diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,57025 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 9500,
+  "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": 4.0849,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-08,
+      "loss": 4.2547,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 3.7748,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-08,
+      "loss": 3.7667,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 3.7999,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 3.7966,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4e-07,
+      "loss": 3.6864,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6e-07,
+      "loss": 3.753,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-07,
+      "loss": 3.8085,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 3.713,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2e-07,
+      "loss": 3.7015,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 3.6297,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6e-07,
+      "loss": 3.8152,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8e-07,
+      "loss": 3.7103,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 3.7804,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2e-07,
+      "loss": 3.7789,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 3.6545,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-07,
+      "loss": 3.6629,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8e-07,
+      "loss": 3.6399,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 3.7148,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 3.8268,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4e-07,
+      "loss": 3.7382,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6000000000000004e-07,
+      "loss": 3.6644,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 3.6612,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 3.5825,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.2e-07,
+      "loss": 3.6453,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.4e-07,
+      "loss": 3.6813,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.6e-07,
+      "loss": 3.6775,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 3.5796,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 3.5652,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.200000000000001e-07,
+      "loss": 3.4863,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.4e-07,
+      "loss": 3.5347,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.6e-07,
+      "loss": 3.5375,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 3.457,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 3.5001,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.2e-07,
+      "loss": 3.3503,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.4e-07,
+      "loss": 3.3336,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.6e-07,
+      "loss": 3.4189,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.8e-07,
+      "loss": 3.3633,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 3.4436,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.200000000000001e-07,
+      "loss": 3.1526,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.400000000000001e-07,
+      "loss": 3.0805,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.6e-07,
+      "loss": 3.1453,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.8e-07,
+      "loss": 3.0905,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 3.166,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 3.0587,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.400000000000001e-07,
+      "loss": 3.0814,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 3.0385,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 3.1109,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 2.9724,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.02e-06,
+      "loss": 2.9547,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.04e-06,
+      "loss": 2.9945,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.06e-06,
+      "loss": 2.8997,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.08e-06,
+      "loss": 2.8656,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1e-06,
+      "loss": 2.8961,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.12e-06,
+      "loss": 2.8771,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.14e-06,
+      "loss": 2.8378,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1600000000000001e-06,
+      "loss": 2.9756,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 2.9886,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 2.918,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2200000000000002e-06,
+      "loss": 2.8971,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2400000000000002e-06,
+      "loss": 2.8768,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.26e-06,
+      "loss": 2.9425,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.28e-06,
+      "loss": 2.8266,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3e-06,
+      "loss": 2.8263,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.32e-06,
+      "loss": 2.8839,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.34e-06,
+      "loss": 2.7258,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 2.8609,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 2.8125,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 2.7838,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.42e-06,
+      "loss": 2.7888,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.44e-06,
+      "loss": 2.8565,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.46e-06,
+      "loss": 2.7278,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.48e-06,
+      "loss": 2.8814,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5e-06,
+      "loss": 2.807,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.52e-06,
+      "loss": 2.8277,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.54e-06,
+      "loss": 2.8267,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.56e-06,
+      "loss": 2.803,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 2.8073,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 2.7907,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6200000000000002e-06,
+      "loss": 2.665,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6400000000000002e-06,
+      "loss": 2.8236,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6600000000000002e-06,
+      "loss": 2.802,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 2.8322,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 2.8089,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.72e-06,
+      "loss": 2.8163,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.74e-06,
+      "loss": 2.6929,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.76e-06,
+      "loss": 2.8663,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 2.8027,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 2.7831,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8200000000000002e-06,
+      "loss": 2.792,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8400000000000002e-06,
+      "loss": 2.8184,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8600000000000002e-06,
+      "loss": 2.8087,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 2.8189,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 2.7829,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 2.8367,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.94e-06,
+      "loss": 2.8516,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9600000000000003e-06,
+      "loss": 2.6574,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.98e-06,
+      "loss": 2.726,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 2.7882,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.02e-06,
+      "loss": 2.7525,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.04e-06,
+      "loss": 2.8489,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.06e-06,
+      "loss": 2.8099,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.08e-06,
+      "loss": 2.7907,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 2.7198,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.12e-06,
+      "loss": 2.7254,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1400000000000003e-06,
+      "loss": 2.6893,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.16e-06,
+      "loss": 2.8345,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 2.6555,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2e-06,
+      "loss": 2.5991,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2200000000000003e-06,
+      "loss": 2.7657,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.24e-06,
+      "loss": 2.694,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2600000000000004e-06,
+      "loss": 2.6142,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.28e-06,
+      "loss": 2.7505,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 2.7498,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3200000000000002e-06,
+      "loss": 2.7914,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3400000000000005e-06,
+      "loss": 2.7842,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3600000000000003e-06,
+      "loss": 2.7574,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.38e-06,
+      "loss": 2.7516,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 2.7145,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.42e-06,
+      "loss": 2.83,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4400000000000004e-06,
+      "loss": 2.7592,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.46e-06,
+      "loss": 2.7365,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 2.7506,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-06,
+      "loss": 2.7578,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.52e-06,
+      "loss": 2.7529,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5400000000000002e-06,
+      "loss": 2.7406,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.56e-06,
+      "loss": 2.6916,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 2.7576,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6e-06,
+      "loss": 2.8568,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6200000000000003e-06,
+      "loss": 2.7901,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.64e-06,
+      "loss": 2.6852,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6600000000000004e-06,
+      "loss": 2.7169,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.68e-06,
+      "loss": 2.7343,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 2.7451,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 2.6833,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7400000000000004e-06,
+      "loss": 2.8169,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7600000000000003e-06,
+      "loss": 2.6993,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 2.6669,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 2.7607,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.82e-06,
+      "loss": 2.7371,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.84e-06,
+      "loss": 2.6837,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.86e-06,
+      "loss": 2.7422,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.88e-06,
+      "loss": 2.7978,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9e-06,
+      "loss": 2.7933,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.92e-06,
+      "loss": 2.7294,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 2.744,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.96e-06,
+      "loss": 2.7404,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 2.665,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3e-06,
+      "loss": 2.8174,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0200000000000003e-06,
+      "loss": 2.7313,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.04e-06,
+      "loss": 2.6594,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0600000000000003e-06,
+      "loss": 2.6314,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.08e-06,
+      "loss": 2.6845,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 2.7209,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.12e-06,
+      "loss": 2.7958,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1400000000000004e-06,
+      "loss": 2.6734,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1600000000000002e-06,
+      "loss": 2.7409,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 2.6351,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 2.5851,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2200000000000005e-06,
+      "loss": 2.7724,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2400000000000003e-06,
+      "loss": 2.6706,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2600000000000006e-06,
+      "loss": 2.7609,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 2.6931,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 2.7064,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 2.6798,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3400000000000006e-06,
+      "loss": 2.7682,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 2.6395,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 2.6804,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 2.6271,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4200000000000007e-06,
+      "loss": 2.6078,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.44e-06,
+      "loss": 2.6168,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.46e-06,
+      "loss": 2.7,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.48e-06,
+      "loss": 2.7068,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5e-06,
+      "loss": 2.7934,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.52e-06,
+      "loss": 2.7884,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.54e-06,
+      "loss": 2.6925,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5600000000000002e-06,
+      "loss": 2.661,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.58e-06,
+      "loss": 2.6431,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 2.6723,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.62e-06,
+      "loss": 2.711,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6400000000000003e-06,
+      "loss": 2.6483,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.66e-06,
+      "loss": 2.709,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 2.7503,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7e-06,
+      "loss": 2.667,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7200000000000004e-06,
+      "loss": 2.7155,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.74e-06,
+      "loss": 2.6617,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 2.5795,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 2.8093,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 2.6018,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.820000000000001e-06,
+      "loss": 2.7081,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 2.6231,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.86e-06,
+      "loss": 2.655,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.88e-06,
+      "loss": 2.728,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 2.7022,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 2.7299,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.94e-06,
+      "loss": 2.7058,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.96e-06,
+      "loss": 2.7342,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 2.6486,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 2.641,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 2.6576,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.04e-06,
+      "loss": 2.6745,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.060000000000001e-06,
+      "loss": 2.66,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.08e-06,
+      "loss": 2.6709,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1e-06,
+      "loss": 2.6594,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.12e-06,
+      "loss": 2.7557,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.14e-06,
+      "loss": 2.779,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.16e-06,
+      "loss": 2.6991,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.18e-06,
+      "loss": 2.6531,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 2.7611,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.22e-06,
+      "loss": 2.716,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.24e-06,
+      "loss": 2.6051,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.26e-06,
+      "loss": 2.6698,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2800000000000005e-06,
+      "loss": 2.6144,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3e-06,
+      "loss": 2.6621,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.32e-06,
+      "loss": 2.5754,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.34e-06,
+      "loss": 2.6295,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.360000000000001e-06,
+      "loss": 2.6479,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.38e-06,
+      "loss": 2.647,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4e-06,
+      "loss": 2.6494,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.42e-06,
+      "loss": 2.6756,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.440000000000001e-06,
+      "loss": 2.7264,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4600000000000005e-06,
+      "loss": 2.7573,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.48e-06,
+      "loss": 2.6999,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5e-06,
+      "loss": 2.7136,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.520000000000001e-06,
+      "loss": 2.704,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.540000000000001e-06,
+      "loss": 2.6073,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.56e-06,
+      "loss": 2.6001,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.58e-06,
+      "loss": 2.5963,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 2.6613,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.620000000000001e-06,
+      "loss": 2.8112,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 2.6685,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.66e-06,
+      "loss": 2.759,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.680000000000001e-06,
+      "loss": 2.6964,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7e-06,
+      "loss": 2.6105,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7200000000000005e-06,
+      "loss": 2.6801,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.74e-06,
+      "loss": 2.7282,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.76e-06,
+      "loss": 2.7238,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.78e-06,
+      "loss": 2.5972,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 2.7104,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8200000000000004e-06,
+      "loss": 2.6989,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.84e-06,
+      "loss": 2.5819,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.86e-06,
+      "loss": 2.5793,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 2.6313,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 2.7209,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.92e-06,
+      "loss": 2.5783,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.94e-06,
+      "loss": 2.574,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 2.7655,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 2.6889,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5e-06,
+      "loss": 2.5236,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.02e-06,
+      "loss": 2.6785,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.04e-06,
+      "loss": 2.6946,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.060000000000001e-06,
+      "loss": 2.6608,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.0800000000000005e-06,
+      "loss": 2.6154,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.1e-06,
+      "loss": 2.5936,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.12e-06,
+      "loss": 2.7424,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.140000000000001e-06,
+      "loss": 2.6006,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.1600000000000006e-06,
+      "loss": 2.6867,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.18e-06,
+      "loss": 2.6521,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.2e-06,
+      "loss": 2.5817,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.220000000000001e-06,
+      "loss": 2.6908,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.240000000000001e-06,
+      "loss": 2.718,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.2600000000000005e-06,
+      "loss": 2.566,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.28e-06,
+      "loss": 2.6142,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.300000000000001e-06,
+      "loss": 2.6191,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.320000000000001e-06,
+      "loss": 2.5594,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.3400000000000005e-06,
+      "loss": 2.6408,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.36e-06,
+      "loss": 2.5974,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 2.6953,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 2.6986,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.420000000000001e-06,
+      "loss": 2.6024,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 2.6452,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.460000000000001e-06,
+      "loss": 2.5447,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.480000000000001e-06,
+      "loss": 2.662,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 2.6479,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.5200000000000005e-06,
+      "loss": 2.631,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.540000000000001e-06,
+      "loss": 2.612,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.560000000000001e-06,
+      "loss": 2.632,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 2.5467,
+      "step": 279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 2.691,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.620000000000001e-06,
+      "loss": 2.638,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.64e-06,
+      "loss": 2.7294,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.66e-06,
+      "loss": 2.6613,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.68e-06,
+      "loss": 2.6462,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.7e-06,
