031001677333746e-06,
+      "loss": 3.7028,
+      "step": 6718
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.029785329418071e-06,
+      "loss": 3.681,
+      "step": 6719
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0285696170574045e-06,
+      "loss": 3.63,
+      "step": 6720
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.027354540446364e-06,
+      "loss": 3.6501,
+      "step": 6721
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0261400997794557e-06,
+      "loss": 3.6331,
+      "step": 6722
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0249262952510918e-06,
+      "loss": 3.7042,
+      "step": 6723
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.023713127055575e-06,
+      "loss": 3.7342,
+      "step": 6724
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0225005953871153e-06,
+      "loss": 3.7602,
+      "step": 6725
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0212887004398104e-06,
+      "loss": 3.7055,
+      "step": 6726
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0200774424076648e-06,
+      "loss": 3.5963,
+      "step": 6727
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0188668214845765e-06,
+      "loss": 3.6348,
+      "step": 6728
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0176568378643432e-06,
+      "loss": 3.5473,
+      "step": 6729
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.016447491740659e-06,
+      "loss": 3.5945,
+      "step": 6730
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.015238783307118e-06,
+      "loss": 3.6988,
+      "step": 6731
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0140307127572125e-06,
+      "loss": 3.6105,
+      "step": 6732
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0128232802843263e-06,
+      "loss": 3.6452,
+      "step": 6733
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0116164860817523e-06,
+      "loss": 3.7016,
+      "step": 6734
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0104103303426694e-06,
+      "loss": 3.7228,
+      "step": 6735
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.009204813260164e-06,
+      "loss": 3.7462,
+      "step": 6736
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0079999350272152e-06,
+      "loss": 3.8091,
+      "step": 6737
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0067956958366998e-06,
+      "loss": 3.6703,
+      "step": 6738
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0055920958813947e-06,
+      "loss": 3.5671,
+      "step": 6739
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.004389135353972e-06,
+      "loss": 3.6331,
+      "step": 6740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.003186814447003e-06,
+      "loss": 3.5579,
+      "step": 6741
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0019851333529568e-06,
+      "loss": 3.6852,
+      "step": 6742
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0007840922641982e-06,
+      "loss": 3.7343,
+      "step": 6743
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.995836913729918e-07,
+      "loss": 3.54,
+      "step": 6744
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.983839308714976e-07,
+      "loss": 3.6196,
+      "step": 6745
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.97184810951775e-07,
+      "loss": 3.7932,
+      "step": 6746
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.959863318057805e-07,
+      "loss": 3.7389,
+      "step": 6747
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.947884936253666e-07,
+      "loss": 3.7596,
+      "step": 6748
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.935912966022843e-07,
+      "loss": 3.6659,
+      "step": 6749
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.92394740928182e-07,
+      "loss": 3.6954,
+      "step": 6750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.911988267946043e-07,
+      "loss": 3.6983,
+      "step": 6751
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.90003554392997e-07,
+      "loss": 3.6324,
+      "step": 6752
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.888089239146963e-07,
+      "loss": 3.8167,
+      "step": 6753
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.87614935550944e-07,
+      "loss": 3.5969,
+      "step": 6754
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.86421589492869e-07,
+      "loss": 3.5688,
+      "step": 6755
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.852288859315096e-07,
+      "loss": 3.6591,
+      "step": 6756
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.84036825057788e-07,
+      "loss": 3.6995,
+      "step": 6757
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.82845407062537e-07,
+      "loss": 3.7616,
+      "step": 6758
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.81654632136473e-07,
+      "loss": 3.6642,
+      "step": 6759
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.804645004702208e-07,
+      "loss": 3.6209,
+      "step": 6760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.792750122542965e-07,
+      "loss": 3.7391,
+      "step": 6761
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.780861676791137e-07,
+      "loss": 3.5421,
+      "step": 6762
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.768979669349837e-07,
+      "loss": 3.6723,
+      "step": 6763
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.757104102121152e-07,
+      "loss": 3.6495,
+      "step": 6764
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.745234977006124e-07,
+      "loss": 3.7325,
+      "step": 6765
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.733372295904776e-07,
+      "loss": 3.6907,
+      "step": 6766
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.721516060716096e-07,
+      "loss": 3.6068,
+      "step": 6767
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.709666273338037e-07,
+      "loss": 3.7117,
+      "step": 6768
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.697822935667517e-07,
+      "loss": 3.6985,
+      "step": 6769
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.685986049600438e-07,
+      "loss": 3.6183,
+      "step": 6770
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.674155617031649e-07,
+      "loss": 3.6332,
+      "step": 6771
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.662331639854977e-07,
+      "loss": 3.7195,
+      "step": 6772
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.650514119963218e-07,
+      "loss": 3.666,
+      "step": 6773
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.638703059248117e-07,
+      "loss": 3.7155,
+      "step": 6774
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.626898459600399e-07,
+      "loss": 3.6371,
+      "step": 6775
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.615100322909787e-07,
+      "loss": 3.6713,
+      "step": 6776
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.60330865106488e-07,
+      "loss": 3.6389,
+      "step": 6777
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.591523445953354e-07,
+      "loss": 3.757,
+      "step": 6778
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.579744709461747e-07,
+      "loss": 3.6423,
+      "step": 6779
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.567972443475648e-07,
+      "loss": 3.6668,
+      "step": 6780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.556206649879523e-07,
+      "loss": 3.7249,
+      "step": 6781
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.544447330556906e-07,
+      "loss": 3.606,
+      "step": 6782
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.532694487390182e-07,
+      "loss": 3.7188,
+      "step": 6783
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.520948122260792e-07,
+      "loss": 3.6177,
+      "step": 6784
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.509208237049094e-07,
+      "loss": 3.5511,
+      "step": 6785
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.497474833634418e-07,
+      "loss": 3.7477,
+      "step": 6786
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.485747913895055e-07,
+      "loss": 3.6666,
+      "step": 6787
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.474027479708254e-07,
+      "loss": 3.5732,
+      "step": 6788
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.462313532950241e-07,
+      "loss": 3.6267,
+      "step": 6789
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.450606075496188e-07,
+      "loss": 3.6322,
+      "step": 6790
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.438905109220237e-07,
+      "loss": 3.6638,
+      "step": 6791
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.427210635995482e-07,
+      "loss": 3.7233,
+      "step": 6792
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.415522657693987e-07,
+      "loss": 3.7576,
+      "step": 6793
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.403841176186773e-07,
+      "loss": 3.6719,
+      "step": 6794
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.392166193343821e-07,
+      "loss": 3.5269,
+      "step": 6795
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.380497711034065e-07,
+      "loss": 3.7434,
+      "step": 6796
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.368835731125415e-07,
+      "loss": 3.6512,
+      "step": 6797
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.357180255484721e-07,
+      "loss": 3.5814,
+      "step": 6798
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.345531285977788e-07,
+      "loss": 3.6798,
+      "step": 6799
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.33388882446944e-07,
+      "loss": 3.6518,
+      "step": 6800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.322252872823351e-07,
+      "loss": 3.6791,
+      "step": 6801
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.31062343290226e-07,
+      "loss": 3.6634,
+      "step": 6802
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.299000506567779e-07,
+      "loss": 3.6806,
+      "step": 6803
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.287384095680558e-07,
+      "loss": 3.7088,
+      "step": 6804
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.275774202100108e-07,
+      "loss": 3.5803,
+      "step": 6805
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.264170827685003e-07,
+      "loss": 3.6345,
+      "step": 6806
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.252573974292672e-07,
+      "loss": 3.6243,
+      "step": 6807
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.240983643779589e-07,
+      "loss": 3.6444,
+      "step": 6808
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.229399838001118e-07,
+      "loss": 3.6645,
+      "step": 6809
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.217822558811612e-07,
+      "loss": 3.7108,
+      "step": 6810
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.206251808064371e-07,
+      "loss": 3.6959,
+      "step": 6811
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.19468758761165e-07,
+      "loss": 3.7052,
+      "step": 6812
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.183129899304654e-07,
+      "loss": 3.7183,
+      "step": 6813
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.171578744993548e-07,
+      "loss": 3.6221,
+      "step": 6814
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.160034126527456e-07,
+      "loss": 3.6739,
+      "step": 6815
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.148496045754441e-07,
+      "loss": 3.6957,
+      "step": 6816
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.136964504521528e-07,
+      "loss": 3.739,
+      "step": 6817
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.125439504674699e-07,
+      "loss": 3.7132,
+      "step": 6818
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.113921048058888e-07,
+      "loss": 3.6879,
+      "step": 6819
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.10240913651797e-07,
+      "loss": 3.7078,
+      "step": 6820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.090903771894788e-07,
+      "loss": 3.6903,
+      "step": 6821
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.079404956031124e-07,
+      "loss": 3.6844,
+      "step": 6822
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.067912690767721e-07,
+      "loss": 3.76,
+      "step": 6823
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.056426977944272e-07,
+      "loss": 3.5974,
+      "step": 6824
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.044947819399408e-07,
+      "loss": 3.676,
+      "step": 6825
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.033475216970733e-07,
+      "loss": 3.584,
+      "step": 6826
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.022009172494784e-07,
+      "loss": 3.6485,
+      "step": 6827
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.010549687807058e-07,
+      "loss": 3.7643,
+      "step": 6828
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.999096764741993e-07,
+      "loss": 3.692,
+      "step": 6829
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.987650405132986e-07,
+      "loss": 3.6709,
+      "step": 6830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.976210610812375e-07,
+      "loss": 3.6255,
+      "step": 6831
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.964777383611445e-07,
+      "loss": 3.6473,
+      "step": 6832
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.953350725360466e-07,
+      "loss": 3.6337,
+      "step": 6833
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.94193063788858e-07,
+      "loss": 3.8011,
+      "step": 6834
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.93051712302398e-07,
+      "loss": 3.6488,
+      "step": 6835
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.919110182593682e-07,
+      "loss": 3.4819,
+      "step": 6836
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.907709818423788e-07,
+      "loss": 3.6384,
+      "step": 6837
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.896316032339214e-07,
+      "loss": 3.5968,
+      "step": 6838
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.884928826163941e-07,
+      "loss": 3.6165,
+      "step": 6839
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.873548201720788e-07,
+      "loss": 3.7393,
+      "step": 6840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.862174160831622e-07,
+      "loss": 3.5672,
+      "step": 6841
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.850806705317183e-07,
+      "loss": 3.7684,
+      "step": 6842
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.839445836997184e-07,
+      "loss": 3.6929,
+      "step": 6843
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.828091557690288e-07,
+      "loss": 3.6381,
+      "step": 6844
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.816743869214095e-07,
+      "loss": 3.6584,
+      "step": 6845
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.805402773385152e-07,
+      "loss": 3.7241,
+      "step": 6846
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.794068272018941e-07,
+      "loss": 3.6165,
+      "step": 6847
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.78274036692991e-07,
+      "loss": 3.6669,
+      "step": 6848
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.771419059931435e-07,
+      "loss": 3.6587,
+      "step": 6849
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.760104352835836e-07,
+      "loss": 3.7087,
+      "step": 6850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.748796247454383e-07,
+      "loss": 3.7276,
+      "step": 6851
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.73749474559728e-07,
+      "loss": 3.7207,
+      "step": 6852
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.726199849073691e-07,
+      "loss": 3.672,
+      "step": 6853
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.714911559691708e-07,
+      "loss": 3.6722,
+      "step": 6854
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.703629879258363e-07,
+      "loss": 3.643,
+      "step": 6855
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.692354809579634e-07,
+      "loss": 3.7286,
+      "step": 6856
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.681086352460466e-07,
+      "loss": 3.6549,
+      "step": 6857
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.669824509704688e-07,
+      "loss": 3.7202,
+      "step": 6858
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.658569283115142e-07,
+      "loss": 3.6402,
+      "step": 6859
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.647320674493531e-07,
+      "loss": 3.7172,
+      "step": 6860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.636078685640587e-07,
+      "loss": 3.6366,
+      "step": 6861
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.624843318355886e-07,
+      "loss": 3.7179,
+      "step": 6862
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.613614574438045e-07,
+      "loss": 3.6478,
+      "step": 6863
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.602392455684522e-07,
+      "loss": 3.6989,
+      "step": 6864
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.591176963891801e-07,
+      "loss": 3.5502,
+      "step": 6865
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.579968100855245e-07,
+      "loss": 3.6092,
+      "step": 6866
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.568765868369188e-07,
+      "loss": 3.6916,
+      "step": 6867
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.557570268226889e-07,
+      "loss": 3.7117,
+      "step": 6868
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.546381302220546e-07,
+      "loss": 3.6065,
+      "step": 6869
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.535198972141295e-07,
+      "loss": 3.7557,
+      "step": 6870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.52402327977922e-07,
+      "loss": 3.7011,
+      "step": 6871
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.512854226923328e-07,
+      "loss": 3.6454,
+      "step": 6872
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.501691815361568e-07,
+      "loss": 3.5241,
+      "step": 6873
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.490536046880832e-07,
+      "loss": 3.8023,
+      "step": 6874
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.479386923266947e-07,
+      "loss": 3.6503,
+      "step": 6875
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.468244446304664e-07,
+      "loss": 3.5947,
+      "step": 6876
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.457108617777688e-07,
+      "loss": 3.6052,
+      "step": 6877
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.445979439468643e-07,
+      "loss": 3.7256,
+      "step": 6878
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.434856913159101e-07,
+      "loss": 3.5552,
+      "step": 6879
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.423741040629557e-07,
+      "loss": 3.7003,
+      "step": 6880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.412631823659473e-07,
+      "loss": 3.7785,
+      "step": 6881
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.401529264027186e-07,
+      "loss": 3.701,
+      "step": 6882
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.390433363510037e-07,
+      "loss": 3.7034,
+      "step": 6883
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.379344123884231e-07,
+      "loss": 3.6479,
+      "step": 6884
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.368261546924977e-07,
+      "loss": 3.7078,
+      "step": 6885
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.357185634406345e-07,
+      "loss": 3.6784,
+      "step": 6886
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.346116388101422e-07,
+      "loss": 3.6626,
+      "step": 6887
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.335053809782129e-07,
+      "loss": 3.6203,
+      "step": 6888
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.323997901219416e-07,
+      "loss": 3.6978,
+      "step": 6889
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.312948664183095e-07,
+      "loss": 3.6442,
+      "step": 6890
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.301906100441954e-07,
+      "loss": 3.6918,
+      "step": 6891
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.29087021176368e-07,
+      "loss": 3.6732,
+      "step": 6892
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.279840999914912e-07,
+      "loss": 3.762,
+      "step": 6893
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.268818466661216e-07,
+      "loss": 3.6623,
+      "step": 6894
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.257802613767091e-07,
+      "loss": 3.6453,
+      "step": 6895
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.246793442995954e-07,
+      "loss": 3.6104,
+      "step": 6896
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.235790956110173e-07,
+      "loss": 3.7514,
+      "step": 6897
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.224795154871024e-07,
+      "loss": 3.5606,
+      "step": 6898
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.21380604103873e-07,
+      "loss": 3.6976,
+      "step": 6899
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.20282361637243e-07,
+      "loss": 3.7255,
+      "step": 6900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.191847882630205e-07,
+      "loss": 3.7025,
+      "step": 6901
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.180878841569062e-07,
+      "loss": 3.7054,
+      "step": 6902
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.169916494944924e-07,
+      "loss": 3.744,
+      "step": 6903
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.158960844512654e-07,
+      "loss": 3.7533,
+      "step": 6904
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.148011892026048e-07,
+      "loss": 3.6663,
+      "step": 6905
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.137069639237815e-07,
+      "loss": 3.7047,
+      "step": 6906
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.126134087899606e-07,
+      "loss": 3.7156,
+      "step": 6907
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.115205239761986e-07,
+      "loss": 3.7658,
+      "step": 6908
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.104283096574461e-07,
+      "loss": 3.6723,
+      "step": 6909
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.093367660085444e-07,
+      "loss": 3.723,
+      "step": 6910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.082458932042298e-07,
+      "loss": 3.6597,
+      "step": 6911
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.071556914191298e-07,
+      "loss": 3.671,
+      "step": 6912
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.060661608277631e-07,
+      "loss": 3.6233,
+      "step": 6913
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.049773016045464e-07,
+      "loss": 3.6205,
+      "step": 6914
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.038891139237808e-07,
+      "loss": 3.6757,
+      "step": 6915
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.028015979596681e-07,
+      "loss": 3.687,
+      "step": 6916
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.017147538862947e-07,
+      "loss": 3.7307,
+      "step": 6917
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.006285818776477e-07,
+      "loss": 3.7748,
+      "step": 6918
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.995430821075978e-07,
+      "loss": 3.6918,
+      "step": 6919
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.984582547499176e-07,
+      "loss": 3.7044,
+      "step": 6920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.97374099978262e-07,
+      "loss": 3.8162,
+      "step": 6921
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.962906179661872e-07,
+      "loss": 3.6371,
+      "step": 6922