+      "loss": 2.6053,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.72e-06,
+      "loss": 2.6851,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.74e-06,
+      "loss": 2.7143,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.76e-06,
+      "loss": 2.7471,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.78e-06,
+      "loss": 2.6588,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.8e-06,
+      "loss": 2.5837,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.82e-06,
+      "loss": 2.6616,
+      "step": 291
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.84e-06,
+      "loss": 2.5706,
+      "step": 292
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.86e-06,
+      "loss": 2.6493,
+      "step": 293
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 2.6529,
+      "step": 294
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9e-06,
+      "loss": 2.7468,
+      "step": 295
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.92e-06,
+      "loss": 2.6457,
+      "step": 296
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.94e-06,
+      "loss": 2.573,
+      "step": 297
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9600000000000005e-06,
+      "loss": 2.5746,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.98e-06,
+      "loss": 2.5662,
+      "step": 299
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6e-06,
+      "loss": 2.5504,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.02e-06,
+      "loss": 2.6326,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.040000000000001e-06,
+      "loss": 2.6348,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 2.5902,
+      "step": 303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.08e-06,
+      "loss": 2.6015,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.1e-06,
+      "loss": 2.6744,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.120000000000001e-06,
+      "loss": 2.6646,
+      "step": 306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.1400000000000005e-06,
+      "loss": 2.6614,
+      "step": 307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.16e-06,
+      "loss": 2.6679,
+      "step": 308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.18e-06,
+      "loss": 2.5792,
+      "step": 309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 2.7111,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.220000000000001e-06,
+      "loss": 2.6208,
+      "step": 311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.24e-06,
+      "loss": 2.6435,
+      "step": 312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.26e-06,
+      "loss": 2.5941,
+      "step": 313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.280000000000001e-06,
+      "loss": 2.6303,
+      "step": 314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 2.6001,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.3200000000000005e-06,
+      "loss": 2.5298,
+      "step": 316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.34e-06,
+      "loss": 2.5679,
+      "step": 317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.360000000000001e-06,
+      "loss": 2.5824,
+      "step": 318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 2.6069,
+      "step": 319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 2.6258,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.42e-06,
+      "loss": 2.5706,
+      "step": 321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.440000000000001e-06,
+      "loss": 2.5985,
+      "step": 322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.460000000000001e-06,
+      "loss": 2.6794,
+      "step": 323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.480000000000001e-06,
+      "loss": 2.5521,
+      "step": 324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 2.7147,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.520000000000001e-06,
+      "loss": 2.5943,
+      "step": 326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.540000000000001e-06,
+      "loss": 2.5422,
+      "step": 327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 2.6089,
+      "step": 328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 2.6447,
+      "step": 329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 2.6386,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.620000000000001e-06,
+      "loss": 2.6691,
+      "step": 331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.640000000000001e-06,
+      "loss": 2.649,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 2.6574,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.680000000000001e-06,
+      "loss": 2.5826,
+      "step": 334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 2.5621,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 2.568,
+      "step": 336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.740000000000001e-06,
+      "loss": 2.6153,
+      "step": 337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.760000000000001e-06,
+      "loss": 2.6382,
+      "step": 338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 2.561,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 2.6414,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.820000000000001e-06,
+      "loss": 2.6471,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.8400000000000014e-06,
+      "loss": 2.6298,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.860000000000001e-06,
+      "loss": 2.6433,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.88e-06,
+      "loss": 2.6215,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.9e-06,
+      "loss": 2.6361,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.92e-06,
+      "loss": 2.6305,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.9400000000000005e-06,
+      "loss": 2.602,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.96e-06,
+      "loss": 2.5801,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.98e-06,
+      "loss": 2.6136,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7e-06,
+      "loss": 2.6049,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 2.6,
+      "step": 351
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.04e-06,
+      "loss": 2.5275,
+      "step": 352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.06e-06,
+      "loss": 2.5743,
+      "step": 353
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.08e-06,
+      "loss": 2.6399,
+      "step": 354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.100000000000001e-06,
+      "loss": 2.6845,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.1200000000000004e-06,
+      "loss": 2.6145,
+      "step": 356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.14e-06,
+      "loss": 2.669,
+      "step": 357
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.16e-06,
+      "loss": 2.5939,
+      "step": 358
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 2.6901,
+      "step": 359
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 2.6088,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.22e-06,
+      "loss": 2.6288,
+      "step": 361
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.24e-06,
+      "loss": 2.639,
+      "step": 362
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.260000000000001e-06,
+      "loss": 2.5805,
+      "step": 363
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.280000000000001e-06,
+      "loss": 2.6444,
+      "step": 364
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.3e-06,
+      "loss": 2.6465,
+      "step": 365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.32e-06,
+      "loss": 2.6,
+      "step": 366
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.340000000000001e-06,
+      "loss": 2.7083,
+      "step": 367
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.360000000000001e-06,
+      "loss": 2.6031,
+      "step": 368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 2.5398,
+      "step": 369
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.4e-06,
+      "loss": 2.6403,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.420000000000001e-06,
+      "loss": 2.5157,
+      "step": 371
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.440000000000001e-06,
+      "loss": 2.6461,
+      "step": 372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.4600000000000006e-06,
+      "loss": 2.6025,
+      "step": 373
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.48e-06,
+      "loss": 2.6397,
+      "step": 374
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 2.601,
+      "step": 375
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.520000000000001e-06,
+      "loss": 2.6305,
+      "step": 376
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.540000000000001e-06,
+      "loss": 2.4958,
+      "step": 377
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 2.582,
+      "step": 378
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.58e-06,
+      "loss": 2.6414,
+      "step": 379
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 2.606,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.620000000000001e-06,
+      "loss": 2.5626,
+      "step": 381
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.640000000000001e-06,
+      "loss": 2.6869,
+      "step": 382
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.660000000000001e-06,
+      "loss": 2.6258,
+      "step": 383
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 2.6287,
+      "step": 384
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.7e-06,
+      "loss": 2.5859,
+      "step": 385
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.72e-06,
+      "loss": 2.641,
+      "step": 386
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.74e-06,
+      "loss": 2.6596,
+      "step": 387
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.76e-06,
+      "loss": 2.5975,
+      "step": 388
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.78e-06,
+      "loss": 2.6006,
+      "step": 389
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 2.6555,
+      "step": 390
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.820000000000001e-06,
+      "loss": 2.5262,
+      "step": 391
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.840000000000001e-06,
+      "loss": 2.6299,
+      "step": 392
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 2.5769,
+      "step": 393
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.88e-06,
+      "loss": 2.6058,
+      "step": 394
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.9e-06,
+      "loss": 2.6013,
+      "step": 395
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.92e-06,
+      "loss": 2.61,
+      "step": 396
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.94e-06,
+      "loss": 2.5893,
+      "step": 397
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.960000000000002e-06,
+      "loss": 2.5121,
+      "step": 398
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 2.5328,
+      "step": 399
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 2.6155,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.020000000000001e-06,
+      "loss": 2.6553,
+      "step": 401
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 2.5966,
+      "step": 402
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.06e-06,
+      "loss": 2.4594,
+      "step": 403
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.08e-06,
+      "loss": 2.6067,
+      "step": 404
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.1e-06,
+      "loss": 2.7325,
+      "step": 405
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.120000000000002e-06,
+      "loss": 2.6326,
+      "step": 406
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.14e-06,
+      "loss": 2.6209,
+      "step": 407
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.16e-06,
+      "loss": 2.5934,
+      "step": 408
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.18e-06,
+      "loss": 2.5808,
+      "step": 409
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.2e-06,
+      "loss": 2.5991,
+      "step": 410
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 2.6196,
+      "step": 411
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.24e-06,
+      "loss": 2.5907,
+      "step": 412
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.26e-06,
+      "loss": 2.5967,
+      "step": 413
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.28e-06,
+      "loss": 2.5882,
+      "step": 414
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.3e-06,
+      "loss": 2.5799,
+      "step": 415
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.32e-06,
+      "loss": 2.6261,
+      "step": 416
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.34e-06,
+      "loss": 2.6446,
+      "step": 417
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.36e-06,
+      "loss": 2.6083,
+      "step": 418
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 2.6294,
+      "step": 419
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 2.6241,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.42e-06,
+      "loss": 2.5419,
+      "step": 421
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.44e-06,
+      "loss": 2.471,
+      "step": 422
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.46e-06,
+      "loss": 2.5545,
+      "step": 423
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.48e-06,
+      "loss": 2.516,
+      "step": 424
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.5e-06,
+      "loss": 2.5959,
+      "step": 425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.52e-06,
+      "loss": 2.5981,
+      "step": 426
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.540000000000001e-06,
+      "loss": 2.5861,
+      "step": 427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.560000000000001e-06,
+      "loss": 2.5671,
+      "step": 428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 2.5829,
+      "step": 429
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.6e-06,
+      "loss": 2.61,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.62e-06,
+      "loss": 2.6632,
+      "step": 431
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.64e-06,
+      "loss": 2.6052,
+      "step": 432
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.66e-06,
+      "loss": 2.5404,
+      "step": 433
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.68e-06,
+      "loss": 2.4941,
+      "step": 434
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.700000000000001e-06,
+      "loss": 2.5619,
+      "step": 435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.720000000000001e-06,
+      "loss": 2.5723,
+      "step": 436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.740000000000001e-06,
+      "loss": 2.6061,
+      "step": 437
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.76e-06,
+      "loss": 2.4716,
+      "step": 438
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.78e-06,
+      "loss": 2.6333,
+      "step": 439
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.8e-06,
+      "loss": 2.5716,
+      "step": 440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.82e-06,
+      "loss": 2.5498,
+      "step": 441
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.84e-06,
+      "loss": 2.5593,
+      "step": 442
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.860000000000002e-06,
+      "loss": 2.5837,
+      "step": 443
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.880000000000001e-06,
+      "loss": 2.5819,
+      "step": 444
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 2.5371,
+      "step": 445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.920000000000001e-06,
+      "loss": 2.5124,
+      "step": 446
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.94e-06,
+      "loss": 2.5585,
+      "step": 447
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.96e-06,
+      "loss": 2.6067,
+      "step": 448
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.98e-06,
+      "loss": 2.5635,
+      "step": 449
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9e-06,
+      "loss": 2.4565,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.020000000000002e-06,
+      "loss": 2.6058,
+      "step": 451
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.040000000000002e-06,
+      "loss": 2.5834,
+      "step": 452
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.060000000000001e-06,
+      "loss": 2.6376,
+      "step": 453
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.080000000000001e-06,
+      "loss": 2.6129,
+      "step": 454
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 2.5389,
+      "step": 455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.12e-06,
+      "loss": 2.5275,
+      "step": 456
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.14e-06,
+      "loss": 2.5832,
+      "step": 457
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.16e-06,
+      "loss": 2.6149,
+      "step": 458
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 2.5558,
+      "step": 459
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 2.5347,
+      "step": 460
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.220000000000002e-06,
+      "loss": 2.5904,
+      "step": 461
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.240000000000001e-06,
+      "loss": 2.5096,
+      "step": 462
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.260000000000001e-06,
+      "loss": 2.6194,
+      "step": 463
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 2.5829,
+      "step": 464
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.3e-06,
+      "loss": 2.6084,
+      "step": 465
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.32e-06,
+      "loss": 2.6,
+      "step": 466
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.340000000000002e-06,
+      "loss": 2.6598,
+      "step": 467
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.360000000000002e-06,
+      "loss": 2.5081,
+      "step": 468
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.38e-06,
+      "loss": 2.5643,
+      "step": 469
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.4e-06,
+      "loss": 2.5061,
+      "step": 470
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.42e-06,
+      "loss": 2.553,
+      "step": 471
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.440000000000001e-06,
+      "loss": 2.5653,
+      "step": 472
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.460000000000001e-06,
+      "loss": 2.5114,
+      "step": 473
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.48e-06,
+      "loss": 2.5247,
+      "step": 474
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.5e-06,
+      "loss": 2.6038,
+      "step": 475
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.52e-06,
+      "loss": 2.5393,
+      "step": 476
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.54e-06,
+      "loss": 2.6075,
+      "step": 477
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.56e-06,
+      "loss": 2.6136,
+      "step": 478
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.58e-06,
+      "loss": 2.599,
+      "step": 479
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 2.6079,
+      "step": 480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.620000000000001e-06,
+      "loss": 2.5842,
+      "step": 481
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.640000000000001e-06,
+      "loss": 2.5095,
+      "step": 482
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.66e-06,
+      "loss": 2.638,
+      "step": 483
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.68e-06,
+      "loss": 2.6381,
+      "step": 484
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.7e-06,
+      "loss": 2.5503,
+      "step": 485
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.72e-06,
+      "loss": 2.6259,
+      "step": 486
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.74e-06,
+      "loss": 2.5611,
+      "step": 487
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 2.6087,
+      "step": 488
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 2.5491,
+      "step": 489
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 2.6071,
+      "step": 490
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.820000000000001e-06,
+      "loss": 2.5927,
+      "step": 491
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.84e-06,
+      "loss": 2.5817,
+      "step": 492
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.86e-06,
+      "loss": 2.6075,
+      "step": 493
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.88e-06,
+      "loss": 2.5269,
+      "step": 494
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9e-06,
+      "loss": 2.5841,
+      "step": 495
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 2.5279,
+      "step": 496
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.940000000000001e-06,
+      "loss": 2.5828,
+      "step": 497
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 2.6522,
+      "step": 498
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 2.6778,
+      "step": 499
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 2.6327,
+      "step": 500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999999695382584e-06,
+      "loss": 2.5393,
+      "step": 501
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999998781530372e-06,
+      "loss": 2.4807,
+      "step": 502
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999997258443473e-06,
+      "loss": 2.5572,
+      "step": 503
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999995126122076e-06,
+      "loss": 2.5176,
+      "step": 504
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99999238456644e-06,
+      "loss": 2.5182,
+      "step": 505
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999989033776898e-06,
+      "loss": 2.5652,
+      "step": 506
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999985073753857e-06,
+      "loss": 2.5868,
+      "step": 507
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999980504497803e-06,
+      "loss": 2.5825,
+      "step": 508
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999975326009292e-06,
+      "loss": 2.6155,
+      "step": 509
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999969538288953e-06,
+      "loss": 2.5543,
+      "step": 510
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999963141337493e-06,
+      "loss": 2.5513,
+      "step": 511
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999956135155688e-06,
+      "loss": 2.5579,
+      "step": 512
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999948519744397e-06,
+      "loss": 2.5298,
+      "step": 513
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999940295104546e-06,
+      "loss": 2.6479,
+      "step": 514