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.952078088871362e-07,
+      "loss": 3.6189,
+      "step": 6923
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.941256729144464e-07,
+      "loss": 3.5843,
+      "step": 6924
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.93044210221347e-07,
+      "loss": 3.6655,
+      "step": 6925
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.919634209809585e-07,
+      "loss": 3.7724,
+      "step": 6926
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.908833053662956e-07,
+      "loss": 3.6409,
+      "step": 6927
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.898038635502631e-07,
+      "loss": 3.6811,
+      "step": 6928
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.887250957056597e-07,
+      "loss": 3.6571,
+      "step": 6929
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.876470020051746e-07,
+      "loss": 3.7122,
+      "step": 6930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.865695826213904e-07,
+      "loss": 3.7767,
+      "step": 6931
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.854928377267812e-07,
+      "loss": 3.6347,
+      "step": 6932
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.844167674937125e-07,
+      "loss": 3.6376,
+      "step": 6933
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.833413720944432e-07,
+      "loss": 3.6375,
+      "step": 6934
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.822666517011235e-07,
+      "loss": 3.6633,
+      "step": 6935
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.81192606485795e-07,
+      "loss": 3.8011,
+      "step": 6936
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.801192366203908e-07,
+      "loss": 3.7191,
+      "step": 6937
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.790465422767396e-07,
+      "loss": 3.6527,
+      "step": 6938
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.779745236265557e-07,
+      "loss": 3.5841,
+      "step": 6939
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.769031808414523e-07,
+      "loss": 3.7146,
+      "step": 6940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.758325140929262e-07,
+      "loss": 3.7334,
+      "step": 6941
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.747625235523759e-07,
+      "loss": 3.7089,
+      "step": 6942
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.73693209391081e-07,
+      "loss": 3.6538,
+      "step": 6943
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.726245717802233e-07,
+      "loss": 3.6987,
+      "step": 6944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.715566108908656e-07,
+      "loss": 3.5958,
+      "step": 6945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.704893268939722e-07,
+      "loss": 3.6848,
+      "step": 6946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.694227199603937e-07,
+      "loss": 3.6483,
+      "step": 6947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.68356790260873e-07,
+      "loss": 3.6208,
+      "step": 6948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.672915379660445e-07,
+      "loss": 3.7312,
+      "step": 6949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.662269632464353e-07,
+      "loss": 3.7428,
+      "step": 6950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.651630662724629e-07,
+      "loss": 3.6525,
+      "step": 6951
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.640998472144373e-07,
+      "loss": 3.718,
+      "step": 6952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.630373062425588e-07,
+      "loss": 3.664,
+      "step": 6953
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.6197544352692e-07,
+      "loss": 3.5382,
+      "step": 6954
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.609142592375046e-07,
+      "loss": 3.7023,
+      "step": 6955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.59853753544188e-07,
+      "loss": 3.7329,
+      "step": 6956
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.587939266167371e-07,
+      "loss": 3.6469,
+      "step": 6957
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.57734778624809e-07,
+      "loss": 3.7366,
+      "step": 6958
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.566763097379537e-07,
+      "loss": 3.6692,
+      "step": 6959
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.556185201256105e-07,
+      "loss": 3.6857,
+      "step": 6960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.545614099571119e-07,
+      "loss": 3.7326,
+      "step": 6961
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.535049794016824e-07,
+      "loss": 3.68,
+      "step": 6962
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.52449228628433e-07,
+      "loss": 3.9021,
+      "step": 6963
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.513941578063732e-07,
+      "loss": 3.6374,
+      "step": 6964
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.503397671043949e-07,
+      "loss": 3.6582,
+      "step": 6965
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.492860566912907e-07,
+      "loss": 3.7267,
+      "step": 6966
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.48233026735734e-07,
+      "loss": 3.7142,
+      "step": 6967
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.471806774063e-07,
+      "loss": 3.6469,
+      "step": 6968
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.461290088714451e-07,
+      "loss": 3.7784,
+      "step": 6969
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.45078021299524e-07,
+      "loss": 3.6424,
+      "step": 6970
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.440277148587788e-07,
+      "loss": 3.5616,
+      "step": 6971
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.429780897173427e-07,
+      "loss": 3.6696,
+      "step": 6972
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.419291460432421e-07,
+      "loss": 3.6712,
+      "step": 6973
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.408808840043913e-07,
+      "loss": 3.68,
+      "step": 6974
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.398333037685973e-07,
+      "loss": 3.65,
+      "step": 6975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.387864055035571e-07,
+      "loss": 3.668,
+      "step": 6976
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.377401893768599e-07,
+      "loss": 3.6708,
+      "step": 6977
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.366946555559846e-07,
+      "loss": 3.6493,
+      "step": 6978
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.356498042083005e-07,
+      "loss": 3.7157,
+      "step": 6979
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.346056355010683e-07,
+      "loss": 3.893,
+      "step": 6980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.335621496014394e-07,
+      "loss": 3.7099,
+      "step": 6981
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.325193466764563e-07,
+      "loss": 3.6409,
+      "step": 6982
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.314772268930508e-07,
+      "loss": 3.6191,
+      "step": 6983
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.304357904180475e-07,
+      "loss": 3.6847,
+      "step": 6984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.29395037418158e-07,
+      "loss": 3.719,
+      "step": 6985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.283549680599905e-07,
+      "loss": 3.5926,
+      "step": 6986
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.273155825100375e-07,
+      "loss": 3.7172,
+      "step": 6987
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.262768809346849e-07,
+      "loss": 3.6711,
+      "step": 6988
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.252388635002089e-07,
+      "loss": 3.6929,
+      "step": 6989
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.242015303727767e-07,
+      "loss": 3.67,
+      "step": 6990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.23164881718445e-07,
+      "loss": 3.7283,
+      "step": 6991
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.221289177031609e-07,
+      "loss": 3.641,
+      "step": 6992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.210936384927631e-07,
+      "loss": 3.7384,
+      "step": 6993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.200590442529786e-07,
+      "loss": 3.6774,
+      "step": 6994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.190251351494292e-07,
+      "loss": 3.5952,
+      "step": 6995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.179919113476192e-07,
+      "loss": 3.5554,
+      "step": 6996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.169593730129526e-07,
+      "loss": 3.6819,
+      "step": 6997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.159275203107141e-07,
+      "loss": 3.6803,
+      "step": 6998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.148963534060888e-07,
+      "loss": 3.6999,
+      "step": 6999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.138658724641417e-07,
+      "loss": 3.6393,
+      "step": 7000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.128360776498372e-07,
+      "loss": 3.6951,
+      "step": 7001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.118069691280211e-07,
+      "loss": 3.6711,
+      "step": 7002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.107785470634381e-07,
+      "loss": 3.6357,
+      "step": 7003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.097508116207169e-07,
+      "loss": 3.7055,
+      "step": 7004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.087237629643778e-07,
+      "loss": 3.6875,
+      "step": 7005
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.076974012588328e-07,
+      "loss": 3.7979,
+      "step": 7006
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.06671726668382e-07,
+      "loss": 3.6514,
+      "step": 7007
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.056467393572158e-07,
+      "loss": 3.742,
+      "step": 7008
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.046224394894158e-07,
+      "loss": 3.7491,
+      "step": 7009
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.035988272289517e-07,
+      "loss": 3.6697,
+      "step": 7010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.02575902739685e-07,
+      "loss": 3.7004,
+      "step": 7011
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.015536661853656e-07,
+      "loss": 3.64,
+      "step": 7012
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.005321177296349e-07,
+      "loss": 3.8296,
+      "step": 7013
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.995112575360219e-07,
+      "loss": 3.6789,
+      "step": 7014
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.984910857679472e-07,
+      "loss": 3.5709,
+      "step": 7015
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.974716025887207e-07,
+      "loss": 3.7593,
+      "step": 7016
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.964528081615424e-07,
+      "loss": 3.6582,
+      "step": 7017
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.954347026495e-07,
+      "loss": 3.6897,
+      "step": 7018
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.944172862155768e-07,
+      "loss": 3.6596,
+      "step": 7019
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.934005590226361e-07,
+      "loss": 3.8022,
+      "step": 7020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.923845212334412e-07,
+      "loss": 3.6759,
+      "step": 7021
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.913691730106359e-07,
+      "loss": 3.7378,
+      "step": 7022
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.903545145167628e-07,
+      "loss": 3.7147,
+      "step": 7023
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.893405459142439e-07,
+      "loss": 3.618,
+      "step": 7024
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.883272673654007e-07,
+      "loss": 3.713,
+      "step": 7025
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.873146790324359e-07,
+      "loss": 3.6622,
+      "step": 7026
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.863027810774475e-07,
+      "loss": 3.6529,
+      "step": 7027
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.85291573662421e-07,
+      "loss": 3.6582,
+      "step": 7028
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.842810569492308e-07,
+      "loss": 3.7067,
+      "step": 7029
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.832712310996408e-07,
+      "loss": 3.6765,
+      "step": 7030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.822620962753057e-07,
+      "loss": 3.7,
+      "step": 7031
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.812536526377673e-07,
+      "loss": 3.6385,
+      "step": 7032
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.80245900348459e-07,
+      "loss": 3.6689,
+      "step": 7033
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.792388395687022e-07,
+      "loss": 3.6649,
+      "step": 7034
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.782324704597076e-07,
+      "loss": 3.6104,
+      "step": 7035
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.772267931825765e-07,
+      "loss": 3.6827,
+      "step": 7036
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.762218078982979e-07,
+      "loss": 3.6861,
+      "step": 7037
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.752175147677504e-07,
+      "loss": 3.6727,
+      "step": 7038
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.742139139517023e-07,
+      "loss": 3.7743,
+      "step": 7039
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.732110056108105e-07,
+      "loss": 3.6987,
+      "step": 7040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.722087899056218e-07,
+      "loss": 3.5908,
+      "step": 7041
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.712072669965703e-07,
+      "loss": 3.6511,
+      "step": 7042
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.702064370439837e-07,
+      "loss": 3.7254,
+      "step": 7043
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.692063002080717e-07,
+      "loss": 3.6556,
+      "step": 7044
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.682068566489408e-07,
+      "loss": 3.6969,
+      "step": 7045
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.67208106526579e-07,
+      "loss": 3.7423,
+      "step": 7046
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.662100500008705e-07,
+      "loss": 3.7484,
+      "step": 7047
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.652126872315812e-07,
+      "loss": 3.6025,
+      "step": 7048
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.642160183783742e-07,
+      "loss": 3.6578,
+      "step": 7049
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.632200436007918e-07,
+      "loss": 3.6643,
+      "step": 7050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.622247630582745e-07,
+      "loss": 3.6413,
+      "step": 7051
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.612301769101464e-07,
+      "loss": 3.6805,
+      "step": 7052
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.602362853156214e-07,
+      "loss": 3.7812,
+      "step": 7053
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.592430884338025e-07,
+      "loss": 3.6643,
+      "step": 7054
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.58250586423681e-07,
+      "loss": 3.6312,
+      "step": 7055
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.572587794441381e-07,
+      "loss": 3.7128,
+      "step": 7056
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.562676676539426e-07,
+      "loss": 3.6845,
+      "step": 7057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.552772512117523e-07,
+      "loss": 3.555,
+      "step": 7058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.542875302761142e-07,
+      "loss": 3.6558,
+      "step": 7059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.532985050054635e-07,
+      "loss": 3.6664,
+      "step": 7060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.52310175558124e-07,
+      "loss": 3.6953,
+      "step": 7061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.51322542092308e-07,
+      "loss": 3.671,
+      "step": 7062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.503356047661175e-07,
+      "loss": 3.6805,
+      "step": 7063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.493493637375414e-07,
+      "loss": 3.7022,
+      "step": 7064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.483638191644575e-07,
+      "loss": 3.6711,
+      "step": 7065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.47378971204633e-07,
+      "loss": 3.6697,
+      "step": 7066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.46394820015725e-07,
+      "loss": 3.6116,
+      "step": 7067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.45411365755273e-07,
+      "loss": 3.7937,
+      "step": 7068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.444286085807144e-07,
+      "loss": 3.6134,
+      "step": 7069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.434465486493641e-07,
+      "loss": 3.7735,
+      "step": 7070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.424651861184361e-07,
+      "loss": 3.634,
+      "step": 7071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.414845211450243e-07,
+      "loss": 3.7517,
+      "step": 7072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.405045538861154e-07,
+      "loss": 3.755,
+      "step": 7073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.395252844985833e-07,
+      "loss": 3.7474,
+      "step": 7074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.385467131391893e-07,
+      "loss": 3.6164,
+      "step": 7075
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.375688399645863e-07,
+      "loss": 3.736,
+      "step": 7076
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.365916651313098e-07,
+      "loss": 3.7062,
+      "step": 7077
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.356151887957901e-07,
+      "loss": 3.774,
+      "step": 7078
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.346394111143384e-07,
+      "loss": 3.6017,
+      "step": 7079
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.336643322431624e-07,
+      "loss": 3.7474,
+      "step": 7080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.326899523383484e-07,
+      "loss": 3.742,
+      "step": 7081
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.317162715558811e-07,
+      "loss": 3.6168,
+      "step": 7082
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.307432900516236e-07,
+      "loss": 3.6966,
+      "step": 7083
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.297710079813346e-07,
+      "loss": 3.6093,
+      "step": 7084
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.287994255006563e-07,
+      "loss": 3.7735,
+      "step": 7085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.278285427651203e-07,
+      "loss": 3.6542,
+      "step": 7086
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.268583599301475e-07,
+      "loss": 3.7104,
+      "step": 7087
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.258888771510441e-07,
+      "loss": 3.5648,
+      "step": 7088
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.249200945830064e-07,
+      "loss": 3.663,
+      "step": 7089
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.239520123811177e-07,
+      "loss": 3.6029,
+      "step": 7090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.229846307003495e-07,
+      "loss": 3.7374,
+      "step": 7091
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.220179496955608e-07,
+      "loss": 3.6649,
+      "step": 7092
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.210519695214984e-07,
+      "loss": 3.6224,
+      "step": 7093
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.200866903327973e-07,
+      "loss": 3.5683,
+      "step": 7094
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.191221122839803e-07,
+      "loss": 3.6853,
+      "step": 7095
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.181582355294579e-07,
+      "loss": 3.6636,
+      "step": 7096
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.171950602235283e-07,
+      "loss": 3.7207,
+      "step": 7097
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.162325865203767e-07,
+      "loss": 3.5732,
+      "step": 7098
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.15270814574076e-07,
+      "loss": 3.6639,
+      "step": 7099
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.143097445385904e-07,
+      "loss": 3.6719,
+      "step": 7100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.13349376567765e-07,
+      "loss": 3.696,
+      "step": 7101
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.123897108153398e-07,
+      "loss": 3.7075,
+      "step": 7102
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.114307474349351e-07,
+      "loss": 3.6393,
+      "step": 7103
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.104724865800665e-07,
+      "loss": 3.6137,
+      "step": 7104
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.09514928404129e-07,
+      "loss": 3.6526,
+      "step": 7105
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.085580730604135e-07,
+      "loss": 3.6338,
+      "step": 7106
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.076019207020894e-07,
+      "loss": 3.7483,
+      "step": 7107
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.066464714822224e-07,
+      "loss": 3.6484,
+      "step": 7108
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.056917255537598e-07,
+      "loss": 3.6351,
+      "step": 7109
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.047376830695378e-07,
+      "loss": 3.5784,
+      "step": 7110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.037843441822816e-07,
+      "loss": 3.6617,
+      "step": 7111
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.028317090446006e-07,
+      "loss": 3.6345,
+      "step": 7112
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.018797778089946e-07,
+      "loss": 3.6781,
+      "step": 7113
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.009285506278495e-07,
+      "loss": 3.6351,
+      "step": 7114
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.999780276534378e-07,
+      "loss": 3.7741,
+      "step": 7115
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.990282090379201e-07,
+      "loss": 3.6347,
+      "step": 7116
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.980790949333448e-07,
+      "loss": 3.6497,
+      "step": 7117
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.971306854916459e-07,
+      "loss": 3.6075,
+      "step": 7118
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.961829808646464e-07,
+      "loss": 3.7048,
+      "step": 7119
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.952359812040548e-07,
+      "loss": 3.7709,
+      "step": 7120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.942896866614678e-07,
+      "loss": 3.6787,
+      "step": 7121
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.933440973883686e-07,
+      "loss": 3.8168,
+      "step": 7122
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.923992135361279e-07,
+      "loss": 3.6694,
+      "step": 7123
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.91455035256005e-07,
+      "loss": 3.6655,
+      "step": 7124
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.90511562699142e-07,
+      "loss": 3.7796,
+      "step": 7125
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.895687960165741e-07,
+      "loss": 3.6411,
+      "step": 7126
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.886267353592163e-07,
+      "loss": 3.6917,
+      "step": 7127