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999931461237135e-06,
+      "loss": 2.6259,
+      "step": 515
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999922018143242e-06,
+      "loss": 2.6709,
+      "step": 516
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999911965824018e-06,
+      "loss": 2.6108,
+      "step": 517
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999901304280686e-06,
+      "loss": 2.5874,
+      "step": 518
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999890033514547e-06,
+      "loss": 2.6436,
+      "step": 519
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999878153526974e-06,
+      "loss": 2.5646,
+      "step": 520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999865664319414e-06,
+      "loss": 2.5191,
+      "step": 521
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99985256589339e-06,
+      "loss": 2.5472,
+      "step": 522
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999838858250497e-06,
+      "loss": 2.6125,
+      "step": 523
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999824541392404e-06,
+      "loss": 2.4955,
+      "step": 524
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999809615320857e-06,
+      "loss": 2.4718,
+      "step": 525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999794080037675e-06,
+      "loss": 2.5558,
+      "step": 526
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99977793554475e-06,
+      "loss": 2.5367,
+      "step": 527
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99976118184405e-06,
+      "loss": 2.5469,
+      "step": 528
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999743818937614e-06,
+      "loss": 2.5502,
+      "step": 529
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999725846827562e-06,
+      "loss": 2.5235,
+      "step": 530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99970726551608e-06,
+      "loss": 2.542,
+      "step": 531
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999688075005434e-06,
+      "loss": 2.5022,
+      "step": 532
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99966827529796e-06,
+      "loss": 2.4506,
+      "step": 533
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999647866396073e-06,
+      "loss": 2.51,
+      "step": 534
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999626848302261e-06,
+      "loss": 2.5538,
+      "step": 535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999605221019082e-06,
+      "loss": 2.6043,
+      "step": 536
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999582984549172e-06,
+      "loss": 2.5454,
+      "step": 537
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999560138895238e-06,
+      "loss": 2.5675,
+      "step": 538
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999536684060071e-06,
+      "loss": 2.5403,
+      "step": 539
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999512620046523e-06,
+      "loss": 2.6247,
+      "step": 540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999487946857526e-06,
+      "loss": 2.5649,
+      "step": 541
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999462664496088e-06,
+      "loss": 2.5092,
+      "step": 542
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99943677296529e-06,
+      "loss": 2.5516,
+      "step": 543
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999410272268285e-06,
+      "loss": 2.5984,
+      "step": 544
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999383162408303e-06,
+      "loss": 2.5509,
+      "step": 545
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999355443388649e-06,
+      "loss": 2.6436,
+      "step": 546
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999327115212698e-06,
+      "loss": 2.5386,
+      "step": 547
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999298177883902e-06,
+      "loss": 2.4774,
+      "step": 548
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99926863140579e-06,
+      "loss": 2.6278,
+      "step": 549
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999238475781957e-06,
+      "loss": 2.5792,
+      "step": 550
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999207711016081e-06,
+      "loss": 2.6663,
+      "step": 551
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999176337111908e-06,
+      "loss": 2.5381,
+      "step": 552
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999144354073264e-06,
+      "loss": 2.5137,
+      "step": 553
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999111761904046e-06,
+      "loss": 2.569,
+      "step": 554
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999078560608221e-06,
+      "loss": 2.5399,
+      "step": 555
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99904475018984e-06,
+      "loss": 2.5287,
+      "step": 556
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999010330653019e-06,
+      "loss": 2.5696,
+      "step": 557
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99897530200195e-06,
+      "loss": 2.5851,
+      "step": 558
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998939664240908e-06,
+      "loss": 2.5465,
+      "step": 559
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998903417374228e-06,
+      "loss": 2.5784,
+      "step": 560
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998866561406331e-06,
+      "loss": 2.5482,
+      "step": 561
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998829096341706e-06,
+      "loss": 2.6141,
+      "step": 562
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998791022184921e-06,
+      "loss": 2.6708,
+      "step": 563
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998752338940612e-06,
+      "loss": 2.5338,
+      "step": 564
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998713046613492e-06,
+      "loss": 2.5508,
+      "step": 565
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998673145208351e-06,
+      "loss": 2.537,
+      "step": 566
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99863263473005e-06,
+      "loss": 2.543,
+      "step": 567
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998591515183524e-06,
+      "loss": 2.5555,
+      "step": 568
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998549786573785e-06,
+      "loss": 2.5366,
+      "step": 569
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998507448905917e-06,
+      "loss": 2.543,
+      "step": 570
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998464502185076e-06,
+      "loss": 2.6095,
+      "step": 571
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9984209464165e-06,
+      "loss": 2.5522,
+      "step": 572
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998376781605493e-06,
+      "loss": 2.5104,
+      "step": 573
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998332007757436e-06,
+      "loss": 2.4816,
+      "step": 574
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998286624877786e-06,
+      "loss": 2.604,
+      "step": 575
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998240632972073e-06,
+      "loss": 2.5333,
+      "step": 576
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9981940320459e-06,
+      "loss": 2.649,
+      "step": 577
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998146822104943e-06,
+      "loss": 2.5308,
+      "step": 578
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99809900315496e-06,
+      "loss": 2.5439,
+      "step": 579
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998050575201772e-06,
+      "loss": 2.5835,
+      "step": 580
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.998001538251283e-06,
+      "loss": 2.5402,
+      "step": 581
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997951892309468e-06,
+      "loss": 2.4621,
+      "step": 582
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997901637382375e-06,
+      "loss": 2.5277,
+      "step": 583
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997850773476126e-06,
+      "loss": 2.5016,
+      "step": 584
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997799300596921e-06,
+      "loss": 2.5245,
+      "step": 585
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997747218751032e-06,
+      "loss": 2.5703,
+      "step": 586
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997694527944804e-06,
+      "loss": 2.5799,
+      "step": 587
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997641228184656e-06,
+      "loss": 2.5474,
+      "step": 588
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997587319477084e-06,
+      "loss": 2.6293,
+      "step": 589
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997532801828659e-06,
+      "loss": 2.5571,
+      "step": 590
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997477675246018e-06,
+      "loss": 2.5766,
+      "step": 591
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997421939735885e-06,
+      "loss": 2.5065,
+      "step": 592
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997365595305045e-06,
+      "loss": 2.604,
+      "step": 593
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997308641960365e-06,
+      "loss": 2.5436,
+      "step": 594
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997251079708788e-06,
+      "loss": 2.6748,
+      "step": 595
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997192908557322e-06,
+      "loss": 2.5585,
+      "step": 596
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99713412851306e-06,
+      "loss": 2.5893,
+      "step": 597
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997074739583162e-06,
+      "loss": 2.4507,
+      "step": 598
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.997014741774866e-06,
+      "loss": 2.5017,
+      "step": 599
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99695413509548e-06,
+      "loss": 2.5674,
+      "step": 600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99689291955239e-06,
+      "loss": 2.6559,
+      "step": 601
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.996831095153054e-06,
+      "loss": 2.5557,
+      "step": 602
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.996768661905008e-06,
+      "loss": 2.5239,
+      "step": 603
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.996705619815857e-06,
+      "loss": 2.5851,
+      "step": 604
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.996641968893281e-06,
+      "loss": 2.5703,
+      "step": 605
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99657770914504e-06,
+      "loss": 2.6728,
+      "step": 606
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99651284057896e-06,
+      "loss": 2.4436,
+      "step": 607
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.996447363202947e-06,
+      "loss": 2.6013,
+      "step": 608
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.996381277024978e-06,
+      "loss": 2.5005,
+      "step": 609
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.996314582053106e-06,
+      "loss": 2.5391,
+      "step": 610
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99624727829546e-06,
+      "loss": 2.5511,
+      "step": 611
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.996179365760235e-06,
+      "loss": 2.547,
+      "step": 612
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99611084445571e-06,
+      "loss": 2.4882,
+      "step": 613
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.996041714390235e-06,
+      "loss": 2.5536,
+      "step": 614
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.995971975572231e-06,
+      "loss": 2.5276,
+      "step": 615
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.995901628010196e-06,
+      "loss": 2.4957,
+      "step": 616
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.995830671712701e-06,
+      "loss": 2.5935,
+      "step": 617
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995759106688394e-06,
+      "loss": 2.508,
+      "step": 618
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995686932945993e-06,
+      "loss": 2.5433,
+      "step": 619
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995614150494293e-06,
+      "loss": 2.5211,
+      "step": 620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995540759342161e-06,
+      "loss": 2.4442,
+      "step": 621
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995466759498543e-06,
+      "loss": 2.4787,
+      "step": 622
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995392150972451e-06,
+      "loss": 2.5415,
+      "step": 623
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995316933772978e-06,
+      "loss": 2.5458,
+      "step": 624
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99524110790929e-06,
+      "loss": 2.6484,
+      "step": 625
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995164673390624e-06,
+      "loss": 2.4705,
+      "step": 626
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995087630226295e-06,
+      "loss": 2.5334,
+      "step": 627
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995009978425692e-06,
+      "loss": 2.5552,
+      "step": 628
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994931717998272e-06,
+      "loss": 2.6088,
+      "step": 629
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994852848953574e-06,
+      "loss": 2.513,
+      "step": 630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994773371301208e-06,
+      "loss": 2.562,
+      "step": 631
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994693285050858e-06,
+      "loss": 2.5564,
+      "step": 632
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99461259021228e-06,
+      "loss": 2.5494,
+      "step": 633
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994531286795309e-06,
+      "loss": 2.5474,
+      "step": 634
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994449374809851e-06,
+      "loss": 2.5486,
+      "step": 635
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994366854265886e-06,
+      "loss": 2.4385,
+      "step": 636
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994283725173468e-06,
+      "loss": 2.5627,
+      "step": 637
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99419998754273e-06,
+      "loss": 2.5568,
+      "step": 638
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994115641383872e-06,
+      "loss": 2.5238,
+      "step": 639
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994030686707171e-06,
+      "loss": 2.5554,
+      "step": 640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99394512352298e-06,
+      "loss": 2.568,
+      "step": 641
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993858951841724e-06,
+      "loss": 2.6618,
+      "step": 642
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993772171673901e-06,
+      "loss": 2.5176,
+      "step": 643
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99368478303009e-06,
+      "loss": 2.443,
+      "step": 644
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993596785920932e-06,
+      "loss": 2.6057,
+      "step": 645
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993508180357154e-06,
+      "loss": 2.5331,
+      "step": 646
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993418966349551e-06,
+      "loss": 2.5112,
+      "step": 647
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993329143908994e-06,
+      "loss": 2.5452,
+      "step": 648
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993238713046428e-06,
+      "loss": 2.5267,
+      "step": 649
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993147673772869e-06,
+      "loss": 2.513,
+      "step": 650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993056026099415e-06,
+      "loss": 2.5051,
+      "step": 651
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.992963770037227e-06,
+      "loss": 2.4977,
+      "step": 652
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.992870905597549e-06,
+      "loss": 2.5676,
+      "step": 653
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.992777432791697e-06,
+      "loss": 2.5637,
+      "step": 654
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99268335163106e-06,
+      "loss": 2.5236,
+      "step": 655
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9925886621271e-06,
+      "loss": 2.5703,
+      "step": 656
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.992493364291356e-06,
+      "loss": 2.528,
+      "step": 657
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.992397458135438e-06,
+      "loss": 2.5707,
+      "step": 658
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.992300943671035e-06,
+      "loss": 2.5729,
+      "step": 659
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.992203820909906e-06,
+      "loss": 2.4917,
+      "step": 660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.992106089863884e-06,
+      "loss": 2.5345,
+      "step": 661
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.992007750544876e-06,
+      "loss": 2.4378,
+      "step": 662
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.991908802964867e-06,
+      "loss": 2.5127,
+      "step": 663
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.991809247135912e-06,
+      "loss": 2.477,
+      "step": 664
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.991709083070143e-06,
+      "loss": 2.5675,
+      "step": 665
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.991608310779762e-06,
+      "loss": 2.5636,
+      "step": 666
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99150693027705e-06,
+      "loss": 2.5898,
+      "step": 667
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99140494157436e-06,
+      "loss": 2.5436,
+      "step": 668
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99130234468412e-06,
+      "loss": 2.4627,
+      "step": 669
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.991199139618828e-06,
+      "loss": 2.5095,
+      "step": 670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.991095326391061e-06,
+      "loss": 2.5023,
+      "step": 671
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.990990905013466e-06,
+      "loss": 2.5546,
+      "step": 672
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99088587549877e-06,
+      "loss": 2.5326,
+      "step": 673
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99078023785977e-06,
+      "loss": 2.5838,
+      "step": 674
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.990673992109335e-06,
+      "loss": 2.466,
+      "step": 675
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.990567138260414e-06,
+      "loss": 2.533,
+      "step": 676
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.990459676326025e-06,
+      "loss": 2.576,
+      "step": 677
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.990351606319261e-06,
+      "loss": 2.5243,
+      "step": 678
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.990242928253291e-06,
+      "loss": 2.5035,
+      "step": 679
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.990133642141359e-06,
+      "loss": 2.4692,
+      "step": 680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.990023747996778e-06,
+      "loss": 2.4879,
+      "step": 681
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.98991324583294e-06,
+      "loss": 2.5204,
+      "step": 682
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.989802135663308e-06,
+      "loss": 2.3926,
+      "step": 683
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.989690417501423e-06,
+      "loss": 2.4331,
+      "step": 684
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.989578091360896e-06,
+      "loss": 2.4863,
+      "step": 685
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.989465157255413e-06,
+      "loss": 2.6121,
+      "step": 686
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.989351615198734e-06,
+      "loss": 2.5101,
+      "step": 687
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.989237465204698e-06,
+      "loss": 2.5139,
+      "step": 688
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.98912270728721e-06,
+      "loss": 2.5652,
+      "step": 689
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.989007341460251e-06,
+      "loss": 2.4658,
+      "step": 690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.988891367737882e-06,
+      "loss": 2.4625,
+      "step": 691
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.988774786134235e-06,
+      "loss": 2.6152,
+      "step": 692
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.988657596663509e-06,
+      "loss": 2.5233,
+      "step": 693
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.988539799339989e-06,
+      "loss": 2.586,
+      "step": 694
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.988421394178027e-06,
+      "loss": 2.5555,
+      "step": 695
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.98830238119205e-06,
+      "loss": 2.4976,
+      "step": 696
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.988182760396557e-06,
+      "loss": 2.5319,
+      "step": 697
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.988062531806127e-06,
+      "loss": 2.5239,
+      "step": 698
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.987941695435409e-06,
+      "loss": 2.4937,
+      "step": 699
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.987820251299121e-06,
+      "loss": 2.6226,
+      "step": 700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.98769819941207e-06,
+      "loss": 2.4845,
+      "step": 701
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.987575539789119e-06,
+      "loss": 2.4652,
+      "step": 702
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.98745227244522e-06,
+      "loss": 2.4892,
+      "step": 703
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.987328397395389e-06,
+      "loss": 2.6209,
+      "step": 704
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.987203914654721e-06,
+      "loss": 2.5036,
+      "step": 705
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.987078824238384e-06,
+      "loss": 2.502,
+      "step": 706
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.98695312616162e-06,
+      "loss": 2.6344,
+      "step": 707
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.986826820439743e-06,
+      "loss": 2.5166,
+      "step": 708
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.986699907088147e-06,
+      "loss": 2.5017,
+      "step": 709
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.98657238612229e-06,
+      "loss": 2.5157,
+      "step": 710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.986444257557717e-06,
+      "loss": 2.5016,
+      "step": 711
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.986315521410035e-06,
+      "loss": 2.5031,
+      "step": 712
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986186177694935e-06,
+      "loss": 2.4967,
+      "step": 713
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98605622642817e-06,
+      "loss": 2.5021,
+      "step": 714
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985925667625581e-06,
+      "loss": 2.4973,
+      "step": 715
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98579450130307e-06,
+      "loss": 2.5149,
+      "step": 716
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985662727476625e-06,
+      "loss": 2.5589,
+      "step": 717
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9855303461623e-06,
+      "loss": 2.493,
+      "step": 718
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985397357376224e-06,
+      "loss": 2.5017,
+      "step": 719
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985263761134602e-06,
+      "loss": 2.5647,
+      "step": 720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985129557453714e-06,
+      "loss": 2.4461,
+      "step": 721
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98499474634991e-06,
+      "loss": 2.5062,
+      "step": 722
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984859327839617e-06,
+      "loss": 2.4892,
+      "step": 723
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984723301939337e-06,
+      "loss": 2.506,
+      "step": 724
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984586668665641e-06,
+      "loss": 2.5006,
+      "step": 725
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98444942803518e-06,
+      "loss": 2.5452,
+      "step": 726
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984311580064676e-06,
+      "loss": 2.5566,
+      "step": 727
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984173124770924e-06,
+      "loss": 2.5026,
+      "step": 728
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984034062170796e-06,
+      "loss": 2.4849,
+      "step": 729
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983894392281237e-06,
+      "loss": 2.5331,
+      "step": 730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983754115119262e-06,
+      "loss": 2.4972,
+      "step": 731
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983613230701967e-06,
+      "loss": 2.5288,
+      "step": 732
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983471739046515e-06,
+      "loss": 2.5854,
+      "step": 733
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98332964017015e-06,
+      "loss": 2.5227,
+      "step": 734
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983186934090183e-06,
+      "loss": 2.5356,
+      "step": 735
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983043620824005e-06,
+      "loss": 2.5444,
+      "step": 736
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.982899700389077e-06,
+      "loss": 2.5282,
+      "step": 737
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.982755172802933e-06,
+      "loss": 2.4933,
+      "step": 738
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.982610038083188e-06,
+      "loss": 2.5409,
+      "step": 739
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.982464296247523e-06,
+      "loss": 2.4976,
+      "step": 740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.982317947313695e-06,
+      "loss": 2.5168,
+      "step": 741
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98217099129954e-06,
+      "loss": 2.4238,
+      "step": 742
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.982023428222963e-06,
+      "loss": 2.517,
+      "step": 743
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.981875258101944e-06,
+      "loss": 2.5476,
+      "step": 744
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.981726480954532e-06,
+      "loss": 2.5152,
+      "step": 745
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.981577096798864e-06,
+      "loss": 2.4839,
+      "step": 746
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.981427105653135e-06,
+      "loss": 2.4795,
+      "step": 747
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.981276507535625e-06,
+      "loss": 2.5031,
+      "step": 748
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.981125302464681e-06,
+      "loss": 2.4245,
+      "step": 749
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.980973490458728e-06,
+      "loss": 2.5285,
+      "step": 750
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.980821071536266e-06,
+      "loss": 2.5549,
+      "step": 751
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.980668045715864e-06,
+      "loss": 2.3846,
+      "step": 752
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.980514413016167e-06,
+      "loss": 2.4627,
+      "step": 753
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.980360173455899e-06,
+      "loss": 2.4511,
+      "step": 754
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98020532705385e-06,
+      "loss": 2.5837,
+      "step": 755
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.980049873828887e-06,
+      "loss": 2.4315,
+      "step": 756
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.979893813799953e-06,
+      "loss": 2.5019,
+      "step": 757
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.979737146986064e-06,
+      "loss": 2.5113,
+      "step": 758
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.97957987340631e-06,
+      "loss": 2.4797,
+      "step": 759
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.979421993079853e-06,
+      "loss": 2.5887,
+      "step": 760
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.97926350602593e-06,
+      "loss": 2.4945,
+      "step": 761
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.979104412263851e-06,
+      "loss": 2.5772,
+      "step": 762
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.978944711813003e-06,
+      "loss": 2.5756,
+      "step": 763
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.978784404692847e-06,
+      "loss": 2.5185,
+      "step": 764
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.978623490922913e-06,
+      "loss": 2.5697,
+      "step": 765
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.978461970522807e-06,
+      "loss": 2.5404,
+      "step": 766
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.97829984351221e-06,
+      "loss": 2.52,
+      "step": 767
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.97813710991088e-06,
+      "loss": 2.5819,
+      "step": 768
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.977973769738642e-06,
+      "loss": 2.4221,
+      "step": 769
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9778098230154e-06,
+      "loss": 2.4466,
+      "step": 770
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.977645269761131e-06,
+      "loss": 2.5037,
+      "step": 771
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.977480109995886e-06,
+      "loss": 2.4762,
+      "step": 772
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.977314343739785e-06,
+      "loss": 2.5069,
+      "step": 773
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.977147971013033e-06,
+      "loss": 2.4696,
+      "step": 774
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.976980991835896e-06,
+      "loss": 2.514,
+      "step": 775
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.97681340622872e-06,
+      "loss": 2.4735,
+      "step": 776
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.976645214211929e-06,
+      "loss": 2.5103,
+      "step": 777
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.976476415806013e-06,
+      "loss": 2.5297,
+      "step": 778
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.976307011031542e-06,
+      "loss": 2.5879,
+      "step": 779
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.976136999909156e-06,
+      "loss": 2.5625,
+      "step": 780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.975966382459571e-06,
+      "loss": 2.4413,
+      "step": 781
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.975795158703576e-06,
+      "loss": 2.5502,
+      "step": 782
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.975623328662036e-06,
+      "loss": 2.5488,
+      "step": 783
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.975450892355882e-06,
+      "loss": 2.5225,
+      "step": 784
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.975277849806133e-06,
+      "loss": 2.5308,
+      "step": 785
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.975104201033868e-06,
+      "loss": 2.3977,
+      "step": 786
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.974929946060246e-06,
+      "loss": 2.5289,
+      "step": 787
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.974755084906503e-06,
+      "loss": 2.4814,
+      "step": 788
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.97457961759394e-06,
+      "loss": 2.4669,
+      "step": 789
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.974403544143942e-06,
+      "loss": 2.5672,
+      "step": 790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.97422686457796e-06,
+      "loss": 2.5671,
+      "step": 791
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.974049578917524e-06,
+      "loss": 2.5506,
+      "step": 792
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.973871687184234e-06,
+      "loss": 2.5542,
+      "step": 793
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.973693189399767e-06,
+      "loss": 2.4818,
+      "step": 794
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.973514085585871e-06,
+      "loss": 2.5224,
+      "step": 795
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.973334375764372e-06,
+      "loss": 2.5112,
+      "step": 796
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.973154059957162e-06,
+      "loss": 2.4854,
+      "step": 797
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.972973138186217e-06,
+      "loss": 2.4906,
+      "step": 798
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.972791610473578e-06,
+      "loss": 2.5121,
+      "step": 799
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.972609476841368e-06,
+      "loss": 2.5144,
+      "step": 800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.972426737311775e-06,
+      "loss": 2.5635,
+      "step": 801
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.972243391907068e-06,
+      "loss": 2.4624,
+      "step": 802
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.972059440649584e-06,
+      "loss": 2.4683,
+      "step": 803
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.97187488356174e-06,
+      "loss": 2.5512,
+      "step": 804
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.971689720666024e-06,
+      "loss": 2.5775,
+      "step": 805
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.971503951984996e-06,
+      "loss": 2.477,
+      "step": 806
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.97131757754129e-06,
+      "loss": 2.4246,
+      "step": 807
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.971130597357618e-06,
+      "loss": 2.5103,
+      "step": 808
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.970943011456762e-06,
+      "loss": 2.5245,
+      "step": 809
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.970754819861577e-06,
+      "loss": 2.496,
+      "step": 810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.970566022594996e-06,
+      "loss": 2.5924,
+      "step": 811
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.970376619680024e-06,
+      "loss": 2.5105,
+      "step": 812
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.970186611139736e-06,
+      "loss": 2.4686,
+      "step": 813
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.969995996997285e-06,
+      "loss": 2.4833,
+      "step": 814
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.9698047772759e-06,
+      "loss": 2.5503,
+      "step": 815
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.969612951998874e-06,
+      "loss": 2.4722,
+      "step": 816
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.969420521189587e-06,
+      "loss": 2.545,
+      "step": 817
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.969227484871485e-06,
+      "loss": 2.423,
+      "step": 818
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.969033843068083e-06,
+      "loss": 2.5963,
+      "step": 819
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.968839595802982e-06,
+      "loss": 2.6228,
+      "step": 820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.968644743099848e-06,
+      "loss": 2.5352,
+      "step": 821
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.968449284982424e-06,
+      "loss": 2.5133,
+      "step": 822
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.968253221474527e-06,
+      "loss": 2.5247,
+      "step": 823
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.968056552600043e-06,
+      "loss": 2.5018,
+      "step": 824
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.967859278382939e-06,
+      "loss": 2.5339,
+      "step": 825
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96766139884725e-06,
+      "loss": 2.5611,
+      "step": 826
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.967462914017087e-06,
+      "loss": 2.5117,
+      "step": 827
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.967263823916638e-06,
+      "loss": 2.5082,
+      "step": 828
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96706412857016e-06,
+      "loss": 2.5511,
+      "step": 829
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.966863828001982e-06,
+      "loss": 2.5143,
+      "step": 830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.966662922236515e-06,
+      "loss": 2.4449,
+      "step": 831
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.966461411298235e-06,
+      "loss": 2.4417,
+      "step": 832
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.966259295211698e-06,
+      "loss": 2.5104,
+      "step": 833
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.966056574001528e-06,
+      "loss": 2.4811,
+      "step": 834
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.965853247692433e-06,
+      "loss": 2.4782,
+      "step": 835
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.965649316309178e-06,
+      "loss": 2.4634,
+      "step": 836
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.965444779876618e-06,
+      "loss": 2.5298,
+      "step": 837
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.965239638419673e-06,
+      "loss": 2.5186,
+      "step": 838
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.965033891963338e-06,
+      "loss": 2.4464,
+      "step": 839
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.964827540532685e-06,
+      "loss": 2.5016,
+      "step": 840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.964620584152858e-06,
+      "loss": 2.5038,
+      "step": 841
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.964413022849069e-06,
+      "loss": 2.4964,
+      "step": 842
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.964204856646613e-06,
+      "loss": 2.4913,
+      "step": 843
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.963996085570854e-06,
+      "loss": 2.4627,
+      "step": 844
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.963786709647228e-06,
+      "loss": 2.4393,
+      "step": 845
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96357672890125e-06,
+      "loss": 2.4487,
+      "step": 846
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.963366143358502e-06,
+      "loss": 2.4926,
+      "step": 847
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.963154953044646e-06,
+      "loss": 2.5192,
+      "step": 848
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.962943157985412e-06,
+      "loss": 2.4483,
+      "step": 849
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.962730758206612e-06,
+      "loss": 2.5023,
+      "step": 850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96251775373412e-06,
+      "loss": 2.5327,
+      "step": 851
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.962304144593893e-06,
+      "loss": 2.5021,
+      "step": 852
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.962089930811959e-06,
+      "loss": 2.4531,
+      "step": 853
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.961875112414417e-06,
+      "loss": 2.5197,
+      "step": 854
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.961659689427444e-06,
+      "loss": 2.4303,
+      "step": 855
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96144366187729e-06,
+      "loss": 2.4524,
+      "step": 856
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.961227029790272e-06,
+      "loss": 2.5114,
+      "step": 857
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.961009793192793e-06,
+      "loss": 2.5113,
+      "step": 858
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.960791952111318e-06,
+      "loss": 2.6098,
+      "step": 859
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.960573506572391e-06,
+      "loss": 2.4632,
+      "step": 860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.960354456602627e-06,
+      "loss": 2.4486,
+      "step": 861
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.960134802228722e-06,
+      "loss": 2.473,
+      "step": 862
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.959914543477436e-06,
+      "loss": 2.4877,
+      "step": 863
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.959693680375608e-06,
+      "loss": 2.4147,
+      "step": 864
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.95947221295015e-06,
+      "loss": 2.4261,
+      "step": 865
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.959250141228046e-06,
+      "loss": 2.5326,
+      "step": 866
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.959027465236354e-06,
+      "loss": 2.4713,
+      "step": 867
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.958804185002209e-06,
+      "loss": 2.4715,
+      "step": 868
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.958580300552816e-06,
+      "loss": 2.5164,
+      "step": 869
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.958355811915452e-06,
+      "loss": 2.48,
+      "step": 870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.958130719117476e-06,
+      "loss": 2.4448,
+      "step": 871
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.957905022186309e-06,
+      "loss": 2.4367,
+      "step": 872
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.957678721149454e-06,
+      "loss": 2.4603,
+      "step": 873
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.957451816034487e-06,
+      "loss": 2.5309,
+      "step": 874
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.957224306869053e-06,
+      "loss": 2.5113,
+      "step": 875
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.956996193680874e-06,
+      "loss": 2.4974,
+      "step": 876
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.956767476497745e-06,
+      "loss": 2.5109,
+      "step": 877
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.956538155347534e-06,
+      "loss": 2.4329,
+      "step": 878
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.956308230258182e-06,
+      "loss": 2.5357,
+      "step": 879
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.95607770125771e-06,
+      "loss": 2.4665,
+      "step": 880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.955846568374201e-06,
+      "loss": 2.443,
+      "step": 881
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.95561483163582e-06,
+      "loss": 2.4592,
+      "step": 882
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.955382491070806e-06,
+      "loss": 2.49,
+      "step": 883
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.955149546707465e-06,
+      "loss": 2.4107,
+      "step": 884
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.954915998574182e-06,
+      "loss": 2.5638,
+      "step": 885
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.954681846699414e-06,
+      "loss": 2.4741,
+      "step": 886
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.954447091111695e-06,
+      "loss": 2.4708,
+      "step": 887
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.954211731839623e-06,
+      "loss": 2.5064,
+      "step": 888
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.953975768911881e-06,
+      "loss": 2.4307,
+      "step": 889
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.953739202357219e-06,
+      "loss": 2.5701,
+      "step": 890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.953502032204461e-06,
+      "loss": 2.4547,
+      "step": 891
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.953264258482505e-06,
+      "loss": 2.4311,
+      "step": 892
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.953025881220325e-06,
+      "loss": 2.5004,
+      "step": 893
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.952786900446964e-06,
+      "loss": 2.497,
+      "step": 894
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.952547316191545e-06,
+      "loss": 2.5596,
+      "step": 895
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.952307128483257e-06,
+      "loss": 2.4631,
+      "step": 896
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.952066337351367e-06,
+      "loss": 2.503,
+      "step": 897
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.951824942825215e-06,
+      "loss": 2.4948,
+      "step": 898
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.951582944934215e-06,
+      "loss": 2.4392,
+      "step": 899
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.951340343707852e-06,
+      "loss": 2.4234,
+      "step": 900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.951097139175688e-06,
+      "loss": 2.5812,
+      "step": 901
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.950853331367356e-06,
+      "loss": 2.5741,
+      "step": 902
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95060892031256e-06,
+      "loss": 2.4917,
+      "step": 903
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.950363906041089e-06,
+      "loss": 2.5381,
+      "step": 904
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95011828858279e-06,
+      "loss": 2.398,
+      "step": 905
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94987206796759e-06,
+      "loss": 2.47,
+      "step": 906
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.949625244225496e-06,
+      "loss": 2.4338,
+      "step": 907
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94937781738658e-06,
+      "loss": 2.535,
+      "step": 908
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.949129787480988e-06,