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.876853808778782e-07,
+      "loss": 3.6914,
+      "step": 7128
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.867447327232484e-07,
+      "loss": 3.7225,
+      "step": 7129
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.858047910459109e-07,
+      "loss": 3.5268,
+      "step": 7130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.848655559963284e-07,
+      "loss": 3.6375,
+      "step": 7131
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.839270277248565e-07,
+      "loss": 3.6466,
+      "step": 7132
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.82989206381736e-07,
+      "loss": 3.7228,
+      "step": 7133
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.820520921170931e-07,
+      "loss": 3.6767,
+      "step": 7134
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.811156850809413e-07,
+      "loss": 3.6509,
+      "step": 7135
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.801799854231826e-07,
+      "loss": 3.6839,
+      "step": 7136
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.792449932936034e-07,
+      "loss": 3.6011,
+      "step": 7137
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.783107088418788e-07,
+      "loss": 3.7049,
+      "step": 7138
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.773771322175692e-07,
+      "loss": 3.6004,
+      "step": 7139
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.764442635701229e-07,
+      "loss": 3.6886,
+      "step": 7140
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.755121030488731e-07,
+      "loss": 3.6422,
+      "step": 7141
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.745806508030422e-07,
+      "loss": 3.732,
+      "step": 7142
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.736499069817364e-07,
+      "loss": 3.6706,
+      "step": 7143
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.727198717339511e-07,
+      "loss": 3.6305,
+      "step": 7144
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.71790545208566e-07,
+      "loss": 3.7002,
+      "step": 7145
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.708619275543493e-07,
+      "loss": 3.601,
+      "step": 7146
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.699340189199532e-07,
+      "loss": 3.6989,
+      "step": 7147
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.690068194539217e-07,
+      "loss": 3.6562,
+      "step": 7148
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.680803293046766e-07,
+      "loss": 3.7593,
+      "step": 7149
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.671545486205359e-07,
+      "loss": 3.6517,
+      "step": 7150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.662294775496951e-07,
+      "loss": 3.6389,
+      "step": 7151
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.653051162402445e-07,
+      "loss": 3.7451,
+      "step": 7152
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.643814648401524e-07,
+      "loss": 3.7421,
+      "step": 7153
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.634585234972812e-07,
+      "loss": 3.6629,
+      "step": 7154
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.625362923593724e-07,
+      "loss": 3.7431,
+      "step": 7155
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.616147715740611e-07,
+      "loss": 3.7746,
+      "step": 7156
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.606939612888629e-07,
+      "loss": 3.7192,
+      "step": 7157
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.597738616511827e-07,
+      "loss": 3.6783,
+      "step": 7158
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.588544728083111e-07,
+      "loss": 3.7574,
+      "step": 7159
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.579357949074221e-07,
+      "loss": 3.693,
+      "step": 7160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.570178280955824e-07,
+      "loss": 3.6267,
+      "step": 7161
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.561005725197366e-07,
+      "loss": 3.7404,
+      "step": 7162
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.551840283267235e-07,
+      "loss": 3.7006,
+      "step": 7163
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.542681956632601e-07,
+      "loss": 3.772,
+      "step": 7164
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.533530746759575e-07,
+      "loss": 3.6357,
+      "step": 7165
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.524386655113067e-07,
+      "loss": 3.7012,
+      "step": 7166
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.515249683156881e-07,
+      "loss": 3.6184,
+      "step": 7167
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.506119832353662e-07,
+      "loss": 3.8494,
+      "step": 7168
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.496997104164931e-07,
+      "loss": 3.7437,
+      "step": 7169
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.487881500051056e-07,
+      "loss": 3.7085,
+      "step": 7170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.478773021471278e-07,
+      "loss": 3.7237,
+      "step": 7171
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.469671669883675e-07,
+      "loss": 3.5398,
+      "step": 7172
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.460577446745213e-07,
+      "loss": 3.807,
+      "step": 7173
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.45149035351169e-07,
+      "loss": 3.6178,
+      "step": 7174
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.442410391637787e-07,
+      "loss": 3.6304,
+      "step": 7175
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.433337562577018e-07,
+      "loss": 3.6124,
+      "step": 7176
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.424271867781778e-07,
+      "loss": 3.7413,
+      "step": 7177
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.41521330870331e-07,
+      "loss": 3.5675,
+      "step": 7178
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.40616188679171e-07,
+      "loss": 3.8092,
+      "step": 7179
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.397117603495927e-07,
+      "loss": 3.5027,
+      "step": 7180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.388080460263812e-07,
+      "loss": 3.5639,
+      "step": 7181
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.379050458541995e-07,
+      "loss": 3.7369,
+      "step": 7182
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.370027599776045e-07,
+      "loss": 3.6582,
+      "step": 7183
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.361011885410311e-07,
+      "loss": 3.6161,
+      "step": 7184
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.352003316888071e-07,
+      "loss": 3.5805,
+      "step": 7185
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.34300189565139e-07,
+      "loss": 3.6292,
+      "step": 7186
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.334007623141252e-07,
+      "loss": 3.5862,
+      "step": 7187
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.325020500797434e-07,
+      "loss": 3.6844,
+      "step": 7188
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.316040530058631e-07,
+      "loss": 3.8214,
+      "step": 7189
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.307067712362357e-07,
+      "loss": 3.567,
+      "step": 7190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.298102049144982e-07,
+      "loss": 3.6831,
+      "step": 7191
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.289143541841735e-07,
+      "loss": 3.5685,
+      "step": 7192
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.280192191886707e-07,
+      "loss": 3.6808,
+      "step": 7193
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.271248000712831e-07,
+      "loss": 3.7103,
+      "step": 7194
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.262310969751905e-07,
+      "loss": 3.5686,
+      "step": 7195
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.253381100434574e-07,
+      "loss": 3.7459,
+      "step": 7196
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.244458394190338e-07,
+      "loss": 3.7119,
+      "step": 7197
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.235542852447544e-07,
+      "loss": 3.6826,
+      "step": 7198
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.226634476633413e-07,
+      "loss": 3.6224,
+      "step": 7199
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.217733268173996e-07,
+      "loss": 3.5993,
+      "step": 7200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2088392284942e-07,
+      "loss": 3.6796,
+      "step": 7201
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.199952359017802e-07,
+      "loss": 3.6244,
+      "step": 7202
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.191072661167412e-07,
+      "loss": 3.6911,
+      "step": 7203
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.182200136364491e-07,
+      "loss": 3.6046,
+      "step": 7204
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.173334786029383e-07,
+      "loss": 3.754,
+      "step": 7205
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.164476611581232e-07,
+      "loss": 3.5726,
+      "step": 7206
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.155625614438086e-07,
+      "loss": 3.7059,
+      "step": 7207
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.146781796016798e-07,
+      "loss": 3.7139,
+      "step": 7208
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.137945157733115e-07,
+      "loss": 3.6559,
+      "step": 7209
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.129115701001586e-07,
+      "loss": 3.6814,
+      "step": 7210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.120293427235667e-07,
+      "loss": 3.7006,
+      "step": 7211
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.111478337847603e-07,
+      "loss": 3.7718,
+      "step": 7212
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.102670434248552e-07,
+      "loss": 3.5453,
+      "step": 7213
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.093869717848471e-07,
+      "loss": 3.6329,
+      "step": 7214
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.085076190056188e-07,
+      "loss": 3.6085,
+      "step": 7215
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.076289852279375e-07,
+      "loss": 3.6792,
+      "step": 7216
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.067510705924567e-07,
+      "loss": 3.6754,
+      "step": 7217
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.058738752397124e-07,
+      "loss": 3.8062,
+      "step": 7218
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.049973993101276e-07,
+      "loss": 3.6692,
+      "step": 7219
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.041216429440088e-07,
+      "loss": 3.6838,
+      "step": 7220
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.032466062815478e-07,
+      "loss": 3.6472,
+      "step": 7221
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.023722894628214e-07,
+      "loss": 3.6145,
+      "step": 7222
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.014986926277904e-07,
+      "loss": 3.7079,
+      "step": 7223
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.006258159163007e-07,
+      "loss": 3.6767,
+      "step": 7224
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.997536594680841e-07,
+      "loss": 3.575,
+      "step": 7225
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.988822234227553e-07,
+      "loss": 3.7306,
+      "step": 7226
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.980115079198144e-07,
+      "loss": 3.648,
+      "step": 7227
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.971415130986457e-07,
+      "loss": 3.6012,
+      "step": 7228
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.962722390985214e-07,
+      "loss": 3.6943,
+      "step": 7229
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.954036860585915e-07,
+      "loss": 3.7011,
+      "step": 7230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.945358541178985e-07,
+      "loss": 3.7361,
+      "step": 7231
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.936687434153619e-07,
+      "loss": 3.7771,
+      "step": 7232
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.92802354089793e-07,
+      "loss": 3.5667,
+      "step": 7233
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.919366862798807e-07,
+      "loss": 3.6519,
+      "step": 7234
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.910717401242054e-07,
+      "loss": 3.7029,
+      "step": 7235
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.902075157612241e-07,
+      "loss": 3.6425,
+      "step": 7236
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.89344013329286e-07,
+      "loss": 3.7326,
+      "step": 7237
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.8848123296662e-07,
+      "loss": 3.6839,
+      "step": 7238
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.876191748113407e-07,
+      "loss": 3.5938,
+      "step": 7239
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.867578390014466e-07,
+      "loss": 3.7164,
+      "step": 7240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.858972256748218e-07,
+      "loss": 3.6282,
+      "step": 7241
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.850373349692334e-07,
+      "loss": 3.654,
+      "step": 7242
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.841781670223334e-07,
+      "loss": 3.6524,
+      "step": 7243
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.833197219716595e-07,
+      "loss": 3.7428,
+      "step": 7244
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.824619999546282e-07,
+      "loss": 3.7661,
+      "step": 7245
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.816050011085494e-07,
+      "loss": 3.702,
+      "step": 7246
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.807487255706078e-07,
+      "loss": 3.6993,
+      "step": 7247
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.798931734778794e-07,
+      "loss": 3.8527,
+      "step": 7248
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.790383449673214e-07,
+      "loss": 3.7065,
+      "step": 7249
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.781842401757741e-07,
+      "loss": 3.7083,
+      "step": 7250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.773308592399644e-07,
+      "loss": 3.709,
+      "step": 7251
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.764782022965014e-07,
+      "loss": 3.7157,
+      "step": 7252
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7562626948187974e-07,
+      "loss": 3.6162,
+      "step": 7253
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.747750609324769e-07,
+      "loss": 3.7387,
+      "step": 7254
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.739245767845552e-07,
+      "loss": 3.7258,
+      "step": 7255
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.730748171742611e-07,
+      "loss": 3.683,
+      "step": 7256
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.722257822376242e-07,
+      "loss": 3.6799,
+      "step": 7257
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7137747211055905e-07,
+      "loss": 3.5333,
+      "step": 7258
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.705298869288638e-07,
+      "loss": 3.5888,
+      "step": 7259
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.696830268282204e-07,
+      "loss": 3.55,
+      "step": 7260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6883689194419436e-07,
+      "loss": 3.6827,
+      "step": 7261
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6799148241223725e-07,
+      "loss": 3.7163,
+      "step": 7262
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.671467983676803e-07,
+      "loss": 3.7177,
+      "step": 7263
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6630283994574475e-07,
+      "loss": 3.6565,
+      "step": 7264
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6545960728152817e-07,
+      "loss": 3.6795,
+      "step": 7265
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.646171005100192e-07,
+      "loss": 3.6241,
+      "step": 7266
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6377531976608434e-07,
+      "loss": 3.6182,
+      "step": 7267
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.629342651844787e-07,
+      "loss": 3.6474,
+      "step": 7268
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6209393689983673e-07,
+      "loss": 3.8106,
+      "step": 7269
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.612543350466814e-07,
+      "loss": 3.7412,
+      "step": 7270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6041545975941404e-07,
+      "loss": 3.705,
+      "step": 7271
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.595773111723245e-07,
+      "loss": 3.6386,
+      "step": 7272
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.587398894195838e-07,
+      "loss": 3.6942,
+      "step": 7273
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5790319463524637e-07,
+      "loss": 3.6124,
+      "step": 7274
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5706722695325167e-07,
+      "loss": 3.6601,
+      "step": 7275
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.562319865074222e-07,
+      "loss": 3.6612,
+      "step": 7276
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5539747343146313e-07,
+      "loss": 3.6886,
+      "step": 7277
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.545636878589643e-07,
+      "loss": 3.6437,
+      "step": 7278
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.53730629923399e-07,
+      "loss": 3.6209,
+      "step": 7279
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.528982997581233e-07,
+      "loss": 3.6551,
+      "step": 7280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5206669749637775e-07,
+      "loss": 3.5972,
+      "step": 7281
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5123582327128543e-07,
+      "loss": 3.6153,
+      "step": 7282
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.504056772158538e-07,
+      "loss": 3.692,
+      "step": 7283
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4957625946297266e-07,
+      "loss": 3.7068,
+      "step": 7284
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4874757014541525e-07,
+      "loss": 3.6903,
+      "step": 7285
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.479196093958421e-07,
+      "loss": 3.6482,
+      "step": 7286
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.470923773467889e-07,
+      "loss": 3.6921,
+      "step": 7287
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.462658741306847e-07,
+      "loss": 3.5767,
+      "step": 7288
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4544009987983195e-07,
+      "loss": 3.6386,
+      "step": 7289
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4461505472642496e-07,
+      "loss": 3.736,
+      "step": 7290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4379073880253477e-07,
+      "loss": 3.7157,
+      "step": 7291
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4296715224012187e-07,
+      "loss": 3.7405,
+      "step": 7292
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4214429517102196e-07,
+      "loss": 3.6651,
+      "step": 7293
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4132216772696355e-07,
+      "loss": 3.7104,
+      "step": 7294
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.405007700395497e-07,
+      "loss": 3.73,
+      "step": 7295
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3968010224027247e-07,
+      "loss": 3.6711,
+      "step": 7296
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3886016446050393e-07,
+      "loss": 3.6681,
+      "step": 7297
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.380409568315014e-07,
+      "loss": 3.6584,
+      "step": 7298
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.372224794844032e-07,
+      "loss": 3.5473,
+      "step": 7299
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.364047325502324e-07,
+      "loss": 3.6256,
+      "step": 7300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.355877161598943e-07,
+      "loss": 3.766,
+      "step": 7301
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.347714304441775e-07,
+      "loss": 3.7432,
+      "step": 7302
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.339558755337542e-07,
+      "loss": 3.7136,
+      "step": 7303
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.331410515591783e-07,
+      "loss": 3.7457,
+      "step": 7304
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.323269586508877e-07,
+      "loss": 3.6531,
+      "step": 7305
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3151359693920315e-07,
+      "loss": 3.7299,
+      "step": 7306
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3070096655432824e-07,
+      "loss": 3.6807,
+      "step": 7307
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.298890676263495e-07,
+      "loss": 3.6894,
+      "step": 7308
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.290779002852352e-07,
+      "loss": 3.6569,
+      "step": 7309
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2826746466084036e-07,
+      "loss": 3.7147,
+      "step": 7310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.274577608828967e-07,
+      "loss": 3.647,
+      "step": 7311
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2664878908102556e-07,
+      "loss": 3.6988,
+      "step": 7312
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.258405493847245e-07,
+      "loss": 3.6638,
+      "step": 7313
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.250330419233806e-07,
+      "loss": 3.6868,
+      "step": 7314
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2422626682625644e-07,
+      "loss": 3.6094,
+      "step": 7315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2342022422250553e-07,
+      "loss": 3.7189,
+      "step": 7316
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.226149142411551e-07,
+      "loss": 3.7425,
+      "step": 7317
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.218103370111237e-07,
+      "loss": 3.6962,
+      "step": 7318
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2100649266120675e-07,
+      "loss": 3.6269,
+      "step": 7319
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2020338132008454e-07,
+      "loss": 3.6246,
+      "step": 7320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.194010031163204e-07,
+      "loss": 3.7294,
+      "step": 7321
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.185993581783587e-07,
+      "loss": 3.653,
+      "step": 7322
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.177984466345275e-07,
+      "loss": 3.6516,
+      "step": 7323
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1699826861303804e-07,
+      "loss": 3.6666,
+      "step": 7324
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1619882424198286e-07,
+      "loss": 3.6689,
+      "step": 7325
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1540011364933796e-07,
+      "loss": 3.6224,
+      "step": 7326
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1460213696296157e-07,
+      "loss": 3.7052,
+      "step": 7327
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.138048943105938e-07,
+      "loss": 3.6125,
+      "step": 7328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.130083858198586e-07,
+      "loss": 3.7755,
+      "step": 7329
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.122126116182612e-07,
+      "loss": 3.5251,
+      "step": 7330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1141757183318977e-07,
+      "loss": 3.6143,
+      "step": 7331
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.106232665919152e-07,