+      "loss": 2.4871,
+      "step": 909
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.948881154538946e-06,
+      "loss": 2.457,
+      "step": 910
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.948631918590746e-06,
+      "loss": 2.4936,
+      "step": 911
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.948382079666756e-06,
+      "loss": 2.4314,
+      "step": 912
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94813163779742e-06,
+      "loss": 2.4148,
+      "step": 913
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.947880593013256e-06,
+      "loss": 2.5468,
+      "step": 914
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.947628945344849e-06,
+      "loss": 2.4106,
+      "step": 915
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.947376694822861e-06,
+      "loss": 2.4614,
+      "step": 916
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.947123841478032e-06,
+      "loss": 2.4957,
+      "step": 917
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.946870385341167e-06,
+      "loss": 2.4909,
+      "step": 918
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.946616326443153e-06,
+      "loss": 2.5706,
+      "step": 919
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.946361664814942e-06,
+      "loss": 2.4635,
+      "step": 920
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.946106400487568e-06,
+      "loss": 2.4782,
+      "step": 921
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.945850533492132e-06,
+      "loss": 2.4466,
+      "step": 922
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94559406385981e-06,
+      "loss": 2.5117,
+      "step": 923
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.945336991621854e-06,
+      "loss": 2.4909,
+      "step": 924
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.945079316809585e-06,
+      "loss": 2.4971,
+      "step": 925
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.944821039454403e-06,
+      "loss": 2.5662,
+      "step": 926
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.944562159587774e-06,
+      "loss": 2.5095,
+      "step": 927
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.944302677241247e-06,
+      "loss": 2.4938,
+      "step": 928
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.944042592446434e-06,
+      "loss": 2.483,
+      "step": 929
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94378190523503e-06,
+      "loss": 2.4676,
+      "step": 930
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.943520615638796e-06,
+      "loss": 2.4567,
+      "step": 931
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94325872368957e-06,
+      "loss": 2.4356,
+      "step": 932
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.942996229419264e-06,
+      "loss": 2.4542,
+      "step": 933
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.942733132859861e-06,
+      "loss": 2.5415,
+      "step": 934
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.942469434043418e-06,
+      "loss": 2.5092,
+      "step": 935
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.942205133002067e-06,
+      "loss": 2.5017,
+      "step": 936
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.941940229768012e-06,
+      "loss": 2.5196,
+      "step": 937
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94167472437353e-06,
+      "loss": 2.4901,
+      "step": 938
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.941408616850974e-06,
+      "loss": 2.4169,
+      "step": 939
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.941141907232766e-06,
+      "loss": 2.5124,
+      "step": 940
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.940874595551403e-06,
+      "loss": 2.5009,
+      "step": 941
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94060668183946e-06,
+      "loss": 2.5311,
+      "step": 942
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.940338166129578e-06,
+      "loss": 2.42,
+      "step": 943
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.940069048454478e-06,
+      "loss": 2.4855,
+      "step": 944
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.939799328846947e-06,
+      "loss": 2.5272,
+      "step": 945
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.939529007339852e-06,
+      "loss": 2.4992,
+      "step": 946
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.93925808396613e-06,
+      "loss": 2.4105,
+      "step": 947
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.938986558758795e-06,
+      "loss": 2.4888,
+      "step": 948
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.938714431750928e-06,
+      "loss": 2.4907,
+      "step": 949
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.938441702975689e-06,
+      "loss": 2.4703,
+      "step": 950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.938168372466308e-06,
+      "loss": 2.4762,
+      "step": 951
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.937894440256091e-06,
+      "loss": 2.4958,
+      "step": 952
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.937619906378413e-06,
+      "loss": 2.471,
+      "step": 953
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.937344770866727e-06,
+      "loss": 2.5342,
+      "step": 954
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.937069033754558e-06,
+      "loss": 2.4644,
+      "step": 955
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.936792695075502e-06,
+      "loss": 2.4877,
+      "step": 956
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.936515754863231e-06,
+      "loss": 2.4167,
+      "step": 957
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.936238213151491e-06,
+      "loss": 2.456,
+      "step": 958
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.935960069974096e-06,
+      "loss": 2.399,
+      "step": 959
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.93568132536494e-06,
+      "loss": 2.4469,
+      "step": 960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.935401979357985e-06,
+      "loss": 2.4427,
+      "step": 961
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.93512203198727e-06,
+      "loss": 2.5516,
+      "step": 962
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.934841483286907e-06,
+      "loss": 2.4803,
+      "step": 963
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.934560333291077e-06,
+      "loss": 2.4728,
+      "step": 964
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.934278582034037e-06,
+      "loss": 2.4626,
+      "step": 965
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.93399622955012e-06,
+      "loss": 2.5142,
+      "step": 966
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.933713275873728e-06,
+      "loss": 2.4862,
+      "step": 967
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.93342972103934e-06,
+      "loss": 2.4432,
+      "step": 968
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.933145565081506e-06,
+      "loss": 2.4892,
+      "step": 969
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.932860808034847e-06,
+      "loss": 2.5269,
+      "step": 970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.932575449934063e-06,
+      "loss": 2.4569,
+      "step": 971
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.932289490813922e-06,
+      "loss": 2.4839,
+      "step": 972
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.932002930709268e-06,
+      "loss": 2.5311,
+      "step": 973
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.931715769655017e-06,
+      "loss": 2.4644,
+      "step": 974
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.931428007686158e-06,
+      "loss": 2.5167,
+      "step": 975
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.931139644837755e-06,
+      "loss": 2.4316,
+      "step": 976
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.930850681144946e-06,
+      "loss": 2.4767,
+      "step": 977
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.930561116642936e-06,
+      "loss": 2.4881,
+      "step": 978
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.930270951367012e-06,
+      "loss": 2.397,
+      "step": 979
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.929980185352525e-06,
+      "loss": 2.5083,
+      "step": 980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.929688818634909e-06,
+      "loss": 2.5191,
+      "step": 981
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.929396851249661e-06,
+      "loss": 2.4851,
+      "step": 982
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.929104283232363e-06,
+      "loss": 2.4141,
+      "step": 983
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.928811114618658e-06,
+      "loss": 2.4767,
+      "step": 984
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.92851734544427e-06,
+      "loss": 2.4311,
+      "step": 985
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.928222975744992e-06,
+      "loss": 2.4768,
+      "step": 986
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.927928005556692e-06,
+      "loss": 2.4952,
+      "step": 987
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.927632434915315e-06,
+      "loss": 2.4158,
+      "step": 988
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.927336263856873e-06,
+      "loss": 2.5188,
+      "step": 989
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.927039492417452e-06,
+      "loss": 2.4864,
+      "step": 990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.926742120633215e-06,
+      "loss": 2.473,
+      "step": 991
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.926444148540394e-06,
+      "loss": 2.4979,
+      "step": 992
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.926145576175297e-06,
+      "loss": 2.4567,
+      "step": 993
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.925846403574306e-06,
+      "loss": 2.5497,
+      "step": 994
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.92554663077387e-06,
+      "loss": 2.536,
+      "step": 995
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.925246257810519e-06,
+      "loss": 2.46,
+      "step": 996
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.924945284720852e-06,
+      "loss": 2.469,
+      "step": 997
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.92464371154154e-06,
+      "loss": 2.5754,
+      "step": 998
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.92434153830933e-06,
+      "loss": 2.4884,
+      "step": 999
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.924038765061042e-06,
+      "loss": 2.5108,
+      "step": 1000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.923735391833564e-06,
+      "loss": 2.4701,
+      "step": 1001
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.923431418663866e-06,
+      "loss": 2.4907,
+      "step": 1002
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.923126845588982e-06,
+      "loss": 2.4711,
+      "step": 1003
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.922821672646028e-06,
+      "loss": 2.4627,
+      "step": 1004
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.922515899872184e-06,
+      "loss": 2.5188,
+      "step": 1005
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.922209527304709e-06,
+      "loss": 2.456,
+      "step": 1006
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.921902554980935e-06,
+      "loss": 2.4872,
+      "step": 1007
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.921594982938262e-06,
+      "loss": 2.4218,
+      "step": 1008
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.921286811214173e-06,
+      "loss": 2.4979,
+      "step": 1009
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.92097803984621e-06,
+      "loss": 2.4051,
+      "step": 1010
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.920668668872002e-06,
+      "loss": 2.5391,
+      "step": 1011
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.920358698329242e-06,
+      "loss": 2.4383,
+      "step": 1012
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.920048128255699e-06,
+      "loss": 2.4622,
+      "step": 1013
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.919736958689216e-06,
+      "loss": 2.5897,
+      "step": 1014
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.91942518966771e-06,
+      "loss": 2.4435,
+      "step": 1015
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.919112821229165e-06,
+      "loss": 2.5461,
+      "step": 1016
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.918799853411642e-06,
+      "loss": 2.4944,
+      "step": 1017
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.918486286253279e-06,
+      "loss": 2.5054,
+      "step": 1018
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.918172119792283e-06,
+      "loss": 2.4937,
+      "step": 1019
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.91785735406693e-06,
+      "loss": 2.5473,
+      "step": 1020
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.917541989115579e-06,
+      "loss": 2.411,
+      "step": 1021
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.91722602497665e-06,
+      "loss": 2.4908,
+      "step": 1022
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.916909461688646e-06,
+      "loss": 2.479,
+      "step": 1023
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.91659229929014e-06,
+      "loss": 2.5249,
+      "step": 1024
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.916274537819774e-06,
+      "loss": 2.4201,
+      "step": 1025
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.915956177316269e-06,
+      "loss": 2.4958,
+      "step": 1026
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.915637217818415e-06,
+      "loss": 2.4876,
+      "step": 1027
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.915317659365078e-06,
+      "loss": 2.4051,
+      "step": 1028
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.914997501995193e-06,
+      "loss": 2.4803,
+      "step": 1029
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.914676745747772e-06,
+      "loss": 2.45,
+      "step": 1030
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.914355390661897e-06,
+      "loss": 2.4193,
+      "step": 1031
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.914033436776724e-06,
+      "loss": 2.3931,
+      "step": 1032
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.913710884131483e-06,
+      "loss": 2.4805,
+      "step": 1033
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.913387732765475e-06,
+      "loss": 2.4983,
+      "step": 1034
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.913063982718076e-06,
+      "loss": 2.3899,
+      "step": 1035
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.912739634028734e-06,
+      "loss": 2.4282,
+      "step": 1036
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.912414686736971e-06,
+      "loss": 2.4395,
+      "step": 1037
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.912089140882377e-06,
+      "loss": 2.4527,
+      "step": 1038
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.911762996504621e-06,
+      "loss": 2.3951,
+      "step": 1039
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.911436253643445e-06,
+      "loss": 2.4808,
+      "step": 1040
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.911108912338656e-06,
+      "loss": 2.4659,
+      "step": 1041
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.910780972630146e-06,
+      "loss": 2.5054,
+      "step": 1042
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.91045243455787e-06,
+      "loss": 2.4695,
+      "step": 1043
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.91012329816186e-06,
+      "loss": 2.4547,
+      "step": 1044
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.90979356348222e-06,
+      "loss": 2.4224,
+      "step": 1045
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.909463230559127e-06,
+      "loss": 2.5103,
+      "step": 1046
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.90913229943283e-06,
+      "loss": 2.5056,
+      "step": 1047
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.908800770143654e-06,
+      "loss": 2.4598,
+      "step": 1048
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.908468642731996e-06,
+      "loss": 2.4294,
+      "step": 1049
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.908135917238321e-06,
+      "loss": 2.4751,
+      "step": 1050
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.907802593703173e-06,
+      "loss": 2.5392,
+      "step": 1051
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.907468672167165e-06,
+      "loss": 2.4591,
+      "step": 1052
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.907134152670987e-06,
+      "loss": 2.4479,
+      "step": 1053
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.906799035255395e-06,
+      "loss": 2.4422,
+      "step": 1054
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.906463319961225e-06,
+      "loss": 2.4893,
+      "step": 1055
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.906127006829385e-06,
+      "loss": 2.5179,
+      "step": 1056
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.905790095900849e-06,
+      "loss": 2.4768,
+      "step": 1057
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.90545258721667e-06,
+      "loss": 2.4889,
+      "step": 1058
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.905114480817976e-06,
+      "loss": 2.4525,
+      "step": 1059
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.904775776745959e-06,
+      "loss": 2.4944,
+      "step": 1060
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.904436475041892e-06,
+      "loss": 2.5283,
+      "step": 1061
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.904096575747117e-06,
+      "loss": 2.4651,
+      "step": 1062
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.90375607890305e-06,
+      "loss": 2.366,
+      "step": 1063
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.903414984551178e-06,
+      "loss": 2.4949,
+      "step": 1064
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.903073292733065e-06,
+      "loss": 2.492,
+      "step": 1065
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.902731003490344e-06,
+      "loss": 2.5042,
+      "step": 1066
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.902388116864723e-06,
+      "loss": 2.5041,
+      "step": 1067
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.90204463289798e-06,
+      "loss": 2.4571,
+      "step": 1068
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.901700551631966e-06,
+      "loss": 2.468,
+      "step": 1069
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.901355873108611e-06,
+      "loss": 2.4957,
+      "step": 1070
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.901010597369908e-06,
+      "loss": 2.4509,
+      "step": 1071
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.900664724457932e-06,
+      "loss": 2.5079,
+      "step": 1072
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.900318254414823e-06,
+      "loss": 2.4366,
+      "step": 1073
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.899971187282799e-06,
+      "loss": 2.481,
+      "step": 1074
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.899623523104149e-06,
+      "loss": 2.4321,
+      "step": 1075
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.899275261921236e-06,
+      "loss": 2.467,
+      "step": 1076
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.898926403776492e-06,
+      "loss": 2.5289,
+      "step": 1077
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.898576948712427e-06,
+      "loss": 2.4666,
+      "step": 1078
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.898226896771619e-06,
+      "loss": 2.4938,
+      "step": 1079
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.89787624799672e-06,
+      "loss": 2.4575,
+      "step": 1080
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.897525002430459e-06,
+      "loss": 2.4337,
+      "step": 1081
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.897173160115633e-06,
+      "loss": 2.4769,
+      "step": 1082
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.89682072109511e-06,
+      "loss": 2.4873,
+      "step": 1083
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.896467685411838e-06,
+      "loss": 2.4163,
+      "step": 1084
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.89611405310883e-06,
+      "loss": 2.389,
+      "step": 1085
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.895759824229176e-06,
+      "loss": 2.4985,
+      "step": 1086
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.895404998816038e-06,
+      "loss": 2.4611,
+      "step": 1087
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.89504957691265e-06,
+      "loss": 2.4895,
+      "step": 1088
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.894693558562319e-06,
+      "loss": 2.4701,
+      "step": 1089
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.894336943808426e-06,
+      "loss": 2.4736,
+      "step": 1090
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.893979732694422e-06,
+      "loss": 2.4966,
+      "step": 1091
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.893621925263832e-06,
+      "loss": 2.4751,
+      "step": 1092
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.893263521560255e-06,
+      "loss": 2.3725,
+      "step": 1093
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.89290452162736e-06,
+      "loss": 2.4425,
+      "step": 1094
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.892544925508894e-06,
+      "loss": 2.469,
+      "step": 1095
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.892184733248666e-06,
+      "loss": 2.46,
+      "step": 1096
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.891823944890569e-06,
+      "loss": 2.4476,
+      "step": 1097
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.891462560478562e-06,
+      "loss": 2.4585,
+      "step": 1098
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.891100580056681e-06,
+      "loss": 2.4242,
+      "step": 1099
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.890738003669029e-06,
+      "loss": 2.4888,
+      "step": 1100
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.890374831359787e-06,
+      "loss": 2.4824,
+      "step": 1101
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.890011063173207e-06,
+      "loss": 2.4096,
+      "step": 1102
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.88964669915361e-06,
+      "loss": 2.4523,
+      "step": 1103
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.889281739345395e-06,
+      "loss": 2.516,
+      "step": 1104
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.88891618379303e-06,