+      "loss": 3.7192,
+      "step": 7332
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.098296960215903e-07,
+      "loss": 3.575,
+      "step": 7333
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.090368602492506e-07,
+      "loss": 3.7345,
+      "step": 7334
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0824475940181296e-07,
+      "loss": 3.8074,
+      "step": 7335
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0745339360607927e-07,
+      "loss": 3.6132,
+      "step": 7336
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.066627629887304e-07,
+      "loss": 3.7507,
+      "step": 7337
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.058728676763313e-07,
+      "loss": 3.6988,
+      "step": 7338
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0508370779533013e-07,
+      "loss": 3.6376,
+      "step": 7339
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.042952834720548e-07,
+      "loss": 3.6299,
+      "step": 7340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.035075948327177e-07,
+      "loss": 3.5977,
+      "step": 7341
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.027206420034119e-07,
+      "loss": 3.5803,
+      "step": 7342
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0193442511011535e-07,
+      "loss": 3.7679,
+      "step": 7343
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.011489442786831e-07,
+      "loss": 3.6255,
+      "step": 7344
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.003641996348595e-07,
+      "loss": 3.8031,
+      "step": 7345
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9958019130426296e-07,
+      "loss": 3.7536,
+      "step": 7346
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.987969194124014e-07,
+      "loss": 3.6561,
+      "step": 7347
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9801438408465895e-07,
+      "loss": 3.6309,
+      "step": 7348
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9723258544630714e-07,
+      "loss": 3.7086,
+      "step": 7349
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9645152362249415e-07,
+      "loss": 3.7391,
+      "step": 7350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9567119873825555e-07,
+      "loss": 3.652,
+      "step": 7351
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9489161091850413e-07,
+      "loss": 3.7381,
+      "step": 7352
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9411276028803793e-07,
+      "loss": 3.5901,
+      "step": 7353
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.933346469715366e-07,
+      "loss": 3.6571,
+      "step": 7354
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.925572710935599e-07,
+      "loss": 3.6541,
+      "step": 7355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.917806327785517e-07,
+      "loss": 3.6616,
+      "step": 7356
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.910047321508359e-07,
+      "loss": 3.7508,
+      "step": 7357
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.902295693346203e-07,
+      "loss": 3.6942,
+      "step": 7358
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.894551444539929e-07,
+      "loss": 3.6691,
+      "step": 7359
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.886814576329245e-07,
+      "loss": 3.7558,
+      "step": 7360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8790850899526723e-07,
+      "loss": 3.6753,
+      "step": 7361
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8713629866475535e-07,
+      "loss": 3.7211,
+      "step": 7362
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8636482676500555e-07,
+      "loss": 3.5711,
+      "step": 7363
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.855940934195146e-07,
+      "loss": 3.6135,
+      "step": 7364
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.848240987516627e-07,
+      "loss": 3.6153,
+      "step": 7365
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.840548428847102e-07,
+      "loss": 3.8049,
+      "step": 7366
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.832863259418024e-07,
+      "loss": 3.7604,
+      "step": 7367
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.825185480459614e-07,
+      "loss": 3.8415,
+      "step": 7368
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.817515093200963e-07,
+      "loss": 3.6533,
+      "step": 7369
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8098520988699264e-07,
+      "loss": 3.6569,
+      "step": 7370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8021964986932283e-07,
+      "loss": 3.7473,
+      "step": 7371
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.794548293896355e-07,
+      "loss": 3.6169,
+      "step": 7372
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.786907485703667e-07,
+      "loss": 3.7175,
+      "step": 7373
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.779274075338285e-07,
+      "loss": 3.6358,
+      "step": 7374
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.771648064022193e-07,
+      "loss": 3.7313,
+      "step": 7375
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7640294529761424e-07,
+      "loss": 3.7468,
+      "step": 7376
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.756418243419746e-07,
+      "loss": 3.691,
+      "step": 7377
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.748814436571413e-07,
+      "loss": 3.7014,
+      "step": 7378
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.741218033648358e-07,
+      "loss": 3.6694,
+      "step": 7379
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7336290358666206e-07,
+      "loss": 3.7126,
+      "step": 7380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7260474444410556e-07,
+      "loss": 3.773,
+      "step": 7381
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.718473260585331e-07,
+      "loss": 3.6697,
+      "step": 7382
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.710906485511928e-07,
+      "loss": 3.7688,
+      "step": 7383
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.703347120432138e-07,
+      "loss": 3.6252,
+      "step": 7384
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.695795166556071e-07,
+      "loss": 3.7339,
+      "step": 7385
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.688250625092649e-07,
+      "loss": 3.6999,
+      "step": 7386
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6807134972496106e-07,
+      "loss": 3.6577,
+      "step": 7387
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6731837842335085e-07,
+      "loss": 3.6425,
+      "step": 7388
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6656614872496945e-07,
+      "loss": 3.6403,
+      "step": 7389
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6581466075023443e-07,
+      "loss": 3.6064,
+      "step": 7390
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6506391461944623e-07,
+      "loss": 3.7208,
+      "step": 7391
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6431391045278263e-07,
+      "loss": 3.7277,
+      "step": 7392
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.635646483703076e-07,
+      "loss": 3.7262,
+      "step": 7393
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.628161284919601e-07,
+      "loss": 3.7488,
+      "step": 7394
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.620683509375672e-07,
+      "loss": 3.8157,
+      "step": 7395
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6132131582683086e-07,
+      "loss": 3.6386,
+      "step": 7396
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.605750232793398e-07,
+      "loss": 3.7546,
+      "step": 7397
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.59829473414558e-07,
+      "loss": 3.6307,
+      "step": 7398
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5908466635183657e-07,
+      "loss": 3.7046,
+      "step": 7399
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.583406022104019e-07,
+      "loss": 3.7051,
+      "step": 7400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.57597281109367e-07,
+      "loss": 3.5118,
+      "step": 7401
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.568547031677222e-07,
+      "loss": 3.8164,
+      "step": 7402
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.561128685043397e-07,
+      "loss": 3.6495,
+      "step": 7403
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5537177723797335e-07,
+      "loss": 3.6376,
+      "step": 7404
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.546314294872577e-07,
+      "loss": 3.6826,
+      "step": 7405
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.538918253707074e-07,
+      "loss": 3.6663,
+      "step": 7406
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5315296500672e-07,
+      "loss": 3.642,
+      "step": 7407
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.52414848513572e-07,
+      "loss": 3.6495,
+      "step": 7408
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5167747600942217e-07,
+      "loss": 3.6305,
+      "step": 7409
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5094084761230883e-07,
+      "loss": 3.6558,
+      "step": 7410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.502049634401533e-07,
+      "loss": 3.6725,
+      "step": 7411
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4946982361075524e-07,
+      "loss": 3.6571,
+      "step": 7412
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4873542824179707e-07,
+      "loss": 3.7002,
+      "step": 7413
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4800177745084094e-07,
+      "loss": 3.6024,
+      "step": 7414
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4726887135533127e-07,
+      "loss": 3.5798,
+      "step": 7415
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4653671007259084e-07,
+      "loss": 3.6742,
+      "step": 7416
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.458052937198253e-07,
+      "loss": 3.7055,
+      "step": 7417
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.450746224141205e-07,
+      "loss": 3.5737,
+      "step": 7418
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.443446962724423e-07,
+      "loss": 3.7405,
+      "step": 7419
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.436155154116383e-07,
+      "loss": 3.7143,
+      "step": 7420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.428870799484363e-07,
+      "loss": 3.6698,
+      "step": 7421
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4215938999944466e-07,
+      "loss": 3.6589,
+      "step": 7422
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4143244568115243e-07,
+      "loss": 3.7662,
+      "step": 7423
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.407062471099298e-07,
+      "loss": 3.5183,
+      "step": 7424
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3998079440202616e-07,
+      "loss": 3.6902,
+      "step": 7425
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.392560876735751e-07,
+      "loss": 3.618,
+      "step": 7426
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3853212704058446e-07,
+      "loss": 3.7102,
+      "step": 7427
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3780891261895043e-07,
+      "loss": 3.5838,
+      "step": 7428
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3708644452444206e-07,
+      "loss": 3.6312,
+      "step": 7429
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3636472287271573e-07,
+      "loss": 3.653,
+      "step": 7430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3564374777930175e-07,
+      "loss": 3.6837,
+      "step": 7431
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.349235193596184e-07,
+      "loss": 3.6635,
+      "step": 7432
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.342040377289568e-07,
+      "loss": 3.6697,
+      "step": 7433
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3348530300249426e-07,
+      "loss": 3.6776,
+      "step": 7434
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3276731529528594e-07,
+      "loss": 3.6317,
+      "step": 7435
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.320500747222677e-07,
+      "loss": 3.7871,
+      "step": 7436
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.31333581398256e-07,
+      "loss": 3.6612,
+      "step": 7437
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3061783543794744e-07,
+      "loss": 3.619,
+      "step": 7438
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2990283695591927e-07,
+      "loss": 3.6532,
+      "step": 7439
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.291885860666294e-07,
+      "loss": 3.5218,
+      "step": 7440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.284750828844152e-07,
+      "loss": 3.7292,
+      "step": 7441
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2776232752349536e-07,
+      "loss": 3.6664,
+      "step": 7442
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2705032009796746e-07,
+      "loss": 3.7036,
+      "step": 7443
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.263390607218103e-07,
+      "loss": 3.6402,
+      "step": 7444
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.25628549508884e-07,
+      "loss": 3.7073,
+      "step": 7445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2491878657292643e-07,
+      "loss": 3.5817,
+      "step": 7446
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2420977202755666e-07,
+      "loss": 3.7198,
+      "step": 7447
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.235015059862767e-07,
+      "loss": 3.6853,
+      "step": 7448
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2279398856246313e-07,
+      "loss": 3.682,
+      "step": 7449
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.220872198693792e-07,
+      "loss": 3.6997,
+      "step": 7450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.213812000201616e-07,
+      "loss": 3.704,
+      "step": 7451
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.206759291278333e-07,
+      "loss": 3.7635,
+      "step": 7452
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.199714073052923e-07,
+      "loss": 3.6445,
+      "step": 7453
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1926763466532174e-07,
+      "loss": 3.6801,
+      "step": 7454
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.185646113205787e-07,
+      "loss": 3.7359,
+      "step": 7455
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.178623373836076e-07,
+      "loss": 3.6396,
+      "step": 7456
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.171608129668252e-07,
+      "loss": 3.7679,
+      "step": 7457
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.164600381825339e-07,
+      "loss": 3.7162,
+      "step": 7458
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1576001314291506e-07,
+      "loss": 3.7681,
+      "step": 7459
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1506073796002734e-07,
+      "loss": 3.6538,
+      "step": 7460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1436221274581293e-07,
+      "loss": 3.725,
+      "step": 7461
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1366443761209117e-07,
+      "loss": 3.6943,
+      "step": 7462
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.129674126705634e-07,
+      "loss": 3.6636,
+      "step": 7463
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1227113803280863e-07,
+      "loss": 3.6385,
+      "step": 7464
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1157561381028824e-07,
+      "loss": 3.5759,
+      "step": 7465
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1088084011434147e-07,
+      "loss": 3.7045,
+      "step": 7466
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1018681705618835e-07,
+      "loss": 3.5929,
+      "step": 7467
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0949354474692937e-07,
+      "loss": 3.6957,
+      "step": 7468
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.08801023297543e-07,
+      "loss": 3.6446,
+      "step": 7469
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.081092528188895e-07,
+      "loss": 3.8164,
+      "step": 7470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0741823342170687e-07,
+      "loss": 3.642,
+      "step": 7471
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0672796521661663e-07,
+      "loss": 3.6456,
+      "step": 7472
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.060384483141138e-07,
+      "loss": 3.6753,
+      "step": 7473
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0534968282458e-07,
+      "loss": 3.7811,
+      "step": 7474
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.046616688582704e-07,
+      "loss": 3.6837,
+      "step": 7475
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0397440652532585e-07,
+      "loss": 3.6636,
+      "step": 7476
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.03287895935761e-07,
+      "loss": 3.7971,
+      "step": 7477
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0260213719947583e-07,
+      "loss": 3.6048,
+      "step": 7478
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0191713042624414e-07,
+      "loss": 3.656,
+      "step": 7479
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0123287572572545e-07,
+      "loss": 3.6618,
+      "step": 7480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0054937320745216e-07,
+      "loss": 3.7068,
+      "step": 7481
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.998666229808433e-07,
+      "loss": 3.7222,
+      "step": 7482
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.991846251551922e-07,
+      "loss": 3.6698,
+      "step": 7483
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.985033798396736e-07,
+      "loss": 3.6123,
+      "step": 7484
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9782288714334306e-07,
+      "loss": 3.6623,
+      "step": 7485
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.971431471751329e-07,
+      "loss": 3.6765,
+      "step": 7486
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9646416004385725e-07,
+      "loss": 3.7357,
+      "step": 7487
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9578592585820856e-07,
+      "loss": 3.6783,
+      "step": 7488
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.951084447267588e-07,
+      "loss": 3.7066,
+      "step": 7489
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9443171675796076e-07,
+      "loss": 3.6061,
+      "step": 7490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9375574206014444e-07,
+      "loss": 3.8074,
+      "step": 7491
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.930805207415205e-07,
+      "loss": 3.6542,
+      "step": 7492
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9240605291017966e-07,
+      "loss": 3.7577,
+      "step": 7493
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9173233867409057e-07,
+      "loss": 3.6734,
+      "step": 7494
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9105937814110196e-07,
+      "loss": 3.6219,
+      "step": 7495
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9038717141894266e-07,
+      "loss": 3.6482,
+      "step": 7496
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.897157186152183e-07,
+      "loss": 3.6536,
+      "step": 7497
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.890450198374184e-07,
+      "loss": 3.6603,
+      "step": 7498
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8837507519290595e-07,
+      "loss": 3.6609,
+      "step": 7499
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8770588478892805e-07,
+      "loss": 3.7529,
+      "step": 7500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.870374487326083e-07,
+      "loss": 3.7237,
+      "step": 7501
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8636976713095065e-07,
+      "loss": 3.6233,
+      "step": 7502
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8570284009083846e-07,
+      "loss": 3.6483,
+      "step": 7503
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.850366677190336e-07,
+      "loss": 3.7492,
+      "step": 7504
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.843712501221774e-07,
+      "loss": 3.639,
+      "step": 7505
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.837065874067901e-07,
+      "loss": 3.7065,
+      "step": 7506
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.830426796792729e-07,
+      "loss": 3.7593,
+      "step": 7507
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.823795270459029e-07,
+      "loss": 3.7009,
+      "step": 7508
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.817171296128396e-07,
+      "loss": 3.8107,
+      "step": 7509
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8105548748611823e-07,
+      "loss": 3.6438,
+      "step": 7510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.803946007716574e-07,
+      "loss": 3.6654,
+      "step": 7511
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.797344695752491e-07,
+      "loss": 3.7284,
+      "step": 7512
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.790750940025716e-07,
+      "loss": 3.7202,
+      "step": 7513
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.784164741591744e-07,
+      "loss": 3.7357,
+      "step": 7514
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7775861015049256e-07,
+      "loss": 3.6835,
+      "step": 7515
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.771015020818363e-07,
+      "loss": 3.6127,
+      "step": 7516
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7644515005839646e-07,
+      "loss": 3.7031,
+      "step": 7517
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7578955418524233e-07,
+      "loss": 3.6891,
+      "step": 7518
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7513471456732165e-07,
+      "loss": 3.638,
+      "step": 7519
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7448063130946224e-07,
+      "loss": 3.6567,
+      "step": 7520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7382730451636973e-07,
+      "loss": 3.634,
+      "step": 7521
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7317473429262995e-07,
+      "loss": 3.7249,
+      "step": 7522
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7252292074270603e-07,
+      "loss": 3.6451,
+      "step": 7523
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.718718639709411e-07,
+      "loss": 3.7721,
+      "step": 7524
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7122156408155686e-07,
+      "loss": 3.6496,
+      "step": 7525
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.705720211786539e-07,
+      "loss": 3.6057,
+      "step": 7526
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.699232353662118e-07,
+      "loss": 3.7303,
+      "step": 7527
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.69275206748088e-07,
+      "loss": 3.7064,
+      "step": 7528
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.686279354280197e-07,
+      "loss": 3.8003,
+      "step": 7529
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6798142150962215e-07,
+      "loss": 3.6557,
+      "step": 7530
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.673356650963915e-07,
+      "loss": 3.6696,
+      "step": 7531
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.666906662916985e-07,
+      "loss": 3.6049,
+      "step": 7532
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6604642519879766e-07,
+      "loss": 3.6702,
+      "step": 7533
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.654029419208165e-07,
+      "loss": 3.6098,
+      "step": 7534
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6476021656076767e-07,
+      "loss": 3.7112,
+      "step": 7535
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.641182492215361e-07,
+      "loss": 3.6684,
+      "step": 7536
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6347704000589115e-07,
+      "loss": 3.7441,
+      "step": 7537
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.628365890164752e-07,
+      "loss": 3.6812,
+      "step": 7538
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.621968963558147e-07,
+      "loss": 3.6487,
+      "step": 7539
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.615579621263109e-07,
+      "loss": 3.5963,
+      "step": 7540
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.609197864302454e-07,
+      "loss": 3.7219,
+      "step": 7541
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.602823693697776e-07,
+      "loss": 3.7222,
+      "step": 7542
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5964571104694524e-07,
+      "loss": 3.6976,
+      "step": 7543
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.590098115636658e-07,
+      "loss": 3.5416,
+      "step": 7544
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5837467102173387e-07,
+      "loss": 3.5192,
+      "step": 7545
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.577402895228243e-07,
+      "loss": 3.6689,
+      "step": 7546
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5710666716848797e-07,
+      "loss": 3.6432,
+      "step": 7547
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5647380406015665e-07,
+      "loss": 3.6887,
+      "step": 7548
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5584170029913923e-07,
+      "loss": 3.7112,
+      "step": 7549
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.552103559866237e-07,
+      "loss": 3.6316,
+      "step": 7550
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5457977122367526e-07,
+      "loss": 3.751,
+      "step": 7551
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5394994611123934e-07,
+      "loss": 3.7199,
+      "step": 7552
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5332088075013864e-07,
+      "loss": 3.7616,
+      "step": 7553
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.526925752410736e-07,
+      "loss": 3.7545,
+      "step": 7554
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.520650296846255e-07,
+      "loss": 3.7812,
+      "step": 7555
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.514382441812502e-07,
+      "loss": 3.6979,
+      "step": 7556
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5081221883128613e-07,
+      "loss": 3.7224,
+      "step": 7557
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.501869537349455e-07,
+      "loss": 3.7801,
+      "step": 7558
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.495624489923243e-07,
+      "loss": 3.6821,
+      "step": 7559
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.489387047033909e-07,
+      "loss": 3.7156,
+      "step": 7560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4831572096799706e-07,
+      "loss": 3.5789,
+      "step": 7561
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4769349788586803e-07,
+      "loss": 3.6741,
+      "step": 7562
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.470720355566114e-07,
+      "loss": 3.677,
+      "step": 7563
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.464513340797114e-07,
+      "loss": 3.6738,
+      "step": 7564
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.458313935545298e-07,
+      "loss": 3.6141,
+      "step": 7565
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.452122140803076e-07,
+      "loss": 3.6315,
+      "step": 7566
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4459379575616336e-07,
+      "loss": 3.6712,
+      "step": 7567
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.439761386810935e-07,
+      "loss": 3.6876,
+      "step": 7568
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4335924295397385e-07,
+      "loss": 3.6231,
+      "step": 7569
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4274310867355654e-07,
+      "loss": 3.7106,
+      "step": 7570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.421277359384738e-07,
+      "loss": 3.685,
+      "step": 7571
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4151312484723465e-07,
+      "loss": 3.7541,
+      "step": 7572
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4089927549822643e-07,
+      "loss": 3.6918,
+      "step": 7573
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4028618798971446e-07,
+      "loss": 3.6453,
+      "step": 7574
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.396738624198425e-07,
+      "loss": 3.6938,
+      "step": 7575
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.39062298886632e-07,
+      "loss": 3.7934,
+      "step": 7576
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3845149748798256e-07,
+      "loss": 3.7196,
+      "step": 7577
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.378414583216715e-07,
+      "loss": 3.6792,
+      "step": 7578
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.372321814853551e-07,
+      "loss": 3.6821,
+      "step": 7579
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3662366707656537e-07,
+      "loss": 3.7806,
+      "step": 7580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3601591519271606e-07,
+      "loss": 3.7826,
+      "step": 7581
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3540892593109433e-07,
+      "loss": 3.7306,
+      "step": 7582
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3480269938886913e-07,
+      "loss": 3.5966,
+      "step": 7583
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.341972356630845e-07,
+      "loss": 3.6558,
+      "step": 7584
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3359253485066402e-07,
+      "loss": 3.6383,
+      "step": 7585
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3298859704840803e-07,
+      "loss": 3.6942,
+      "step": 7586
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.323854223529959e-07,
+      "loss": 3.6493,
+      "step": 7587
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3178301086098532e-07,
+      "loss": 3.5287,
+      "step": 7588
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3118136266880864e-07,
+      "loss": 3.7255,
+      "step": 7589
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3058047787278103e-07,
+      "loss": 3.7282,
+      "step": 7590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.299803565690889e-07,
+      "loss": 3.7075,
+      "step": 7591
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.293809988538037e-07,
+      "loss": 3.7436,
+      "step": 7592
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2878240482286818e-07,
+      "loss": 3.7142,
+      "step": 7593
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.281845745721084e-07,
+      "loss": 3.7515,
+      "step": 7594
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.275875081972223e-07,
+      "loss": 3.7746,
+      "step": 7595
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2699120579379174e-07,
+      "loss": 3.6956,
+      "step": 7596
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2639566745727203e-07,
+      "loss": 3.7298,
+      "step": 7597
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2580089328299747e-07,
+      "loss": 3.589,
+      "step": 7598
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.252068833661797e-07,
+      "loss": 3.7153,
+      "step": 7599
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.246136378019087e-07,
+      "loss": 3.641,
+      "step": 7600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2402115668515135e-07,
+      "loss": 3.7147,
+      "step": 7601
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2342944011075286e-07,
+      "loss": 3.6983,
+      "step": 7602
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2283848817343523e-07,
+      "loss": 3.6446,
+      "step": 7603
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2224830096779837e-07,
+      "loss": 3.612,
+      "step": 7604
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2165887858832057e-07,
+      "loss": 3.6475,
+      "step": 7605
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2107022112935695e-07,
+      "loss": 3.6334,
+      "step": 7606
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2048232868513986e-07,
+      "loss": 3.7768,
+      "step": 7607
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.198952013497796e-07,
+      "loss": 3.6929,
+      "step": 7608
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1930883921726376e-07,
+      "loss": 3.6176,
+      "step": 7609
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1872324238145837e-07,
+      "loss": 3.6724,
+      "step": 7610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.181384109361051e-07,
+      "loss": 3.6851,
+      "step": 7611
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.175543449748263e-07,
+      "loss": 3.7163,
+      "step": 7612
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1697104459111662e-07,
+      "loss": 3.6579,
+      "step": 7613
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.163885098783547e-07,
+      "loss": 3.6907,
+      "step": 7614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1580674092978982e-07,
+      "loss": 3.6736,
+      "step": 7615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1522573783855473e-07,
+      "loss": 3.703,
+      "step": 7616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1464550069765444e-07,
+      "loss": 3.6322,
+      "step": 7617
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1406602959997636e-07,
+      "loss": 3.6078,
+      "step": 7618
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1348732463827958e-07,
+      "loss": 3.7292,
+      "step": 7619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.129093859052067e-07,
+      "loss": 3.6737,
+      "step": 7620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.123322134932726e-07,
+      "loss": 3.8039,
+      "step": 7621
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.117558074948728e-07,
+      "loss": 3.5875,
+      "step": 7622
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1118016800227737e-07,
+      "loss": 3.6456,
+      "step": 7623
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.106052951076365e-07,
+      "loss": 3.6934,
+      "step": 7624
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1003118890297548e-07,
+      "loss": 3.6195,
+      "step": 7625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.09457849480198e-07,
+      "loss": 3.8047,
+      "step": 7626
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0888527693108395e-07,
+      "loss": 3.6792,
+      "step": 7627
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.083134713472923e-07,
+      "loss": 3.6536,
+      "step": 7628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0774243282035756e-07,
+      "loss": 3.7898,
+      "step": 7629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0717216144169216e-07,
+      "loss": 3.6942,
+      "step": 7630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0660265730258533e-07,
+      "loss": 3.7205,
+      "step": 7631
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0603392049420357e-07,
+      "loss": 3.6946,
+      "step": 7632
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0546595110759127e-07,
+      "loss": 3.7059,
+      "step": 7633
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0489874923366958e-07,
+      "loss": 3.64,
+      "step": 7634
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.043323149632348e-07,
+      "loss": 3.6643,
+      "step": 7635
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0376664838696546e-07,
+      "loss": 3.7462,
+      "step": 7636
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0320174959541028e-07,
+      "loss": 3.6795,
+      "step": 7637
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.026376186790019e-07,
+      "loss": 3.6929,
+      "step": 7638
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0207425572804362e-07,
+      "loss": 3.6945,
+      "step": 7639
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0151166083272222e-07,
+      "loss": 3.6257,
+      "step": 7640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0094983408309564e-07,
+      "loss": 3.6598,
+      "step": 7641
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.003887755691042e-07,
+      "loss": 3.6653,
+      "step": 7642
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9982848538055932e-07,
+      "loss": 3.784,
+      "step": 7643
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9926896360715542e-07,
+      "loss": 3.6527,
+      "step": 7644
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9871021033846026e-07,
+      "loss": 3.7312,
+      "step": 7645
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9815222566391957e-07,
+      "loss": 3.7189,
+      "step": 7646
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9759500967285518e-07,
+      "loss": 3.6499,
+      "step": 7647
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9703856245446795e-07,
+      "loss": 3.6891,
+      "step": 7648
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9648288409783334e-07,
+      "loss": 3.8991,
+      "step": 7649
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.959279746919057e-07,
+      "loss": 3.669,
+      "step": 7650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9537383432551404e-07,
+      "loss": 3.6774,
+      "step": 7651
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.948204630873668e-07,
+      "loss": 3.7105,
+      "step": 7652
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9426786106604755e-07,
+      "loss": 3.767,
+      "step": 7653
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.937160283500178e-07,
+      "loss": 3.6849,
+      "step": 7654
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9316496502761407e-07,
+      "loss": 3.687,
+      "step": 7655
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9261467118705246e-07,
+      "loss": 3.7129,
+      "step": 7656
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9206514691642363e-07,
+      "loss": 3.6859,
+      "step": 7657
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9151639230369556e-07,
+      "loss": 3.6014,
+      "step": 7658
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9096840743671353e-07,
+      "loss": 3.6722,
+      "step": 7659
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9042119240320067e-07,
+      "loss": 3.6683,
+      "step": 7660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8987474729075362e-07,
+      "loss": 3.7411,
+      "step": 7661
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.893290721868496e-07,
+      "loss": 3.6833,
+      "step": 7662
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.887841671788382e-07,
+      "loss": 3.652,
+      "step": 7663
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8824003235395128e-07,
+      "loss": 3.7403,
+      "step": 7664
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8769666779929142e-07,
+      "loss": 3.6776,
+      "step": 7665
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8715407360184402e-07,
+      "loss": 3.775,
+      "step": 7666
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.866122498484646e-07,
+      "loss": 3.5813,
+      "step": 7667
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8607119662589045e-07,
+      "loss": 3.728,
+      "step": 7668
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8553091402073452e-07,
+      "loss": 3.7211,
+      "step": 7669
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.849914021194843e-07,
+      "loss": 3.7051,
+      "step": 7670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8445266100850623e-07,
+      "loss": 3.6181,
+      "step": 7671
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8391469077404134e-07,
+      "loss": 3.6493,
+      "step": 7672
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8337749150220962e-07,
+      "loss": 3.7317,
+      "step": 7673
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8284106327900453e-07,
+      "loss": 3.745,
+      "step": 7674
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8230540619030012e-07,
+      "loss": 3.6013,
+      "step": 7675
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8177052032184285e-07,
+      "loss": 3.6255,
+      "step": 7676
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.812364057592586e-07,
+      "loss": 3.6956,
+      "step": 7677
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8070306258804904e-07,
+      "loss": 3.693,
+      "step": 7678
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8017049089359195e-07,
+      "loss": 3.6317,
+      "step": 7679
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7963869076114193e-07,
+      "loss": 3.8415,
+      "step": 7680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7910766227582922e-07,
+      "loss": 3.7191,
+      "step": 7681
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.785774055226619e-07,
+      "loss": 3.6351,
+      "step": 7682
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7804792058652376e-07,
+      "loss": 3.6643,
+      "step": 7683
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7751920755217532e-07,
+      "loss": 3.6721,
+      "step": 7684
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7699126650425336e-07,
+      "loss": 3.8073,
+      "step": 7685
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7646409752727078e-07,
+      "loss": 3.6397,
+      "step": 7686
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7593770070561732e-07,
+      "loss": 3.6426,
+      "step": 7687
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7541207612355894e-07,
+      "loss": 3.6169,
+      "step": 7688
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7488722386523826e-07,
+      "loss": 3.7269,
+      "step": 7689
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7436314401467369e-07,
+      "loss": 3.7138,
+      "step": 7690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7383983665576088e-07,
+      "loss": 3.6833,
+      "step": 7691
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7331730187227003e-07,
+      "loss": 3.6546,
+      "step": 7692
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.727955397478509e-07,
+      "loss": 3.6338,
+      "step": 7693
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7227455036602558e-07,
+      "loss": 3.6425,
+      "step": 7694
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7175433381019623e-07,
+      "loss": 3.7413,
+      "step": 7695
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7123489016363793e-07,
+      "loss": 3.6078,
+      "step": 7696
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7071621950950467e-07,
+      "loss": 3.7145,
+      "step": 7697
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7019832193082507e-07,
+      "loss": 3.5986,
+      "step": 7698
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6968119751050506e-07,
+      "loss": 3.5237,
+      "step": 7699
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6916484633132558e-07,
+      "loss": 3.7276,
+      "step": 7700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.68649268475945e-07,
+      "loss": 3.6157,
+      "step": 7701
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6813446402689783e-07,
+      "loss": 3.6781,
+      "step": 7702
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.676204330665937e-07,
+      "loss": 3.5528,
+      "step": 7703
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.671071756773196e-07,
+      "loss": 3.6175,
+      "step": 7704
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6659469194123755e-07,
+      "loss": 3.7441,
+      "step": 7705
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6608298194038697e-07,
+      "loss": 3.6562,
+      "step": 7706
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.655720457566823e-07,
+      "loss": 3.7182,
+      "step": 7707
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6506188347191477e-07,
+      "loss": 3.7506,
+      "step": 7708
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6455249516775186e-07,
+      "loss": 3.5931,
+      "step": 7709
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.640438809257361e-07,
+      "loss": 3.6399,
+      "step": 7710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6353604082728792e-07,
+      "loss": 3.6068,
+      "step": 7711
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6302897495370175e-07,
+      "loss": 3.6779,
+      "step": 7712
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625226833861493e-07,
+      "loss": 3.7382,
+      "step": 7713
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6201716620567853e-07,
+      "loss": 3.634,
+      "step": 7714
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.615124234932125e-07,
+      "loss": 3.6477,
+      "step": 7715
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.61008455329551e-07,
+      "loss": 3.766,
+      "step": 7716
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6050526179537062e-07,
+      "loss": 3.6184,
+      "step": 7717
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.600028429712208e-07,
+      "loss": 3.5964,
+      "step": 7718
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.595011989375317e-07,
+      "loss": 3.6,
+      "step": 7719
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.59000329774604e-07,
+      "loss": 3.6924,
+      "step": 7720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5850023556262028e-07,
+      "loss": 3.8253,
+      "step": 7721
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5800091638163374e-07,
+      "loss": 3.674,
+      "step": 7722
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5750237231157705e-07,
+      "loss": 3.7315,
+      "step": 7723
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5700460343225644e-07,
+      "loss": 3.8141,
+      "step": 7724
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.565076098233559e-07,
+      "loss": 3.7422,
+      "step": 7725
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5601139156443456e-07,
+      "loss": 3.7026,
+      "step": 7726
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5551594873492726e-07,
+      "loss": 3.7675,
+      "step": 7727
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5502128141414496e-07,
+      "loss": 3.692,
+      "step": 7728
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.545273896812738e-07,
+      "loss": 3.6276,
+      "step": 7729
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5403427361537714e-07,
+      "loss": 3.8178,
+      "step": 7730
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.53541933295393e-07,
+      "loss": 3.6814,
+      "step": 7731
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5305036880013612e-07,
+      "loss": 3.6507,
+      "step": 7732
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5255958020829575e-07,
+      "loss": 3.6172,
+      "step": 7733
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5206956759843794e-07,
+      "loss": 3.6359,
+      "step": 7734
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5158033104900437e-07,
+      "loss": 3.6542,
+      "step": 7735
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5109187063831243e-07,
+      "loss": 3.5989,
+      "step": 7736
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.506041864445551e-07,
+      "loss": 3.5796,
+      "step": 7737
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5011727854580105e-07,
+      "loss": 3.5935,
+      "step": 7738
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.496311470199957e-07,
+      "loss": 3.6754,
+      "step": 7739
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4914579194495794e-07,
+      "loss": 3.6539,
+      "step": 7740
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4866121339838613e-07,