+      "loss": 2.4601,
+      "step": 1105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.88855003254106e-06,
+      "loss": 2.5807,
+      "step": 1106
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.888183285634097e-06,
+      "loss": 2.5105,
+      "step": 1107
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.887815943116827e-06,
+      "loss": 2.4794,
+      "step": 1108
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.887448005034011e-06,
+      "loss": 2.3977,
+      "step": 1109
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.887079471430481e-06,
+      "loss": 2.4681,
+      "step": 1110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.88671034235114e-06,
+      "loss": 2.5993,
+      "step": 1111
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.886340617840968e-06,
+      "loss": 2.499,
+      "step": 1112
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.885970297945013e-06,
+      "loss": 2.4189,
+      "step": 1113
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.8855993827084e-06,
+      "loss": 2.458,
+      "step": 1114
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.88522787217632e-06,
+      "loss": 2.429,
+      "step": 1115
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.884855766394041e-06,
+      "loss": 2.4341,
+      "step": 1116
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.884483065406905e-06,
+      "loss": 2.3764,
+      "step": 1117
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.884109769260326e-06,
+      "loss": 2.4526,
+      "step": 1118
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.883735877999785e-06,
+      "loss": 2.4453,
+      "step": 1119
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.883361391670841e-06,
+      "loss": 2.4138,
+      "step": 1120
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.882986310319124e-06,
+      "loss": 2.5722,
+      "step": 1121
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.882610633990337e-06,
+      "loss": 2.4202,
+      "step": 1122
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.882234362730255e-06,
+      "loss": 2.3941,
+      "step": 1123
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.881857496584726e-06,
+      "loss": 2.4993,
+      "step": 1124
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.881480035599667e-06,
+      "loss": 2.4692,
+      "step": 1125
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.881101979821075e-06,
+      "loss": 2.4985,
+      "step": 1126
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.880723329295012e-06,
+      "loss": 2.492,
+      "step": 1127
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.880344084067616e-06,
+      "loss": 2.4972,
+      "step": 1128
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.879964244185098e-06,
+      "loss": 2.489,
+      "step": 1129
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.879583809693737e-06,
+      "loss": 2.502,
+      "step": 1130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.879202780639893e-06,
+      "loss": 2.4111,
+      "step": 1131
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.878821157069988e-06,
+      "loss": 2.437,
+      "step": 1132
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.878438939030526e-06,
+      "loss": 2.4884,
+      "step": 1133
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.878056126568077e-06,
+      "loss": 2.428,
+      "step": 1134
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.877672719729283e-06,
+      "loss": 2.4556,
+      "step": 1135
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.877288718560866e-06,
+      "loss": 2.4862,
+      "step": 1136
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.876904123109613e-06,
+      "loss": 2.4885,
+      "step": 1137
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.876518933422385e-06,
+      "loss": 2.4985,
+      "step": 1138
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.876133149546117e-06,
+      "loss": 2.4929,
+      "step": 1139
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.875746771527817e-06,
+      "loss": 2.4229,
+      "step": 1140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.87535979941456e-06,
+      "loss": 2.4761,
+      "step": 1141
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.874972233253503e-06,
+      "loss": 2.4624,
+      "step": 1142
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.874584073091867e-06,
+      "loss": 2.5101,
+      "step": 1143
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.874195318976945e-06,
+      "loss": 2.4693,
+      "step": 1144
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.87380597095611e-06,
+      "loss": 2.4254,
+      "step": 1145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.873416029076801e-06,
+      "loss": 2.456,
+      "step": 1146
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.873025493386531e-06,
+      "loss": 2.3734,
+      "step": 1147
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.872634363932887e-06,
+      "loss": 2.5086,
+      "step": 1148
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.872242640763525e-06,
+      "loss": 2.4989,
+      "step": 1149
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.871850323926178e-06,
+      "loss": 2.4635,
+      "step": 1150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.871457413468645e-06,
+      "loss": 2.4457,
+      "step": 1151
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.871063909438803e-06,
+      "loss": 2.4934,
+      "step": 1152
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.8706698118846e-06,
+      "loss": 2.4577,
+      "step": 1153
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.870275120854055e-06,
+      "loss": 2.4449,
+      "step": 1154
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.86987983639526e-06,
+      "loss": 2.4107,
+      "step": 1155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.869483958556376e-06,
+      "loss": 2.5338,
+      "step": 1156
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.869087487385644e-06,
+      "loss": 2.4706,
+      "step": 1157
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.868690422931372e-06,
+      "loss": 2.4764,
+      "step": 1158
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.86829276524194e-06,
+      "loss": 2.4526,
+      "step": 1159
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.867894514365802e-06,
+      "loss": 2.4437,
+      "step": 1160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.867495670351483e-06,
+      "loss": 2.4787,
+      "step": 1161
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.867096233247581e-06,
+      "loss": 2.4917,
+      "step": 1162
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.866696203102765e-06,
+      "loss": 2.4755,
+      "step": 1163
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.866295579965782e-06,
+      "loss": 2.3669,
+      "step": 1164
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.865894363885442e-06,
+      "loss": 2.5013,
+      "step": 1165
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.865492554910634e-06,
+      "loss": 2.5075,
+      "step": 1166
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.865090153090315e-06,
+      "loss": 2.4579,
+      "step": 1167
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.86468715847352e-06,
+      "loss": 2.352,
+      "step": 1168
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.864283571109352e-06,
+      "loss": 2.4661,
+      "step": 1169
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.863879391046985e-06,
+      "loss": 2.4381,
+      "step": 1170
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.863474618335666e-06,
+      "loss": 2.3679,
+      "step": 1171
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.863069253024719e-06,
+      "loss": 2.4201,
+      "step": 1172
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.862663295163533e-06,
+      "loss": 2.4328,
+      "step": 1173
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.862256744801576e-06,
+      "loss": 2.485,
+      "step": 1174
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.861849601988384e-06,
+      "loss": 2.4746,
+      "step": 1175
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.861441866773564e-06,
+      "loss": 2.4763,
+      "step": 1176
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.8610335392068e-06,
+      "loss": 2.4623,
+      "step": 1177
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.860624619337844e-06,
+      "loss": 2.4285,
+      "step": 1178
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.860215107216523e-06,
+      "loss": 2.4692,
+      "step": 1179
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.859805002892733e-06,
+      "loss": 2.4237,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.859394306416443e-06,
+      "loss": 2.4506,
+      "step": 1181
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.8589830178377e-06,
+      "loss": 2.4327,
+      "step": 1182
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.858571137206611e-06,
+      "loss": 2.3846,
+      "step": 1183
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.85815866457337e-06,
+      "loss": 2.47,
+      "step": 1184
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.857745599988231e-06,
+      "loss": 2.4258,
+      "step": 1185
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.857331943501527e-06,
+      "loss": 2.3954,
+      "step": 1186
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.856917695163659e-06,
+      "loss": 2.4905,
+      "step": 1187
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.856502855025101e-06,
+      "loss": 2.4138,
+      "step": 1188
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.856087423136403e-06,
+      "loss": 2.4585,
+      "step": 1189
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.85567139954818e-06,
+      "loss": 2.4263,
+      "step": 1190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.85525478431113e-06,
+      "loss": 2.4332,
+      "step": 1191
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.854837577476008e-06,
+      "loss": 2.4929,
+      "step": 1192
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.854419779093656e-06,
+      "loss": 2.4805,
+      "step": 1193
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.854001389214979e-06,
+      "loss": 2.4132,
+      "step": 1194
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.853582407890954e-06,
+      "loss": 2.4217,
+      "step": 1195
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.853162835172638e-06,
+      "loss": 2.3681,
+      "step": 1196
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.852742671111151e-06,
+      "loss": 2.4736,
+      "step": 1197
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.852321915757688e-06,
+      "loss": 2.4375,
+      "step": 1198
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.85190056916352e-06,
+      "loss": 2.5094,
+      "step": 1199
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.851478631379982e-06,
+      "loss": 2.4479,
+      "step": 1200
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.851056102458492e-06,
+      "loss": 2.4882,
+      "step": 1201
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.85063298245053e-06,
+      "loss": 2.3989,
+      "step": 1202
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.850209271407653e-06,
+      "loss": 2.4631,
+      "step": 1203
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.849784969381488e-06,
+      "loss": 2.4702,
+      "step": 1204
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.849360076423736e-06,
+      "loss": 2.4087,
+      "step": 1205
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.848934592586165e-06,
+      "loss": 2.4926,
+      "step": 1206
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.848508517920626e-06,
+      "loss": 2.3849,
+      "step": 1207
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.84808185247903e-06,
+      "loss": 2.5261,
+      "step": 1208
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.847654596313368e-06,
+      "loss": 2.5043,
+      "step": 1209
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.847226749475696e-06,
+      "loss": 2.5002,
+      "step": 1210
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.846798312018147e-06,
+      "loss": 2.4505,
+      "step": 1211
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.846369283992927e-06,
+      "loss": 2.4221,
+      "step": 1212
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.845939665452309e-06,
+      "loss": 2.4933,
+      "step": 1213
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.845509456448642e-06,
+      "loss": 2.4668,
+      "step": 1214
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.845078657034348e-06,
+      "loss": 2.3861,
+      "step": 1215
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.844647267261915e-06,
+      "loss": 2.5391,
+      "step": 1216
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.84421528718391e-06,
+      "loss": 2.4951,
+      "step": 1217
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.843782716852963e-06,
+      "loss": 2.4783,
+      "step": 1218
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.843349556321787e-06,
+      "loss": 2.4201,
+      "step": 1219
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.842915805643156e-06,
+      "loss": 2.4859,
+      "step": 1220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.842481464869926e-06,
+      "loss": 2.4492,
+      "step": 1221
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.84204653405502e-06,
+      "loss": 2.4495,
+      "step": 1222
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.841611013251428e-06,
+      "loss": 2.4941,
+      "step": 1223
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.841174902512223e-06,
+      "loss": 2.4113,
+      "step": 1224
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.84073820189054e-06,
+      "loss": 2.4982,
+      "step": 1225
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.84030091143959e-06,
+      "loss": 2.5057,
+      "step": 1226
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.839863031212657e-06,
+      "loss": 2.4132,
+      "step": 1227
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.839424561263094e-06,
+      "loss": 2.4216,
+      "step": 1228
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.838985501644329e-06,
+      "loss": 2.4981,
+      "step": 1229
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.838545852409857e-06,
+      "loss": 2.405,
+      "step": 1230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.83810561361325e-06,
+      "loss": 2.5445,
+      "step": 1231
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.83766478530815e-06,
+      "loss": 2.4352,
+      "step": 1232
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.837223367548271e-06,
+      "loss": 2.4048,
+      "step": 1233
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.836781360387396e-06,
+      "loss": 2.423,
+      "step": 1234
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.836338763879386e-06,
+      "loss": 2.4872,
+      "step": 1235
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.835895578078165e-06,
+      "loss": 2.5009,
+      "step": 1236
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.835451803037738e-06,
+      "loss": 2.4383,
+      "step": 1237
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.835007438812177e-06,
+      "loss": 2.4893,
+      "step": 1238
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.834562485455622e-06,
+      "loss": 2.4498,
+      "step": 1239
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.834116943022299e-06,
+      "loss": 2.4699,
+      "step": 1240
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.833670811566485e-06,
+      "loss": 2.3854,
+      "step": 1241
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.833224091142548e-06,
+      "loss": 2.4565,
+      "step": 1242
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.832776781804913e-06,
+      "loss": 2.4473,
+      "step": 1243
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.832328883608088e-06,
+      "loss": 2.4407,
+      "step": 1244
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.831880396606649e-06,
+      "loss": 2.4216,
+      "step": 1245
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.831431320855235e-06,
+      "loss": 2.4294,
+      "step": 1246
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.830981656408575e-06,
+      "loss": 2.4202,
+      "step": 1247
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.830531403321451e-06,
+      "loss": 2.4778,
+      "step": 1248
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.83008056164873e-06,
+      "loss": 2.4299,
+      "step": 1249
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.829629131445342e-06,
+      "loss": 2.4616,
+      "step": 1250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.829177112766295e-06,
+      "loss": 2.517,
+      "step": 1251
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.828724505666664e-06,
+      "loss": 2.5293,
+      "step": 1252
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.828271310201601e-06,
+      "loss": 2.4994,
+      "step": 1253
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.827817526426324e-06,
+      "loss": 2.4691,
+      "step": 1254
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.827363154396126e-06,
+      "loss": 2.492,
+      "step": 1255
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.82690819416637e-06,
+      "loss": 2.4103,
+      "step": 1256
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.826452645792493e-06,
+      "loss": 2.451,
+      "step": 1257
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.825996509330001e-06,
+      "loss": 2.4292,
+      "step": 1258
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.825539784834472e-06,
+      "loss": 2.5445,
+      "step": 1259
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.825082472361558e-06,
+      "loss": 2.4738,
+      "step": 1260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.824624571966982e-06,
+      "loss": 2.4608,
+      "step": 1261
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.824166083706534e-06,
+      "loss": 2.4333,
+      "step": 1262
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.823707007636085e-06,
+      "loss": 2.4042,
+      "step": 1263
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.823247343811567e-06,
+      "loss": 2.3932,
+      "step": 1264
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.822787092288991e-06,
+      "loss": 2.4415,
+      "step": 1265
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.822326253124436e-06,
+      "loss": 2.3972,
+      "step": 1266
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.821864826374057e-06,
+      "loss": 2.4971,
+      "step": 1267
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.821402812094074e-06,
+      "loss": 2.3812,
+      "step": 1268
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.820940210340784e-06,
+      "loss": 2.5255,
+      "step": 1269
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.82047702117055e-06,
+      "loss": 2.3465,
+      "step": 1270
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.820013244639817e-06,
+      "loss": 2.4133,
+      "step": 1271
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.819548880805087e-06,
+      "loss": 2.467,
+      "step": 1272
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.819083929722947e-06,
+      "loss": 2.3595,
+      "step": 1273
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.81861839145005e-06,
+      "loss": 2.4171,
+      "step": 1274
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.818152266043115e-06,
+      "loss": 2.4692,
+      "step": 1275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.817685553558945e-06,
+      "loss": 2.3832,
+      "step": 1276
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.8172182540544e-06,
+      "loss": 2.4008,
+      "step": 1277
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.816750367586424e-06,
+      "loss": 2.3742,
+      "step": 1278
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.816281894212028e-06,
+      "loss": 2.544,
+      "step": 1279
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.815812833988292e-06,
+      "loss": 2.4484,
+      "step": 1280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.815343186972369e-06,
+      "loss": 2.4733,
+      "step": 1281
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.814872953221487e-06,
+      "loss": 2.4152,
+      "step": 1282
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.814402132792939e-06,
+      "loss": 2.4105,
+      "step": 1283
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.813930725744095e-06,
+      "loss": 2.4007,
+      "step": 1284
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.813458732132395e-06,
+      "loss": 2.4782,
+      "step": 1285
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.812986152015349e-06,
+      "loss": 2.4505,
+      "step": 1286
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.812512985450539e-06,
+      "loss": 2.4459,
+      "step": 1287
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.81203923249562e-06,
+      "loss": 2.3959,
+      "step": 1288
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.811564893208317e-06,
+      "loss": 2.4742,
+      "step": 1289
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.811089967646427e-06,
+      "loss": 2.3631,
+      "step": 1290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.810614455867818e-06,
+      "loss": 2.458,
+      "step": 1291
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.81013835793043e-06,
+      "loss": 2.4132,
+      "step": 1292
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.809661673892274e-06,
+      "loss": 2.4495,
+      "step": 1293
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.809184403811432e-06,
+      "loss": 2.4011,
+      "step": 1294
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.808706547746057e-06,
+      "loss": 2.4624,
+      "step": 1295
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.808228105754378e-06,
+      "loss": 2.5053,
+      "step": 1296
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.807749077894686e-06,
+      "loss": 2.4173,
+      "step": 1297
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.807269464225355e-06,
+      "loss": 2.4974,
+      "step": 1298
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.806789264804821e-06,
+      "loss": 2.4274,
+      "step": 1299
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.806308479691595e-06,