+      "loss": 3.6702,
+      "step": 7741
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.481774114578488e-07,
+      "loss": 3.6712,
+      "step": 7742
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4769438620079678e-07,
+      "loss": 3.726,
+      "step": 7743
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4721213770454988e-07,
+      "loss": 3.6004,
+      "step": 7744
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4673066604630914e-07,
+      "loss": 3.632,
+      "step": 7745
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4624997130314788e-07,
+      "loss": 3.6695,
+      "step": 7746
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.457700535520168e-07,
+      "loss": 3.7206,
+      "step": 7747
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4529091286973994e-07,
+      "loss": 3.6227,
+      "step": 7748
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4481254933301992e-07,
+      "loss": 3.5688,
+      "step": 7749
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4433496301843318e-07,
+      "loss": 3.6621,
+      "step": 7750
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4385815400243254e-07,
+      "loss": 3.7074,
+      "step": 7751
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4338212236134518e-07,
+      "loss": 3.6263,
+      "step": 7752
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4290686817137522e-07,
+      "loss": 3.6917,
+      "step": 7753
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4243239150860123e-07,
+      "loss": 3.7063,
+      "step": 7754
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4195869244897797e-07,
+      "loss": 3.8053,
+      "step": 7755
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.414857710683365e-07,
+      "loss": 3.713,
+      "step": 7756
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4101362744238013e-07,
+      "loss": 3.6036,
+      "step": 7757
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4054226164669172e-07,
+      "loss": 3.7314,
+      "step": 7758
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4007167375672813e-07,
+      "loss": 3.5608,
+      "step": 7759
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3960186384782025e-07,
+      "loss": 3.6263,
+      "step": 7760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3913283199517623e-07,
+      "loss": 3.7106,
+      "step": 7761
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3866457827387935e-07,
+      "loss": 3.6609,
+      "step": 7762
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3819710275888742e-07,
+      "loss": 3.7398,
+      "step": 7763
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3773040552503447e-07,
+      "loss": 3.6241,
+      "step": 7764
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3726448664703017e-07,
+      "loss": 3.6773,
+      "step": 7765
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3679934619945822e-07,
+      "loss": 3.6493,
+      "step": 7766
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3633498425678016e-07,
+      "loss": 3.5869,
+      "step": 7767
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3587140089332984e-07,
+      "loss": 3.7099,
+      "step": 7768
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3540859618331848e-07,
+      "loss": 3.693,
+      "step": 7769
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3494657020083291e-07,
+      "loss": 3.6166,
+      "step": 7770
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3448532301983397e-07,
+      "loss": 3.6586,
+      "step": 7771
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3402485471415871e-07,
+      "loss": 3.6242,
+      "step": 7772
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3356516535751875e-07,
+      "loss": 3.7255,
+      "step": 7773
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3310625502350294e-07,
+      "loss": 3.6714,
+      "step": 7774
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3264812378557145e-07,
+      "loss": 3.6591,
+      "step": 7775
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3219077171706563e-07,
+      "loss": 3.6186,
+      "step": 7776
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3173419889119522e-07,
+      "loss": 3.7407,
+      "step": 7777
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3127840538105175e-07,
+      "loss": 3.5844,
+      "step": 7778
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3082339125959685e-07,
+      "loss": 3.7242,
+      "step": 7779
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.303691565996712e-07,
+      "loss": 3.8606,
+      "step": 7780
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2991570147398768e-07,
+      "loss": 3.7485,
+      "step": 7781
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2946302595513604e-07,
+      "loss": 3.6809,
+      "step": 7782
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2901113011558163e-07,
+      "loss": 3.6355,
+      "step": 7783
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2856001402766384e-07,
+      "loss": 3.7345,
+      "step": 7784
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2810967776359763e-07,
+      "loss": 3.656,
+      "step": 7785
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2766012139547313e-07,
+      "loss": 3.6963,
+      "step": 7786
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2721134499525555e-07,
+      "loss": 3.6853,
+      "step": 7787
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.267633486347858e-07,
+      "loss": 3.6412,
+      "step": 7788
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2631613238577923e-07,
+      "loss": 3.6586,
+      "step": 7789
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2586969631982636e-07,
+      "loss": 3.6613,
+      "step": 7790
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.254240405083934e-07,
+      "loss": 3.5724,
+      "step": 7791
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2497916502282104e-07,
+      "loss": 3.6764,
+      "step": 7792
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2453506993432507e-07,
+      "loss": 3.7068,
+      "step": 7793
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2409175531399697e-07,
+      "loss": 3.7025,
+      "step": 7794
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.236492212328033e-07,
+      "loss": 3.6327,
+      "step": 7795
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.232074677615841e-07,
+      "loss": 3.6408,
+      "step": 7796
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.227664949710561e-07,
+      "loss": 3.638,
+      "step": 7797
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2232630293181114e-07,
+      "loss": 3.6603,
+      "step": 7798
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2188689171431457e-07,
+      "loss": 3.6884,
+      "step": 7799
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2144826138890897e-07,
+      "loss": 3.7797,
+      "step": 7800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2101041202580876e-07,
+      "loss": 3.7693,
+      "step": 7801
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2057334369510732e-07,
+      "loss": 3.6812,
+      "step": 7802
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.201370564667692e-07,
+      "loss": 3.6723,
+      "step": 7803
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1970155041063636e-07,
+      "loss": 3.6401,
+      "step": 7804
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.192668255964241e-07,
+      "loss": 3.6816,
+      "step": 7805
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1883288209372512e-07,
+      "loss": 3.6663,
+      "step": 7806
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1839971997200383e-07,
+      "loss": 3.6609,
+      "step": 7807
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1796733930060256e-07,
+      "loss": 3.7577,
+      "step": 7808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1753574014873593e-07,
+      "loss": 3.6792,
+      "step": 7809
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1710492258549478e-07,
+      "loss": 3.5751,
+      "step": 7810
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1667488667984561e-07,
+      "loss": 3.6225,
+      "step": 7811
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1624563250062836e-07,
+      "loss": 3.6342,
+      "step": 7812
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1581716011655808e-07,
+      "loss": 3.6069,
+      "step": 7813
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1538946959622543e-07,
+      "loss": 3.6775,
+      "step": 7814
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1496256100809511e-07,
+      "loss": 3.6548,
+      "step": 7815
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1453643442050744e-07,
+      "loss": 3.5891,
+      "step": 7816
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1411108990167618e-07,
+      "loss": 3.6573,
+      "step": 7817
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1368652751969189e-07,
+      "loss": 3.6859,
+      "step": 7818
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1326274734251796e-07,
+      "loss": 3.6812,
+      "step": 7819
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1283974943799403e-07,
+      "loss": 3.6977,
+      "step": 7820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1241753387383314e-07,
+      "loss": 3.6679,
+      "step": 7821
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1199610071762567e-07,
+      "loss": 3.5773,
+      "step": 7822
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1157545003683268e-07,
+      "loss": 3.8217,
+      "step": 7823
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1115558189879416e-07,
+      "loss": 3.7951,
+      "step": 7824
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1073649637072137e-07,
+      "loss": 3.7348,
+      "step": 7825
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1031819351970341e-07,
+      "loss": 3.5786,
+      "step": 7826
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0990067341270061e-07,
+      "loss": 3.6382,
+      "step": 7827
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0948393611655229e-07,
+      "loss": 3.6628,
+      "step": 7828
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0906798169796729e-07,
+      "loss": 3.7766,
+      "step": 7829
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0865281022353457e-07,
+      "loss": 3.6425,
+      "step": 7830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0823842175971322e-07,
+      "loss": 3.6484,
+      "step": 7831
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0782481637284014e-07,
+      "loss": 3.7714,
+      "step": 7832
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0741199412912461e-07,
+      "loss": 3.7276,
+      "step": 7833
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0699995509465211e-07,
+      "loss": 3.6968,
+      "step": 7834
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0658869933538207e-07,
+      "loss": 3.6561,
+      "step": 7835
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0617822691714796e-07,
+      "loss": 3.6565,
+      "step": 7836
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0576853790566e-07,
+      "loss": 3.5469,
+      "step": 7837
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0535963236650016e-07,
+      "loss": 3.6372,
+      "step": 7838
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0495151036512664e-07,
+      "loss": 3.6568,
+      "step": 7839
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0454417196687216e-07,
+      "loss": 3.6567,
+      "step": 7840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0413761723694405e-07,
+      "loss": 3.733,
+      "step": 7841
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0373184624042354e-07,
+      "loss": 3.5851,
+      "step": 7842
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0332685904226647e-07,
+      "loss": 3.6022,
+      "step": 7843
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0292265570730431e-07,
+      "loss": 3.6867,
+      "step": 7844
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.025192363002414e-07,
+      "loss": 3.8039,
+      "step": 7845
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0211660088565777e-07,
+      "loss": 3.5384,
+      "step": 7846
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0171474952800798e-07,
+      "loss": 3.6805,
+      "step": 7847
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0131368229161997e-07,
+      "loss": 3.6077,
+      "step": 7848
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0091339924069743e-07,
+      "loss": 3.682,
+      "step": 7849
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0051390043931741e-07,
+      "loss": 3.6331,
+      "step": 7850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0011518595143266e-07,
+      "loss": 3.6424,
+      "step": 7851
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.971725584086933e-08,
+      "loss": 3.6789,
+      "step": 7852
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.93201101713287e-08,
+      "loss": 3.5994,
+      "step": 7853
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.892374900638546e-08,
+      "loss": 3.6335,
+      "step": 7854
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.852817240949053e-08,
+      "loss": 3.7233,
+      "step": 7855
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.813338044396714e-08,
+      "loss": 3.7058,
+      "step": 7856
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.773937317301474e-08,
+      "loss": 3.6894,
+      "step": 7857
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.734615065970454e-08,
+      "loss": 3.7709,
+      "step": 7858
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.695371296698619e-08,
+      "loss": 3.5942,
+      "step": 7859
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.656206015768e-08,
+      "loss": 3.8411,
+      "step": 7860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.617119229448302e-08,
+      "loss": 3.627,
+      "step": 7861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.578110943996411e-08,
+      "loss": 3.6851,
+      "step": 7862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.539181165656941e-08,
+      "loss": 3.6677,
+      "step": 7863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.500329900661742e-08,
+      "loss": 3.6912,
+      "step": 7864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.461557155230227e-08,
+      "loss": 3.8052,
+      "step": 7865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.422862935569099e-08,
+      "loss": 3.6982,
+      "step": 7866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.384247247872513e-08,
+      "loss": 3.7065,
+      "step": 7867
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.345710098322247e-08,
+      "loss": 3.6312,
+      "step": 7868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.307251493087199e-08,
+      "loss": 3.5529,
+      "step": 7869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.268871438323945e-08,
+      "loss": 3.6755,
+      "step": 7870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.230569940176348e-08,
+      "loss": 3.7209,
+      "step": 7871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.192347004775781e-08,
+      "loss": 3.6377,
+      "step": 7872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.154202638241016e-08,
+      "loss": 3.6175,
+      "step": 7873
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.11613684667817e-08,
+      "loss": 3.6489,
+      "step": 7874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.078149636180866e-08,
+      "loss": 3.6208,
+      "step": 7875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.040241012830131e-08,
+      "loss": 3.8813,
+      "step": 7876
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.002410982694388e-08,
+      "loss": 3.6298,
+      "step": 7877
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.964659551829513e-08,
+      "loss": 3.6537,
+      "step": 7878
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.926986726278897e-08,
+      "loss": 3.7059,
+      "step": 7879
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.889392512072992e-08,
+      "loss": 3.6231,
+      "step": 7880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.851876915230151e-08,
+      "loss": 3.6364,
+      "step": 7881
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.814439941755682e-08,
+      "loss": 3.7185,
+      "step": 7882
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.777081597642733e-08,
+      "loss": 3.6602,
+      "step": 7883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.739801888871468e-08,
+      "loss": 3.7126,
+      "step": 7884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.702600821409834e-08,
+      "loss": 3.5079,
+      "step": 7885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.665478401212845e-08,
+      "loss": 3.708,
+      "step": 7886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.628434634223193e-08,
+      "loss": 3.7514,
+      "step": 7887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.591469526370799e-08,
+      "loss": 3.751,
+      "step": 7888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.554583083573098e-08,
+      "loss": 3.6689,
+      "step": 7889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.517775311734922e-08,
+      "loss": 3.6861,
+      "step": 7890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.481046216748501e-08,
+      "loss": 3.629,
+      "step": 7891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.444395804493411e-08,
+      "loss": 3.7268,
+      "step": 7892
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.40782408083668e-08,
+      "loss": 3.711,
+      "step": 7893
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.371331051632736e-08,
+      "loss": 3.7614,
+      "step": 7894
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.334916722723407e-08,
+      "loss": 3.7822,
+      "step": 7895
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.298581099937975e-08,
+      "loss": 3.65,
+      "step": 7896
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.262324189093063e-08,
+      "loss": 3.7357,
+      "step": 7897
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.226145995992641e-08,
+      "loss": 3.6891,
+      "step": 7898
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.190046526428241e-08,
+      "loss": 3.627,
+      "step": 7899
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.154025786178577e-08,
+      "loss": 3.7463,
+      "step": 7900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.118083781009978e-08,
+      "loss": 3.6476,
+      "step": 7901
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.082220516675954e-08,
+      "loss": 3.5715,
+      "step": 7902
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.046435998917745e-08,
+      "loss": 3.7369,
+      "step": 7903
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.010730233463493e-08,
+      "loss": 3.676,
+      "step": 7904
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.975103226029179e-08,
+      "loss": 3.6619,
+      "step": 7905
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.93955498231791e-08,
+      "loss": 3.6402,
+      "step": 7906
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.90408550802041e-08,
+      "loss": 3.6104,
+      "step": 7907
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.86869480881447e-08,
+      "loss": 3.7105,
+      "step": 7908
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.833382890365615e-08,
+      "loss": 3.6377,
+      "step": 7909
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.79814975832649e-08,
+      "loss": 3.6737,
+      "step": 7910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.76299541833736e-08,
+      "loss": 3.7209,
+      "step": 7911
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.727919876025669e-08,
+      "loss": 3.6451,
+      "step": 7912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.69292313700637e-08,
+      "loss": 3.7552,
+      "step": 7913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.658005206881758e-08,
+      "loss": 3.651,
+      "step": 7914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.623166091241529e-08,
+      "loss": 3.6355,
+      "step": 7915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.588405795662779e-08,
+      "loss": 3.6397,
+      "step": 7916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.553724325709888e-08,
+      "loss": 3.7323,
+      "step": 7917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.519121686934805e-08,
+      "loss": 3.7317,
+      "step": 7918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.484597884876598e-08,
+      "loss": 3.6864,
+      "step": 7919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.450152925062015e-08,
+      "loss": 3.7468,
+      "step": 7920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.415786813004977e-08,
+      "loss": 3.6597,
+      "step": 7921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.381499554206806e-08,
+      "loss": 3.7023,
+      "step": 7922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.347291154156222e-08,
+      "loss": 3.6897,
+      "step": 7923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.3131616183294e-08,
+      "loss": 3.634,
+      "step": 7924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.27911095218975e-08,
+      "loss": 3.7778,
+      "step": 7925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.245139161188186e-08,
+      "loss": 3.7727,
+      "step": 7926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.211246250762916e-08,
+      "loss": 3.7164,
+      "step": 7927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.177432226339542e-08,
+      "loss": 3.7298,
+      "step": 7928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.143697093331015e-08,
+      "loss": 3.6078,
+      "step": 7929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.110040857137734e-08,
+      "loss": 3.7313,
+      "step": 7930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.076463523147392e-08,
+      "loss": 3.6666,
+      "step": 7931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.042965096735076e-08,
+      "loss": 3.6427,
+      "step": 7932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.009545583263223e-08,
+      "loss": 3.71,
+      "step": 7933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.976204988081658e-08,
+      "loss": 3.668,
+      "step": 7934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.942943316527561e-08,
+      "loss": 3.7824,
+      "step": 7935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.909760573925561e-08,
+      "loss": 3.741,
+      "step": 7936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.876656765587409e-08,
+      "loss": 3.5446,
+      "step": 7937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.843631896812641e-08,
+      "loss": 3.7102,
+      "step": 7938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.810685972887642e-08,
+      "loss": 3.6212,
+      "step": 7939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.777818999086582e-08,
+      "loss": 3.6903,
+      "step": 7940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.745030980670753e-08,
+      "loss": 3.6399,
+      "step": 7941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.712321922889009e-08,
+      "loss": 3.6917,
+      "step": 7942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.67969183097722e-08,
+      "loss": 3.5691,
+      "step": 7943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.647140710159039e-08,
+      "loss": 3.5831,
+      "step": 7944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.614668565645244e-08,
+      "loss": 3.7407,
+      "step": 7945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.582275402633898e-08,