+      "loss": 2.4445,
+      "step": 1300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.80582710894426e-06,
+      "loss": 2.4217,
+      "step": 1301
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.80534515262147e-06,
+      "loss": 2.4244,
+      "step": 1302
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.804862610781949e-06,
+      "loss": 2.4089,
+      "step": 1303
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.804379483484493e-06,
+      "loss": 2.4354,
+      "step": 1304
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.803895770787972e-06,
+      "loss": 2.4464,
+      "step": 1305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.803411472751321e-06,
+      "loss": 2.3986,
+      "step": 1306
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.802926589433553e-06,
+      "loss": 2.4104,
+      "step": 1307
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.80244112089375e-06,
+      "loss": 2.4411,
+      "step": 1308
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.801955067191062e-06,
+      "loss": 2.4531,
+      "step": 1309
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.801468428384716e-06,
+      "loss": 2.4369,
+      "step": 1310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.800981204534006e-06,
+      "loss": 2.4551,
+      "step": 1311
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.8004933956983e-06,
+      "loss": 2.5355,
+      "step": 1312
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.800005001937034e-06,
+      "loss": 2.4434,
+      "step": 1313
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.799516023309719e-06,
+      "loss": 2.4704,
+      "step": 1314
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.799026459875935e-06,
+      "loss": 2.4722,
+      "step": 1315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.798536311695334e-06,
+      "loss": 2.4848,
+      "step": 1316
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.79804557882764e-06,
+      "loss": 2.4078,
+      "step": 1317
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.797554261332644e-06,
+      "loss": 2.4218,
+      "step": 1318
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.797062359270215e-06,
+      "loss": 2.385,
+      "step": 1319
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.796569872700287e-06,
+      "loss": 2.3863,
+      "step": 1320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.796076801682873e-06,
+      "loss": 2.473,
+      "step": 1321
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.795583146278047e-06,
+      "loss": 2.3674,
+      "step": 1322
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.795088906545959e-06,
+      "loss": 2.4401,
+      "step": 1323
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.794594082546835e-06,
+      "loss": 2.4573,
+      "step": 1324
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.794098674340966e-06,
+      "loss": 2.3879,
+      "step": 1325
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.793602681988714e-06,
+      "loss": 2.4368,
+      "step": 1326
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.793106105550518e-06,
+      "loss": 2.4143,
+      "step": 1327
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.79260894508688e-06,
+      "loss": 2.5011,
+      "step": 1328
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.79211120065838e-06,
+      "loss": 2.4995,
+      "step": 1329
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.791612872325667e-06,
+      "loss": 2.4798,
+      "step": 1330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.791113960149458e-06,
+      "loss": 2.3745,
+      "step": 1331
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.79061446419055e-06,
+      "loss": 2.4028,
+      "step": 1332
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.790114384509796e-06,
+      "loss": 2.4668,
+      "step": 1333
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.789613721168138e-06,
+      "loss": 2.3998,
+      "step": 1334
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.789112474226575e-06,
+      "loss": 2.5133,
+      "step": 1335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.788610643746184e-06,
+      "loss": 2.409,
+      "step": 1336
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.788108229788111e-06,
+      "loss": 2.448,
+      "step": 1337
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.787605232413575e-06,
+      "loss": 2.3807,
+      "step": 1338
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.787101651683864e-06,
+      "loss": 2.5208,
+      "step": 1339
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.786597487660336e-06,
+      "loss": 2.4373,
+      "step": 1340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.786092740404424e-06,
+      "loss": 2.4349,
+      "step": 1341
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.785587409977632e-06,
+      "loss": 2.4775,
+      "step": 1342
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.785081496441528e-06,
+      "loss": 2.3889,
+      "step": 1343
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.784574999857757e-06,
+      "loss": 2.396,
+      "step": 1344
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.78406792028804e-06,
+      "loss": 2.4476,
+      "step": 1345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.783560257794153e-06,
+      "loss": 2.484,
+      "step": 1346
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.783052012437962e-06,
+      "loss": 2.5627,
+      "step": 1347
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.78254318428139e-06,
+      "loss": 2.3795,
+      "step": 1348
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.782033773386439e-06,
+      "loss": 2.4221,
+      "step": 1349
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.781523779815178e-06,
+      "loss": 2.4893,
+      "step": 1350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.781013203629748e-06,
+      "loss": 2.4214,
+      "step": 1351
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.780502044892363e-06,
+      "loss": 2.4112,
+      "step": 1352
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.779990303665303e-06,
+      "loss": 2.4269,
+      "step": 1353
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.779477980010924e-06,
+      "loss": 2.4368,
+      "step": 1354
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.778965073991652e-06,
+      "loss": 2.4846,
+      "step": 1355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.778451585669982e-06,
+      "loss": 2.3845,
+      "step": 1356
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.777937515108478e-06,
+      "loss": 2.3947,
+      "step": 1357
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.777422862369782e-06,
+      "loss": 2.4626,
+      "step": 1358
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.776907627516604e-06,
+      "loss": 2.5333,
+      "step": 1359
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.776391810611719e-06,
+      "loss": 2.4309,
+      "step": 1360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.775875411717981e-06,
+      "loss": 2.4366,
+      "step": 1361
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.775358430898311e-06,
+      "loss": 2.5014,
+      "step": 1362
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.7748408682157e-06,
+      "loss": 2.5078,
+      "step": 1363
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.774322723733216e-06,
+      "loss": 2.4704,
+      "step": 1364
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.77380399751399e-06,
+      "loss": 2.4706,
+      "step": 1365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.773284689621223e-06,
+      "loss": 2.4213,
+      "step": 1366
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.7727648001182e-06,
+      "loss": 2.4248,
+      "step": 1367
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.772244329068261e-06,
+      "loss": 2.4715,
+      "step": 1368
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.771723276534825e-06,
+      "loss": 2.4322,
+      "step": 1369
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.771201642581384e-06,
+      "loss": 2.4272,
+      "step": 1370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.770679427271496e-06,
+      "loss": 2.4757,
+      "step": 1371
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.77015663066879e-06,
+      "loss": 2.5005,
+      "step": 1372
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.769633252836969e-06,
+      "loss": 2.4525,
+      "step": 1373
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.769109293839803e-06,
+      "loss": 2.4267,
+      "step": 1374
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.768584753741134e-06,
+      "loss": 2.4343,
+      "step": 1375
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.768059632604881e-06,
+      "loss": 2.4352,
+      "step": 1376
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.767533930495023e-06,
+      "loss": 2.3775,
+      "step": 1377
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.767007647475618e-06,
+      "loss": 2.47,
+      "step": 1378
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.766480783610789e-06,
+      "loss": 2.4113,
+      "step": 1379
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.765953338964736e-06,
+      "loss": 2.4366,
+      "step": 1380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.765425313601726e-06,
+      "loss": 2.3463,
+      "step": 1381
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.764896707586095e-06,
+      "loss": 2.4181,
+      "step": 1382
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.764367520982255e-06,
+      "loss": 2.3602,
+      "step": 1383
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.763837753854684e-06,
+      "loss": 2.3904,
+      "step": 1384
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.763307406267933e-06,
+      "loss": 2.4067,
+      "step": 1385
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.762776478286622e-06,
+      "loss": 2.4143,
+      "step": 1386
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.762244969975446e-06,
+      "loss": 2.4358,
+      "step": 1387
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.761712881399164e-06,
+      "loss": 2.5048,
+      "step": 1388
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.761180212622613e-06,
+      "loss": 2.5022,
+      "step": 1389
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.760646963710694e-06,
+      "loss": 2.4031,
+      "step": 1390
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.760113134728383e-06,
+      "loss": 2.3606,
+      "step": 1391
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.759578725740726e-06,
+      "loss": 2.4424,
+      "step": 1392
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.75904373681284e-06,
+      "loss": 2.4185,
+      "step": 1393
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.758508168009908e-06,
+      "loss": 2.4684,
+      "step": 1394
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.757972019397192e-06,
+      "loss": 2.4405,
+      "step": 1395
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.757435291040016e-06,
+      "loss": 2.3608,
+      "step": 1396
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.756897983003782e-06,
+      "loss": 2.4881,
+      "step": 1397
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.756360095353957e-06,
+      "loss": 2.4903,
+      "step": 1398
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.755821628156083e-06,
+      "loss": 2.3571,
+      "step": 1399
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.755282581475769e-06,
+      "loss": 2.4273,
+      "step": 1400
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.754742955378697e-06,
+      "loss": 2.4829,
+      "step": 1401
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.754202749930618e-06,
+      "loss": 2.4188,
+      "step": 1402
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.753661965197355e-06,
+      "loss": 2.4487,
+      "step": 1403
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.7531206012448e-06,
+      "loss": 2.4779,
+      "step": 1404
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.75257865813892e-06,
+      "loss": 2.3492,
+      "step": 1405
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.752036135945743e-06,
+      "loss": 2.4777,
+      "step": 1406
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.75149303473138e-06,
+      "loss": 2.3301,
+      "step": 1407
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.750949354562006e-06,
+      "loss": 2.4457,
+      "step": 1408
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.750405095503859e-06,
+      "loss": 2.4701,
+      "step": 1409
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.749860257623262e-06,
+      "loss": 2.4751,
+      "step": 1410
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.749314840986604e-06,
+      "loss": 2.4202,
+      "step": 1411
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.748768845660335e-06,
+      "loss": 2.5285,
+      "step": 1412
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.748222271710988e-06,
+      "loss": 2.4421,
+      "step": 1413
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.74767511920516e-06,
+      "loss": 2.4437,
+      "step": 1414
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.74712738820952e-06,
+      "loss": 2.3568,
+      "step": 1415
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.746579078790808e-06,
+      "loss": 2.3604,
+      "step": 1416
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.746030191015831e-06,
+      "loss": 2.3484,
+      "step": 1417
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.745480724951473e-06,
+      "loss": 2.3764,
+      "step": 1418
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.744930680664685e-06,
+      "loss": 2.3722,
+      "step": 1419
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.744380058222483e-06,
+      "loss": 2.4844,
+      "step": 1420
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.743828857691964e-06,
+      "loss": 2.4175,
+      "step": 1421
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.743277079140288e-06,
+      "loss": 2.4129,
+      "step": 1422
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.742724722634688e-06,
+      "loss": 2.4284,
+      "step": 1423
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.742171788242468e-06,
+      "loss": 2.4331,
+      "step": 1424
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.741618276030998e-06,
+      "loss": 2.4219,
+      "step": 1425
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.741064186067723e-06,
+      "loss": 2.3602,
+      "step": 1426
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.74050951842016e-06,
+      "loss": 2.4359,
+      "step": 1427
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.739954273155892e-06,
+      "loss": 2.4299,
+      "step": 1428
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.739398450342573e-06,
+      "loss": 2.4866,
+      "step": 1429
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.73884205004793e-06,
+      "loss": 2.4114,
+      "step": 1430
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.738285072339756e-06,
+      "loss": 2.4362,
+      "step": 1431
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.73772751728592e-06,
+      "loss": 2.4306,
+      "step": 1432
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.737169384954356e-06,
+      "loss": 2.4225,
+      "step": 1433
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.736610675413073e-06,
+      "loss": 2.4139,
+      "step": 1434
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.736051388730146e-06,
+      "loss": 2.3212,
+      "step": 1435
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.735491524973723e-06,
+      "loss": 2.3839,
+      "step": 1436
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.734931084212021e-06,
+      "loss": 2.4427,
+      "step": 1437
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.73437006651333e-06,
+      "loss": 2.3835,
+      "step": 1438
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.73380847194601e-06,
+      "loss": 2.4475,
+      "step": 1439
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.733246300578482e-06,
+      "loss": 2.3744,
+      "step": 1440
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.732683552479253e-06,
+      "loss": 2.4163,
+      "step": 1441
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.732120227716887e-06,
+      "loss": 2.4526,
+      "step": 1442
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.731556326360027e-06,
+      "loss": 2.4835,
+      "step": 1443
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.73099184847738e-06,
+      "loss": 2.4732,
+      "step": 1444
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.730426794137727e-06,
+      "loss": 2.4059,
+      "step": 1445
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.72986116340992e-06,
+      "loss": 2.4855,
+      "step": 1446
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.729294956362873e-06,
+      "loss": 2.3942,
+      "step": 1447
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.728728173065584e-06,
+      "loss": 2.4319,
+      "step": 1448
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.728160813587111e-06,
+      "loss": 2.4277,
+      "step": 1449
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.727592877996585e-06,
+      "loss": 2.3971,
+      "step": 1450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.727024366363208e-06,
+      "loss": 2.4322,
+      "step": 1451
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.726455278756249e-06,
+      "loss": 2.4048,
+      "step": 1452
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.72588561524505e-06,
+      "loss": 2.4402,
+      "step": 1453
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.725315375899025e-06,
+      "loss": 2.385,
+      "step": 1454
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.724744560787655e-06,
+      "loss": 2.3625,
+      "step": 1455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.724173169980492e-06,
+      "loss": 2.426,
+      "step": 1456
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.723601203547158e-06,
+      "loss": 2.495,
+      "step": 1457
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.723028661557345e-06,
+      "loss": 2.3845,
+      "step": 1458
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.722455544080818e-06,
+      "loss": 2.3654,
+      "step": 1459
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.721881851187406e-06,
+      "loss": 2.476,
+      "step": 1460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.721307582947014e-06,
+      "loss": 2.4328,
+      "step": 1461
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.720732739429614e-06,
+      "loss": 2.4427,
+      "step": 1462
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.72015732070525e-06,
+      "loss": 2.4237,
+      "step": 1463
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.719581326844033e-06,
+      "loss": 2.4195,
+      "step": 1464
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.719004757916149e-06,
+      "loss": 2.4395,
+      "step": 1465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.718427613991848e-06,
+      "loss": 2.4905,
+      "step": 1466
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.717849895141455e-06,
+      "loss": 2.4569,
+      "step": 1467
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.717271601435363e-06,
+      "loss": 2.4063,
+      "step": 1468
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.716692732944036e-06,
+      "loss": 2.5053,
+      "step": 1469
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.716113289738005e-06,
+      "loss": 2.5018,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.715533271887877e-06,
+      "loss": 2.467,
+      "step": 1471
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.714952679464324e-06,
+      "loss": 2.482,
+      "step": 1472
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.714371512538088e-06,
+      "loss": 2.5207,
+      "step": 1473
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.713789771179983e-06,
+      "loss": 2.4406,
+      "step": 1474
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.713207455460893e-06,
+      "loss": 2.3858,
+      "step": 1475
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.712624565451772e-06,
+      "loss": 2.3897,
+      "step": 1476
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.71204110122364e-06,
+      "loss": 2.4721,
+      "step": 1477
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.711457062847596e-06,
+      "loss": 2.4676,
+      "step": 1478
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.7108724503948e-06,
+      "loss": 2.3315,
+      "step": 1479
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.710287263936485e-06,
+      "loss": 2.3404,
+      "step": 1480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.709701503543954e-06,
+      "loss": 2.3993,
+      "step": 1481
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.709115169288582e-06,
+      "loss": 2.319,
+      "step": 1482
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.70852826124181e-06,
+      "loss": 2.43,
+      "step": 1483
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.707940779475151e-06,
+      "loss": 2.3089,
+      "step": 1484
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.70735272406019e-06,
+      "loss": 2.4254,
+      "step": 1485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.706764095068579e-06,
+      "loss": 2.4871,
+      "step": 1486
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.706174892572038e-06,
+      "loss": 2.3975,
+      "step": 1487
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.705585116642364e-06,
+      "loss": 2.4653,
+      "step": 1488
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.704994767351417e-06,
+      "loss": 2.4269,
+      "step": 1489
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.704403844771128e-06,
+      "loss": 2.4449,
+      "step": 1490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.703812348973501e-06,
+      "loss": 2.4293,
+      "step": 1491
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.703220280030607e-06,
+      "loss": 2.4266,
+      "step": 1492
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.70262763801459e-06,
+      "loss": 2.4153,
+      "step": 1493
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.702034422997658e-06,
+      "loss": 2.359,
+      "step": 1494
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.701440635052094e-06,