+      "loss": 3.6202,
+      "step": 7946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.549961226310631e-08,
+      "loss": 3.6565,
+      "step": 7947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.51772604184825e-08,
+      "loss": 3.6387,
+      "step": 7948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.485569854407014e-08,
+      "loss": 3.6958,
+      "step": 7949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.453492669134476e-08,
+      "loss": 3.7513,
+      "step": 7950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.421494491165692e-08,
+      "loss": 3.5517,
+      "step": 7951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.389575325622787e-08,
+      "loss": 3.7697,
+      "step": 7952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.357735177615566e-08,
+      "loss": 3.71,
+      "step": 7953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.32597405224089e-08,
+      "loss": 3.6364,
+      "step": 7954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.294291954583198e-08,
+      "loss": 3.6597,
+      "step": 7955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.262688889714152e-08,
+      "loss": 3.782,
+      "step": 7956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.231164862692763e-08,
+      "loss": 3.597,
+      "step": 7957
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.199719878565491e-08,
+      "loss": 3.684,
+      "step": 7958
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.168353942365979e-08,
+      "loss": 3.6902,
+      "step": 7959
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.137067059115431e-08,
+      "loss": 3.7515,
+      "step": 7960
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.105859233822176e-08,
+      "loss": 3.8047,
+      "step": 7961
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.074730471482049e-08,
+      "loss": 3.7241,
+      "step": 7962
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.043680777078065e-08,
+      "loss": 3.6367,
+      "step": 7963
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.012710155580858e-08,
+      "loss": 3.6924,
+      "step": 7964
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.98181861194802e-08,
+      "loss": 3.5718,
+      "step": 7965
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.951006151124927e-08,
+      "loss": 3.7132,
+      "step": 7966
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.9202727780438006e-08,
+      "loss": 3.6361,
+      "step": 7967
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.889618497624649e-08,
+      "loss": 3.6321,
+      "step": 7968
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.859043314774604e-08,
+      "loss": 3.6798,
+      "step": 7969
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.8285472343881374e-08,
+      "loss": 3.7434,
+      "step": 7970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.798130261347068e-08,
+      "loss": 3.6699,
+      "step": 7971
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.767792400520556e-08,
+      "loss": 3.696,
+      "step": 7972
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.737533656765215e-08,
+      "loss": 3.651,
+      "step": 7973
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.7073540349247835e-08,
+      "loss": 3.701,
+      "step": 7974
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.677253539830452e-08,
+      "loss": 3.7042,
+      "step": 7975
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.647232176300754e-08,
+      "loss": 3.7204,
+      "step": 7976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.617289949141569e-08,
+      "loss": 3.5577,
+      "step": 7977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.58742686314595e-08,
+      "loss": 3.7156,
+      "step": 7978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.557642923094575e-08,
+      "loss": 3.6995,
+      "step": 7979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.5279381337551286e-08,
+      "loss": 3.6095,
+      "step": 7980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.498312499882808e-08,
+      "loss": 3.6032,
+      "step": 7981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.468766026220207e-08,
+      "loss": 3.7306,
+      "step": 7982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.439298717496988e-08,
+      "loss": 3.634,
+      "step": 7983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.409910578430488e-08,
+      "loss": 3.7508,
+      "step": 7984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.3806016137250007e-08,
+      "loss": 3.6495,
+      "step": 7985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.351371828072494e-08,
+      "loss": 3.6619,
+      "step": 7986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.3222212261518914e-08,
+      "loss": 3.7507,
+      "step": 7987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.2931498126298495e-08,
+      "loss": 3.611,
+      "step": 7988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.264157592160035e-08,
+      "loss": 3.6718,
+      "step": 7989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.235244569383569e-08,
+      "loss": 3.7486,
+      "step": 7990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.206410748928858e-08,
+      "loss": 3.6086,
+      "step": 7991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.177656135411657e-08,
+      "loss": 3.7475,
+      "step": 7992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.148980733435005e-08,
+      "loss": 3.6732,
+      "step": 7993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1203845475892855e-08,
+      "loss": 3.6457,
+      "step": 7994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.0918675824522815e-08,
+      "loss": 3.5992,
+      "step": 7995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.063429842588841e-08,
+      "loss": 3.714,
+      "step": 7996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.035071332551489e-08,
+      "loss": 3.7293,
+      "step": 7997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.006792056879706e-08,
+      "loss": 3.7139,
+      "step": 7998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.978592020100592e-08,
+      "loss": 3.6151,
+      "step": 7999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.950471226728371e-08,
+      "loss": 3.6785,
+      "step": 8000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.922429681264662e-08,
+      "loss": 3.6503,
+      "step": 8001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8944673881983205e-08,
+      "loss": 3.6849,
+      "step": 8002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8665843520056544e-08,
+      "loss": 3.6842,
+      "step": 8003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.838780577150093e-08,
+      "loss": 3.7875,
+      "step": 8004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8110560680826314e-08,
+      "loss": 3.657,
+      "step": 8005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.783410829241331e-08,
+      "loss": 3.7619,
+      "step": 8006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.755844865051651e-08,
+      "loss": 3.7157,
+      "step": 8007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.728358179926451e-08,
+      "loss": 3.753,
+      "step": 8008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.700950778265767e-08,
+      "loss": 3.7763,
+      "step": 8009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6736226644570336e-08,
+      "loss": 3.648,
+      "step": 8010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.646373842874863e-08,
+      "loss": 3.5949,
+      "step": 8011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.619204317881376e-08,
+      "loss": 3.7506,
+      "step": 8012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.592114093825872e-08,
+      "loss": 3.6563,
+      "step": 8013
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.565103175044883e-08,
+      "loss": 3.6857,
+      "step": 8014
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.538171565862448e-08,
+      "loss": 3.5947,
+      "step": 8015
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.511319270589731e-08,
+      "loss": 3.6168,
+      "step": 8016
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.484546293525349e-08,
+      "loss": 3.6576,
+      "step": 8017
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4578526389550406e-08,
+      "loss": 3.6084,
+      "step": 8018
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4312383111519974e-08,
+      "loss": 3.6987,
+      "step": 8019
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.404703314376646e-08,
+      "loss": 3.6727,
+      "step": 8020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.37824765287681e-08,
+      "loss": 3.6951,
+      "step": 8021
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3518713308873805e-08,
+      "loss": 3.8315,
+      "step": 8022
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3255743526308125e-08,
+      "loss": 3.7456,
+      "step": 8023
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.299356722316683e-08,
+      "loss": 3.7359,
+      "step": 8024
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.273218444141969e-08,
+      "loss": 3.5435,
+      "step": 8025
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.247159522290933e-08,
+      "loss": 3.6613,
+      "step": 8026
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.221179960935074e-08,
+      "loss": 3.6855,
+      "step": 8027
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1952797642331736e-08,
+      "loss": 3.6355,
+      "step": 8028
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1694589363314165e-08,
+      "loss": 3.6607,
+      "step": 8029
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1437174813632187e-08,
+      "loss": 3.6725,
+      "step": 8030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1180554034492836e-08,
+      "loss": 3.6407,
+      "step": 8031
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.092472706697603e-08,
+      "loss": 3.5762,
+      "step": 8032
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0669693952035126e-08,
+      "loss": 3.6854,
+      "step": 8033
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0415454730495243e-08,
+      "loss": 3.6532,
+      "step": 8034
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.016200944305604e-08,
+      "loss": 3.621,
+      "step": 8035
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.99093581302884e-08,
+      "loss": 3.7151,
+      "step": 8036
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.965750083263775e-08,
+      "loss": 3.7294,
+      "step": 8037
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9406437590421284e-08,
+      "loss": 3.676,
+      "step": 8038
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.915616844382908e-08,
+      "loss": 3.6191,
+      "step": 8039
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.890669343292464e-08,
+      "loss": 3.6551,
+      "step": 8040
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8658012597644344e-08,
+      "loss": 3.6833,
+      "step": 8041
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.841012597779692e-08,
+      "loss": 3.7042,
+      "step": 8042
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.816303361306506e-08,
+      "loss": 3.7114,
+      "step": 8043
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.791673554300157e-08,
+      "loss": 3.7319,
+      "step": 8044
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.767123180703658e-08,
+      "loss": 3.6456,
+      "step": 8045
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.742652244446809e-08,
+      "loss": 3.6576,
+      "step": 8046
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7182607494471444e-08,
+      "loss": 3.6983,
+      "step": 8047
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6939486996090953e-08,
+      "loss": 3.7299,
+      "step": 8048
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6697160988246607e-08,
+      "loss": 3.6314,
+      "step": 8049
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.645562950973014e-08,
+      "loss": 3.6322,
+      "step": 8050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.621489259920563e-08,
+      "loss": 3.7159,
+      "step": 8051
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.597495029521059e-08,
+      "loss": 3.6611,
+      "step": 8052
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.573580263615539e-08,
+      "loss": 3.6593,
+      "step": 8053
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5497449660322756e-08,
+      "loss": 3.6771,
+      "step": 8054
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.525989140586827e-08,
+      "loss": 3.7934,
+      "step": 8055
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5023127910820966e-08,
+      "loss": 3.6065,
+      "step": 8056
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.478715921308162e-08,
+      "loss": 3.6604,
+      "step": 8057
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4551985350424474e-08,
+      "loss": 3.6914,
+      "step": 8058
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.431760636049608e-08,
+      "loss": 3.7372,
+      "step": 8059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.408402228081642e-08,
+      "loss": 3.791,
+      "step": 8060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3851233148777805e-08,
+      "loss": 3.708,
+      "step": 8061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3619239001644877e-08,
+      "loss": 3.6509,
+      "step": 8062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.33880398765557e-08,
+      "loss": 3.6686,
+      "step": 8063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.315763581052067e-08,
+      "loss": 3.6549,
+      "step": 8064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.292802684042362e-08,
+      "loss": 3.5909,
+      "step": 8065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.269921300301959e-08,
+      "loss": 3.6768,
+      "step": 8066
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.247119433493817e-08,
+      "loss": 3.6682,
+      "step": 8067
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.224397087267961e-08,
+      "loss": 3.6788,
+      "step": 8068
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.20175426526198e-08,
+      "loss": 3.5119,
+      "step": 8069
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1791909711003657e-08,
+      "loss": 3.6591,
+      "step": 8070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1567072083952265e-08,
+      "loss": 3.7518,
+      "step": 8071
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1343029807456296e-08,
+      "loss": 3.7262,
+      "step": 8072
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.111978291738149e-08,
+      "loss": 3.8551,
+      "step": 8073
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.089733144946594e-08,
+      "loss": 3.6975,
+      "step": 8074
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.067567543931838e-08,
+      "loss": 3.7479,
+      "step": 8075
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.045481492242319e-08,
+      "loss": 3.7231,
+      "step": 8076
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.023474993413489e-08,
+      "loss": 3.6144,
+      "step": 8077
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.001548050968195e-08,
+      "loss": 3.7375,
+      "step": 8078
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9797006684165185e-08,
+      "loss": 3.52,
+      "step": 8079
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9579328492557734e-08,
+      "loss": 3.7343,
+      "step": 8080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9362445969706166e-08,
+      "loss": 3.767,
+      "step": 8081
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9146359150329373e-08,
+      "loss": 3.7523,
+      "step": 8082
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.893106806901802e-08,
+      "loss": 3.6158,
+      "step": 8083
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8716572760236205e-08,
+      "loss": 3.6884,
+      "step": 8084
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8502873258320908e-08,
+      "loss": 3.5539,
+      "step": 8085
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8289969597481424e-08,
+      "loss": 3.6685,
+      "step": 8086
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8077861811798835e-08,
+      "loss": 3.7008,
+      "step": 8087
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7866549935227638e-08,
+      "loss": 3.696,
+      "step": 8088
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7656034001595778e-08,
+      "loss": 3.631,
+      "step": 8089
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7446314044601298e-08,
+      "loss": 3.7503,
+      "step": 8090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7237390097817895e-08,
+      "loss": 3.6373,
+      "step": 8091
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.702926219468882e-08,
+      "loss": 3.6768,
+      "step": 8092
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6821930368532423e-08,
+      "loss": 3.7557,
+      "step": 8093
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.661539465253826e-08,
+      "loss": 3.5238,
+      "step": 8094
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6409655079768225e-08,
+      "loss": 3.723,
+      "step": 8095
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.620471168315819e-08,
+      "loss": 3.7358,
+      "step": 8096
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6000564495514134e-08,
+      "loss": 3.6627,
+      "step": 8097
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.579721354951825e-08,
+      "loss": 3.6777,
+      "step": 8098
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5594658877721168e-08,
+      "loss": 3.6834,
+      "step": 8099
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5392900512549168e-08,
+      "loss": 3.654,
+      "step": 8100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5191938486299196e-08,
+      "loss": 3.6812,
+      "step": 8101
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4991772831141626e-08,
+      "loss": 3.6806,
+      "step": 8102
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4792403579119163e-08,
+      "loss": 3.6364,
+      "step": 8103
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.459383076214794e-08,
+      "loss": 3.6261,
+      "step": 8104
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4396054412013647e-08,
+      "loss": 3.6246,
+      "step": 8105
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4199074560377623e-08,
+      "loss": 3.5387,
+      "step": 8106
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4002891238772975e-08,
+      "loss": 3.6617,
+      "step": 8107
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3807504478604583e-08,
+      "loss": 3.7161,
+      "step": 8108
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3612914311149093e-08,
+      "loss": 3.6564,
+      "step": 8109
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3419120767558256e-08,
+      "loss": 3.6428,
+      "step": 8110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.322612387885337e-08,
+      "loss": 3.6522,
+      "step": 8111
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.303392367593027e-08,
+      "loss": 3.8098,
+      "step": 8112
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.284252018955657e-08,
+      "loss": 3.6678,
+      "step": 8113
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2651913450371655e-08,
+      "loss": 3.7074,
+      "step": 8114
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2462103488888332e-08,
+      "loss": 3.7595,
+      "step": 8115
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2273090335491744e-08,
+      "loss": 3.693,
+      "step": 8116
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.208487402043935e-08,
+      "loss": 3.6578,
+      "step": 8117
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1897454573860388e-08,
+      "loss": 3.5992,
+      "step": 8118
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1710832025757522e-08,
+      "loss": 3.7712,
+      "step": 8119
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.152500640600519e-08,
+      "loss": 3.5895,
+      "step": 8120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1339977744350148e-08,
+      "loss": 3.6195,
+      "step": 8121
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1155746070413153e-08,
+      "loss": 3.6886,
+      "step": 8122
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0972311413685052e-08,
+      "loss": 3.5918,
+      "step": 8123
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0789673803530696e-08,
+      "loss": 3.701,
+      "step": 8124
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0607833269186695e-08,
+      "loss": 3.7418,
+      "step": 8125
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.042678983976254e-08,
+      "loss": 3.664,
+      "step": 8126
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.024654354423894e-08,
+      "loss": 3.6094,
+      "step": 8127
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.006709441147059e-08,
+      "loss": 3.6912,
+      "step": 8128
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9888442470183957e-08,
+      "loss": 3.6899,
+      "step": 8129
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.971058774897727e-08,
+      "loss": 3.7249,
+      "step": 8130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9533530276321656e-08,
+      "loss": 3.6884,
+      "step": 8131
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9357270080561654e-08,
+      "loss": 3.6838,
+      "step": 8132
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9181807189911362e-08,
+      "loss": 3.6264,
+      "step": 8133
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9007141632460534e-08,
+      "loss": 3.6828,
+      "step": 8134
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.883327343616903e-08,
+      "loss": 3.6167,
+      "step": 8135
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.866020262887014e-08,
+      "loss": 3.461,
+      "step": 8136
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.848792923826892e-08,
+      "loss": 3.6464,
+      "step": 8137
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8316453291943314e-08,
+      "loss": 3.571,
+      "step": 8138
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8145774817342477e-08,
+      "loss": 3.779,
+      "step": 8139
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7975893841790105e-08,
+      "loss": 3.8034,
+      "step": 8140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.780681039248e-08,
+      "loss": 3.7026,
+      "step": 8141
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7638524496478847e-08,
+      "loss": 3.6119,
+      "step": 8142
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7471036180726764e-08,
+      "loss": 3.6851,
+      "step": 8143
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7304345472035634e-08,
+      "loss": 3.5909,
+      "step": 8144
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7138452397088e-08,
+      "loss": 3.6129,
+      "step": 8145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.697335698244207e-08,
+      "loss": 3.6889,
+      "step": 8146
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6809059254524474e-08,
+      "loss": 3.7012,
+      "step": 8147
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6645559239638066e-08,
+      "loss": 3.6858,
+      "step": 8148
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.648285696395413e-08,
+      "loss": 3.7114,
+      "step": 8149
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.632095245351961e-08,
+      "loss": 3.6692,
+      "step": 8150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6159845734251557e-08,
+      "loss": 3.6982,
+      "step": 8151
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5999536831941e-08,
+      "loss": 3.6001,
+      "step": 8152
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5840025772248524e-08,
+      "loss": 3.7614,
+      "step": 8153
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.568131258071093e-08,
+      "loss": 3.7008,
+      "step": 8154
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5523397282733444e-08,
+      "loss": 3.7088,
+      "step": 8155
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.536627990359585e-08,
+      "loss": 3.6025,
+      "step": 8156
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5209960468450247e-08,
+      "loss": 3.717,
+      "step": 8157
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.50544390023194e-08,
+      "loss": 3.7337,
+      "step": 8158
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4899715530100057e-08,
+      "loss": 3.6939,
+      "step": 8159
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.474579007655963e-08,
+      "loss": 3.646,
+      "step": 8160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4592662666339518e-08,
+      "loss": 3.7257,
+      "step": 8161
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4440333323952337e-08,
+      "loss": 3.57,
+      "step": 8162
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.428880207378247e-08,
+      "loss": 3.7414,
+      "step": 8163
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.413806894008718e-08,
+      "loss": 3.8237,
+      "step": 8164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3988133946996608e-08,
+      "loss": 3.5671,
+      "step": 8165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3838997118512664e-08,
+      "loss": 3.6513,
+      "step": 8166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3690658478507924e-08,
+      "loss": 3.7012,
+      "step": 8167
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3543118050730053e-08,
+      "loss": 3.7048,
+      "step": 8168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3396375858796273e-08,
+      "loss": 3.6315,
+      "step": 8169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3250431926197793e-08,
+      "loss": 3.6959,
+      "step": 8170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3105286276297036e-08,
+      "loss": 3.5412,
+      "step": 8171
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2960938932329858e-08,
+      "loss": 3.7578,
+      "step": 8172
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2817389917402778e-08,
+      "loss": 3.6632,
+      "step": 8173
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2674639254495192e-08,
+      "loss": 3.6256,
+      "step": 8174
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2532686966459373e-08,
+      "loss": 3.6403,
+      "step": 8175
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2391533076018258e-08,
+      "loss": 3.6275,
+      "step": 8176
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2251177605768771e-08,
+      "loss": 3.5689,
+      "step": 8177
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2111620578178495e-08,
+      "loss": 3.7112,
+      "step": 8178
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1972862015587894e-08,
+      "loss": 3.7609,
+      "step": 8179
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1834901940209752e-08,
+      "loss": 3.6822,
+      "step": 8180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.169774037412863e-08,
+      "loss": 3.7187,
+      "step": 8181
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1561377339301405e-08,
+      "loss": 3.6089,
+      "step": 8182
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1425812857557838e-08,
+      "loss": 3.6979,
+      "step": 8183
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.129104695059835e-08,
+      "loss": 3.7063,
+      "step": 8184
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1157079639996792e-08,
+      "loss": 3.5817,
+      "step": 8185
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.102391094719879e-08,
+      "loss": 3.6202,
+      "step": 8186
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0891540893522289e-08,
+      "loss": 3.7101,
+      "step": 8187
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0759969500155897e-08,
+      "loss": 3.6779,
+      "step": 8188
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.062919678816332e-08,
+      "loss": 3.6532,
+      "step": 8189
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0499222778478369e-08,
+      "loss": 3.6805,
+      "step": 8190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0370047491906621e-08,
+      "loss": 3.7125,
+      "step": 8191
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0241670949127091e-08,
+      "loss": 3.618,
+      "step": 8192
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0114093170690564e-08,
+      "loss": 3.6996,
+      "step": 8193
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.987314177019036e-09,
+      "loss": 3.7566,
+      "step": 8194
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.861333988408495e-09,
+      "loss": 3.7374,
+      "step": 8195
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.73615262502503e-09,
+      "loss": 3.5921,
+      "step": 8196
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.611770106908169e-09,
+      "loss": 3.6611,
+      "step": 8197
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.488186453969206e-09,
+      "loss": 3.6395,
+      "step": 8198
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.365401685991204e-09,
+      "loss": 3.6797,
+      "step": 8199
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.243415822629553e-09,
+      "loss": 3.6633,
+      "step": 8200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.12222888341252e-09,
+      "loss": 3.6139,
+      "step": 8201
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.001840887739588e-09,
+      "loss": 3.7494,
+      "step": 8202
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.882251854882007e-09,
+      "loss": 3.5993,
+      "step": 8203
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.763461803983907e-09,
+      "loss": 3.7143,
+      "step": 8204
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.645470754062302e-09,
+      "loss": 3.7292,
+      "step": 8205
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.528278724003747e-09,
+      "loss": 3.803,
+      "step": 8206
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.411885732568792e-09,
+      "loss": 3.648,
+      "step": 8207
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.29629179839031e-09,
+      "loss": 3.6795,
+      "step": 8208
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.181496939972944e-09,
+      "loss": 3.7321,
+      "step": 8209
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.067501175691994e-09,
+      "loss": 3.5783,
+      "step": 8210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.954304523796753e-09,
+      "loss": 3.7744,
+      "step": 8211
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.841907002407723e-09,
+      "loss": 3.5525,
+      "step": 8212
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.73030862951718e-09,
+      "loss": 3.5861,
+      "step": 8213
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.61950942299028e-09,
+      "loss": 3.5959,
+      "step": 8214
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.509509400563942e-09,
+      "loss": 3.6621,
+      "step": 8215
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.40030857984686e-09,
+      "loss": 3.7215,
+      "step": 8216
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.291906978320606e-09,
+      "loss": 3.6225,
+      "step": 8217
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.184304613336857e-09,
+      "loss": 3.631,
+      "step": 8218
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.0775015021218304e-09,
+      "loss": 3.6856,
+      "step": 8219
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.971497661771854e-09,
+      "loss": 3.8455,
+      "step": 8220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.866293109256683e-09,
+      "loss": 3.6626,
+      "step": 8221
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.761887861417293e-09,
+      "loss": 3.6866,
+      "step": 8222
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.658281934966981e-09,
+      "loss": 3.6064,
+      "step": 8223
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.555475346491369e-09,
+      "loss": 3.5518,
+      "step": 8224
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.453468112447292e-09,
+      "loss": 3.6817,
+      "step": 8225
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.3522602491644655e-09,
+      "loss": 3.6324,
+      "step": 8226
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.251851772844375e-09,
+      "loss": 3.7427,
+      "step": 8227
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.152242699560273e-09,
+      "loss": 3.7078,
+      "step": 8228
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.053433045257739e-09,
+      "loss": 3.6992,
+      "step": 8229
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.9554228257541205e-09,
+      "loss": 3.7746,
+      "step": 8230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.858212056739643e-09,
+      "loss": 3.6857,
+      "step": 8231
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.761800753775193e-09,
+      "loss": 3.7558,
+      "step": 8232
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.666188932294536e-09,
+      "loss": 3.6572,
+      "step": 8233
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.571376607603207e-09,
+      "loss": 3.6605,
+      "step": 8234
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.477363794879065e-09,
+      "loss": 3.7111,
+      "step": 8235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.384150509171737e-09,
+      "loss": 3.793,
+      "step": 8236
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.291736765402622e-09,
+      "loss": 3.6444,
+      "step": 8237
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.200122578365441e-09,
+      "loss": 3.6674,
+      "step": 8238
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.10930796272624e-09,
+      "loss": 3.6738,
+      "step": 8239
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.019292933022279e-09,
+      "loss": 3.7838,
+      "step": 8240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9300775036637e-09,
+      "loss": 3.6953,
+      "step": 8241
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.841661688931299e-09,
+      "loss": 3.692,
+      "step": 8242
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.754045502979865e-09,
+      "loss": 3.6885,
+      "step": 8243
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6672289598337364e-09,
+      "loss": 3.651,
+      "step": 8244
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.581212073391794e-09,
+      "loss": 3.7517,
+      "step": 8245
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.49599485742358e-09,
+      "loss": 3.6519,
+      "step": 8246
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4115773255698494e-09,
+      "loss": 3.6621,
+      "step": 8247
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.327959491344791e-09,
+      "loss": 3.7524,
+      "step": 8248
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.24514136813381e-09,
+      "loss": 3.6997,
+      "step": 8249
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.163122969195188e-09,
+      "loss": 3.6766,
+      "step": 8250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.081904307657869e-09,
+      "loss": 3.7497,
+      "step": 8251
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.001485396523119e-09,
+      "loss": 3.6729,
+      "step": 8252
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.921866248665085e-09,
+      "loss": 3.5554,
+      "step": 8253
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.843046876828571e-09,
+      "loss": 3.774,
+      "step": 8254
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7650272936323725e-09,
+      "loss": 3.7269,
+      "step": 8255
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6878075115642785e-09,
+      "loss": 3.6267,
+      "step": 8256
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6113875429866217e-09,
+      "loss": 3.6463,
+      "step": 8257
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.535767400132395e-09,
+      "loss": 3.6844,
+      "step": 8258
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.46094709510747e-09,
+      "loss": 3.7456,
+      "step": 8259
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3869266398889323e-09,
+      "loss": 3.6818,
+      "step": 8260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3137060463256378e-09,
+      "loss": 3.7094,
+      "step": 8261
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2412853261387657e-09,
+      "loss": 3.5565,
+      "step": 8262
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.169664490922375e-09,
+      "loss": 3.6004,
+      "step": 8263
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.098843552140629e-09,
+      "loss": 3.6804,
+      "step": 8264
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.028822521131125e-09,
+      "loss": 3.7601,
+      "step": 8265
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9596014091026748e-09,
+      "loss": 3.664,
+      "step": 8266
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.891180227135859e-09,
+      "loss": 3.5391,
+      "step": 8267
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8235589861846935e-09,
+      "loss": 3.628,
+      "step": 8268
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.756737697072742e-09,
+      "loss": 3.5087,
+      "step": 8269
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6907163704975592e-09,
+      "loss": 3.7365,
+      "step": 8270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6254950170279126e-09,
+      "loss": 3.644,
+      "step": 8271
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5610736471043398e-09,
+      "loss": 3.6772,
+      "step": 8272
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4974522710391468e-09,
+      "loss": 3.7431,
+      "step": 8273
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4346308990175205e-09,
+      "loss": 3.7136,
+      "step": 8274
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3726095410958606e-09,
+      "loss": 3.7718,
+      "step": 8275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.311388207201781e-09,
+      "loss": 3.6409,
+      "step": 8276
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2509669071363316e-09,
+      "loss": 3.781,
+      "step": 8277
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1913456505723297e-09,
+      "loss": 3.6247,
+      "step": 8278
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1325244470526973e-09,
+      "loss": 3.6938,
+      "step": 8279
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0745033059943463e-09,
+      "loss": 3.6407,
+      "step": 8280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.017282236685403e-09,
+      "loss": 3.693,
+      "step": 8281
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9608612482852064e-09,
+      "loss": 3.6596,
+      "step": 8282
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9052403498265316e-09,
+      "loss": 3.6159,
+      "step": 8283
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.850419550212812e-09,
+      "loss": 3.7448,
+      "step": 8284
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7963988582198056e-09,
+      "loss": 3.5046,
+      "step": 8285
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7431782824955946e-09,
+      "loss": 3.6982,
+      "step": 8286
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6907578315589202e-09,
+      "loss": 3.6098,
+      "step": 8287
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6391375138019583e-09,
+      "loss": 3.5818,
+      "step": 8288
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5883173374875438e-09,
+      "loss": 3.685,
+      "step": 8289
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.538297310751946e-09,
+      "loss": 3.6253,
+      "step": 8290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4890774416015385e-09,
+      "loss": 3.607,
+      "step": 8291
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4406577379155739e-09,
+      "loss": 3.6119,
+      "step": 8292
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3930382074456295e-09,
+      "loss": 3.6405,
+      "step": 8293
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3462188578144964e-09,
+      "loss": 3.7311,
+      "step": 8294
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3001996965167352e-09,
+      "loss": 3.6262,
+      "step": 8295
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2549807309192308e-09,
+      "loss": 3.6284,
+      "step": 8296
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2105619682606374e-09,
+      "loss": 3.6125,
+      "step": 8297
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1669434156519332e-09,
+      "loss": 3.6345,
+      "step": 8298
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1241250800753112e-09,
+      "loss": 3.6509,
+      "step": 8299
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.082106968385288e-09,
+      "loss": 3.6506,
+      "step": 8300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0408890873081502e-09,
+      "loss": 3.5462,
+      "step": 8301
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0004714434419527e-09,
+      "loss": 3.7413,
+      "step": 8302
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.608540432565206e-10,
+      "loss": 3.663,
+      "step": 8303
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.220368930945578e-10,
+      "loss": 3.6521,
+      "step": 8304
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.840199991688725e-10,
+      "loss": 3.7249,
+      "step": 8305
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.468033675668174e-10,
+      "loss": 3.6814,
+      "step": 8306
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.103870042441841e-10,
+      "loss": 3.7762,
+      "step": 8307
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.74770915032419e-10,
+      "loss": 3.6902,
+      "step": 8308
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.399551056314069e-10,
+      "loss": 3.7299,
+      "step": 8309
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.059395816155779e-10,
+      "loss": 3.6384,
+      "step": 8310
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.727243484294654e-10,
+      "loss": 3.6139,
+      "step": 8311
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.403094113904828e-10,
+      "loss": 3.8301,
+      "step": 8312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.086947756872574e-10,
+      "loss": 3.6781,
+      "step": 8313
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.77880446381296e-10,
+      "loss": 3.6799,
+      "step": 8314
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.478664284053192e-10,
+      "loss": 3.6987,
+      "step": 8315
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.186527265638175e-10,
+      "loss": 3.5925,
+      "step": 8316
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.902393455330501e-10,
+      "loss": 3.6994,
+      "step": 8317
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6262628986160074e-10,
+      "loss": 3.7339,
+      "step": 8318
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3581356397037756e-10,
+      "loss": 3.7353,
+      "step": 8319
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.098011721503925e-10,
+      "loss": 3.7228,
+      "step": 8320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.845891185672024e-10,
+      "loss": 3.7723,
+      "step": 8321
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6017740725535763e-10,
+      "loss": 3.6006,
+      "step": 8322
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.365660421233985e-10,
+      "loss": 3.6418,
+      "step": 8323
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.137550269516343e-10,
+      "loss": 3.7536,
+      "step": 8324
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9174436539047834e-10,
+      "loss": 3.6405,
+      "step": 8325
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7053406096433366e-10,
+      "loss": 3.6957,
+      "step": 8326
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5012411706770714e-10,
+      "loss": 3.6044,
+      "step": 8327
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.305145369685402e-10,
+      "loss": 3.7754,
+      "step": 8328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.117053238059885e-10,
+      "loss": 3.7004,
+      "step": 8329
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9369648059097689e-10,
+      "loss": 3.651,
+      "step": 8330
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7648801020564432e-10,
+      "loss": 3.7107,
+      "step": 8331
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6007991540556434e-10,
+      "loss": 3.7321,
+      "step": 8332
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.444721988169695e-10,
+      "loss": 3.6408,
+      "step": 8333
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2966486293841673e-10,
+      "loss": 3.6608,
+      "step": 8334
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1565791014078732e-10,
+      "loss": 3.74,
+      "step": 8335
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0245134266562152e-10,
+      "loss": 3.7125,
+      "step": 8336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.004516262678397e-11,
+      "loss": 3.7119,
+      "step": 8337
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.843937201157393e-11,
+      "loss": 3.6551,
+      "step": 8338
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.763397267628424e-11,
+      "loss": 3.7448,
+      "step": 8339
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.762896635175263e-11,
+      "loss": 3.6737,
+      "step": 8340
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.842435463947581e-11,
+      "loss": 3.7594,
+      "step": 8341
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.002013901271973e-11,
+      "loss": 3.6485,
+      "step": 8342
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.241632081651958e-11,
+      "loss": 3.6767,
+      "step": 8343
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.56129012682349e-11,
+      "loss": 3.7185,
+      "step": 8344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9609881457549608e-11,
+      "loss": 3.6751,
+      "step": 8345
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4407262344806605e-11,
+      "loss": 3.5813,
+      "step": 8346
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0005044763228277e-11,
+      "loss": 3.6931,
+      "step": 8347
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.403229417251134e-12,
+      "loss": 3.6698,
+      "step": 8348
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6018168836360333e-12,
+      "loss": 3.6223,
+      "step": 8349
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6008076103579683e-12,
+      "loss": 3.7611,
+      "step": 8350
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0020191882650385e-13,
+      "loss": 3.7207,
+      "step": 8351
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 3.8362,
+      "step": 8352
+    },
+    {
+      "epoch": 1.0,
+      "step": 8352,
+      "total_flos": 1.0768015748329636e+18,
+      "train_loss": 3.708707153426048,
+      "train_runtime": 9172.4719,
+      "train_samples_per_second": 58.27,
+      "train_steps_per_second": 0.911
+    }
+  ],
+  "max_steps": 8352,
+  "num_train_epochs": 1,
+  "total_flos": 1.0768015748329636e+18,
+  "trial_name": null,
+  "trial_params": null
+}