+      "loss": 2.417,
+      "step": 1495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.700846274250252e-06,
+      "loss": 2.4287,
+      "step": 1496
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.70025134066455e-06,
+      "loss": 2.362,
+      "step": 1497
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.699655834367479e-06,
+      "loss": 2.4776,
+      "step": 1498
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.699059755431599e-06,
+      "loss": 2.4327,
+      "step": 1499
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.698463103929542e-06,
+      "loss": 2.3676,
+      "step": 1500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.69786587993401e-06,
+      "loss": 2.476,
+      "step": 1501
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.697268083517767e-06,
+      "loss": 2.4289,
+      "step": 1502
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.696669714753658e-06,
+      "loss": 2.4629,
+      "step": 1503
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.696070773714592e-06,
+      "loss": 2.4003,
+      "step": 1504
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.695471260473546e-06,
+      "loss": 2.3979,
+      "step": 1505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.69487117510357e-06,
+      "loss": 2.4287,
+      "step": 1506
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.694270517677782e-06,
+      "loss": 2.3366,
+      "step": 1507
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.693669288269371e-06,
+      "loss": 2.3733,
+      "step": 1508
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.693067486951595e-06,
+      "loss": 2.4773,
+      "step": 1509
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.69246511379778e-06,
+      "loss": 2.4903,
+      "step": 1510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.691862168881325e-06,
+      "loss": 2.4459,
+      "step": 1511
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.691258652275698e-06,
+      "loss": 2.3518,
+      "step": 1512
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.690654564054433e-06,
+      "loss": 2.3875,
+      "step": 1513
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.690049904291139e-06,
+      "loss": 2.4683,
+      "step": 1514
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.68944467305949e-06,
+      "loss": 2.3752,
+      "step": 1515
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.68883887043323e-06,
+      "loss": 2.3862,
+      "step": 1516
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.688232496486179e-06,
+      "loss": 2.4325,
+      "step": 1517
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.687625551292219e-06,
+      "loss": 2.342,
+      "step": 1518
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.687018034925304e-06,
+      "loss": 2.46,
+      "step": 1519
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.68640994745946e-06,
+      "loss": 2.4694,
+      "step": 1520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.685801288968777e-06,
+      "loss": 2.4482,
+      "step": 1521
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.68519205952742e-06,
+      "loss": 2.4694,
+      "step": 1522
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.684582259209625e-06,
+      "loss": 2.3946,
+      "step": 1523
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.68397188808969e-06,
+      "loss": 2.3697,
+      "step": 1524
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.683360946241988e-06,
+      "loss": 2.4085,
+      "step": 1525
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.682749433740963e-06,
+      "loss": 2.4795,
+      "step": 1526
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.682137350661123e-06,
+      "loss": 2.3974,
+      "step": 1527
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.681524697077047e-06,
+      "loss": 2.4211,
+      "step": 1528
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.68091147306339e-06,
+      "loss": 2.4512,
+      "step": 1529
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.680297678694867e-06,
+      "loss": 2.4399,
+      "step": 1530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.67968331404627e-06,
+      "loss": 2.4114,
+      "step": 1531
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.679068379192455e-06,
+      "loss": 2.3532,
+      "step": 1532
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.678452874208352e-06,
+      "loss": 2.3615,
+      "step": 1533
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.677836799168958e-06,
+      "loss": 2.4041,
+      "step": 1534
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.677220154149338e-06,
+      "loss": 2.48,
+      "step": 1535
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.67660293922463e-06,
+      "loss": 2.3872,
+      "step": 1536
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.67598515447004e-06,
+      "loss": 2.4493,
+      "step": 1537
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.675366799960842e-06,
+      "loss": 2.3993,
+      "step": 1538
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.674747875772381e-06,
+      "loss": 2.4145,
+      "step": 1539
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.674128381980073e-06,
+      "loss": 2.4772,
+      "step": 1540
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.673508318659399e-06,
+      "loss": 2.404,
+      "step": 1541
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.672887685885913e-06,
+      "loss": 2.3626,
+      "step": 1542
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.672266483735235e-06,
+      "loss": 2.4528,
+      "step": 1543
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.671644712283061e-06,
+      "loss": 2.4071,
+      "step": 1544
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.671022371605148e-06,
+      "loss": 2.4745,
+      "step": 1545
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.670399461777328e-06,
+      "loss": 2.3945,
+      "step": 1546
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.6697759828755e-06,
+      "loss": 2.3771,
+      "step": 1547
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.669151934975635e-06,
+      "loss": 2.3772,
+      "step": 1548
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.668527318153769e-06,
+      "loss": 2.3804,
+      "step": 1549
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.667902132486009e-06,
+      "loss": 2.4626,
+      "step": 1550
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.667276378048535e-06,
+      "loss": 2.4275,
+      "step": 1551
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.666650054917591e-06,
+      "loss": 2.3445,
+      "step": 1552
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.666023163169493e-06,
+      "loss": 2.3898,
+      "step": 1553
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.665395702880627e-06,
+      "loss": 2.4111,
+      "step": 1554
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.664767674127447e-06,
+      "loss": 2.3183,
+      "step": 1555
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.664139076986473e-06,
+      "loss": 2.4455,
+      "step": 1556
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.663509911534302e-06,
+      "loss": 2.394,
+      "step": 1557
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.662880177847595e-06,
+      "loss": 2.3365,
+      "step": 1558
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.66224987600308e-06,
+      "loss": 2.3831,
+      "step": 1559
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.661619006077562e-06,
+      "loss": 2.3736,
+      "step": 1560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.660987568147907e-06,
+      "loss": 2.4451,
+      "step": 1561
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.660355562291055e-06,
+      "loss": 2.4139,
+      "step": 1562
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.659722988584015e-06,
+      "loss": 2.4728,
+      "step": 1563
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.659089847103863e-06,
+      "loss": 2.518,
+      "step": 1564
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.658456137927745e-06,
+      "loss": 2.3542,
+      "step": 1565
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.65782186113288e-06,
+      "loss": 2.3591,
+      "step": 1566
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.657187016796546e-06,
+      "loss": 2.4363,
+      "step": 1567
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.656551604996102e-06,
+      "loss": 2.4724,
+      "step": 1568
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.655915625808971e-06,
+      "loss": 2.3799,
+      "step": 1569
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.655279079312643e-06,
+      "loss": 2.3806,
+      "step": 1570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.65464196558468e-06,
+      "loss": 2.4391,
+      "step": 1571
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.654004284702712e-06,
+      "loss": 2.4382,
+      "step": 1572
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.65336603674444e-06,
+      "loss": 2.3505,
+      "step": 1573
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.65272722178763e-06,
+      "loss": 2.4737,
+      "step": 1574
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.652087839910123e-06,
+      "loss": 2.4185,
+      "step": 1575
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.651447891189824e-06,
+      "loss": 2.4556,
+      "step": 1576
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.650807375704708e-06,
+      "loss": 2.4015,
+      "step": 1577
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.650166293532822e-06,
+      "loss": 2.3959,
+      "step": 1578
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.649524644752278e-06,
+      "loss": 2.4512,
+      "step": 1579
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.648882429441258e-06,
+      "loss": 2.4137,
+      "step": 1580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.648239647678017e-06,
+      "loss": 2.4003,
+      "step": 1581
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.647596299540874e-06,
+      "loss": 2.3513,
+      "step": 1582
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.64695238510822e-06,
+      "loss": 2.4211,
+      "step": 1583
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.646307904458513e-06,
+      "loss": 2.3474,
+      "step": 1584
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.645662857670281e-06,
+      "loss": 2.3868,
+      "step": 1585
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.645017244822124e-06,
+      "loss": 2.4094,
+      "step": 1586
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.644371065992704e-06,
+      "loss": 2.4055,
+      "step": 1587
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.643724321260757e-06,
+      "loss": 2.4023,
+      "step": 1588
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.643077010705088e-06,
+      "loss": 2.4141,
+      "step": 1589
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.642429134404568e-06,
+      "loss": 2.3983,
+      "step": 1590
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.641780692438142e-06,
+      "loss": 2.4629,
+      "step": 1591
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.641131684884817e-06,
+      "loss": 2.3798,
+      "step": 1592
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.640482111823675e-06,
+      "loss": 2.4018,
+      "step": 1593
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.639831973333864e-06,
+      "loss": 2.4228,
+      "step": 1594
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.6391812694946e-06,
+      "loss": 2.4629,
+      "step": 1595
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.638530000385171e-06,
+      "loss": 2.4489,
+      "step": 1596
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.637878166084932e-06,
+      "loss": 2.3684,
+      "step": 1597
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.637225766673309e-06,
+      "loss": 2.3938,
+      "step": 1598
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.636572802229789e-06,
+      "loss": 2.4545,
+      "step": 1599
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.635919272833938e-06,
+      "loss": 2.4562,
+      "step": 1600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.635265178565386e-06,
+      "loss": 2.3661,
+      "step": 1601
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.634610519503833e-06,
+      "loss": 2.3609,
+      "step": 1602
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.633955295729044e-06,
+      "loss": 2.4158,
+      "step": 1603
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.633299507320862e-06,
+      "loss": 2.3816,
+      "step": 1604
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.632643154359187e-06,
+      "loss": 2.4718,
+      "step": 1605
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.631986236923998e-06,
+      "loss": 2.4749,
+      "step": 1606
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.631328755095334e-06,
+      "loss": 2.4917,
+      "step": 1607
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.630670708953311e-06,
+      "loss": 2.3673,
+      "step": 1608
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.630012098578108e-06,
+      "loss": 2.4743,
+      "step": 1609
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.629352924049975e-06,
+      "loss": 2.3263,
+      "step": 1610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.62869318544923e-06,
+      "loss": 2.376,
+      "step": 1611
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.628032882856262e-06,
+      "loss": 2.3623,
+      "step": 1612
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.627372016351524e-06,
+      "loss": 2.3761,
+      "step": 1613
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.626710586015543e-06,
+      "loss": 2.3602,
+      "step": 1614
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.626048591928911e-06,
+      "loss": 2.4479,
+      "step": 1615
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.62538603417229e-06,
+      "loss": 2.4219,
+      "step": 1616
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.624722912826412e-06,
+      "loss": 2.4271,
+      "step": 1617
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.624059227972077e-06,
+      "loss": 2.4132,
+      "step": 1618
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.623394979690149e-06,
+      "loss": 2.4518,
+      "step": 1619
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.622730168061568e-06,
+      "loss": 2.4321,
+      "step": 1620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.622064793167336e-06,
+      "loss": 2.493,
+      "step": 1621
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.62139885508853e-06,
+      "loss": 2.3702,
+      "step": 1622
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.620732353906293e-06,
+      "loss": 2.3908,
+      "step": 1623
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.620065289701835e-06,
+      "loss": 2.4054,
+      "step": 1624
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.619397662556434e-06,
+      "loss": 2.3585,
+      "step": 1625
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.61872947255144e-06,
+      "loss": 2.3843,
+      "step": 1626
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.618060719768272e-06,
+      "loss": 2.4047,
+      "step": 1627
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.617391404288412e-06,
+      "loss": 2.4179,
+      "step": 1628
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.616721526193416e-06,
+      "loss": 2.4303,
+      "step": 1629
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.616051085564905e-06,
+      "loss": 2.4696,
+      "step": 1630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.615380082484573e-06,
+      "loss": 2.4684,
+      "step": 1631
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.614708517034176e-06,
+      "loss": 2.4373,
+      "step": 1632
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.614036389295545e-06,
+      "loss": 2.4082,
+      "step": 1633
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.613363699350575e-06,
+      "loss": 2.3051,
+      "step": 1634
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.612690447281233e-06,
+      "loss": 2.3886,
+      "step": 1635
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.612016633169552e-06,
+      "loss": 2.4359,
+      "step": 1636
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.611342257097632e-06,
+      "loss": 2.3424,
+      "step": 1637
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.610667319147648e-06,
+      "loss": 2.4176,
+      "step": 1638
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.609991819401836e-06,
+      "loss": 2.3684,
+      "step": 1639
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.609315757942504e-06,
+      "loss": 2.4018,
+      "step": 1640
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.608639134852028e-06,
+      "loss": 2.4151,
+      "step": 1641
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.607961950212855e-06,
+      "loss": 2.3866,
+      "step": 1642
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.607284204107493e-06,
+      "loss": 2.4581,
+      "step": 1643
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.606605896618528e-06,
+      "loss": 2.4755,
+      "step": 1644
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.605927027828608e-06,
+      "loss": 2.4465,
+      "step": 1645
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.605247597820448e-06,
+      "loss": 2.3435,
+      "step": 1646
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.60456760667684e-06,
+      "loss": 2.4643,
+      "step": 1647
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.603887054480636e-06,
+      "loss": 2.3904,
+      "step": 1648
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.60320594131476e-06,
+      "loss": 2.4309,
+      "step": 1649
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.602524267262202e-06,
+      "loss": 2.4093,
+      "step": 1650
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.601842032406024e-06,
+      "loss": 2.4242,
+      "step": 1651
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.601159236829353e-06,
+      "loss": 2.3275,
+      "step": 1652
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.600475880615385e-06,
+      "loss": 2.3109,
+      "step": 1653
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.599791963847388e-06,
+      "loss": 2.4858,
+      "step": 1654
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.59910748660869e-06,
+      "loss": 2.4248,
+      "step": 1655
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.598422448982697e-06,
+      "loss": 2.4494,
+      "step": 1656
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.597736851052874e-06,
+      "loss": 2.4365,
+      "step": 1657
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.597050692902765e-06,
+      "loss": 2.4036,
+      "step": 1658
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.596363974615973e-06,
+      "loss": 2.4502,
+      "step": 1659
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.595676696276173e-06,
+      "loss": 2.3951,
+      "step": 1660
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.594988857967107e-06,
+      "loss": 2.404,
+      "step": 1661
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.594300459772588e-06,
+      "loss": 2.3487,
+      "step": 1662
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.593611501776493e-06,
+      "loss": 2.3392,
+      "step": 1663
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.592921984062771e-06,
+      "loss": 2.4406,
+      "step": 1664
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.592231906715437e-06,
+      "loss": 2.4443,
+      "step": 1665
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.591541269818574e-06,
+      "loss": 2.4627,
+      "step": 1666
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.590850073456335e-06,
+      "loss": 2.3891,
+      "step": 1667
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.590158317712941e-06,
+      "loss": 2.4448,
+      "step": 1668
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.58946600267268e-06,
+      "loss": 2.378,
+      "step": 1669
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.588773128419907e-06,
+      "loss": 2.3825,
+      "step": 1670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.588079695039048e-06,
+      "loss": 2.3594,
+      "step": 1671
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.587385702614593e-06,
+      "loss": 2.4858,
+      "step": 1672
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.586691151231109e-06,
+      "loss": 2.4428,
+      "step": 1673
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.585996040973218e-06,
+      "loss": 2.3869,
+      "step": 1674
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.58530037192562e-06,
+      "loss": 2.393,
+      "step": 1675
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.584604144173084e-06,
+      "loss": 2.4102,
+      "step": 1676
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.583907357800436e-06,
+      "loss": 2.4663,
+      "step": 1677
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.583210012892582e-06,
+      "loss": 2.4728,
+      "step": 1678
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.58251210953449e-06,
+      "loss": 2.4176,
+      "step": 1679
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.581813647811199e-06,
+      "loss": 2.3756,
+      "step": 1680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.581114627807812e-06,
+      "loss": 2.4067,
+      "step": 1681
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.580415049609503e-06,
+      "loss": 2.4231,
+      "step": 1682
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.579714913301516e-06,
+      "loss": 2.4199,
+      "step": 1683
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.579014218969158e-06,
+      "loss": 2.3861,
+      "step": 1684
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.578312966697807e-06,
+      "loss": 2.3878,
+      "step": 1685
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.577611156572908e-06,
+      "loss": 2.4103,
+      "step": 1686
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.576908788679975e-06,
+      "loss": 2.4047,
+      "step": 1687
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.576205863104588e-06,
+      "loss": 2.4258,
+      "step": 1688
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.5755023799324e-06,
+      "loss": 2.407,
+      "step": 1689
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.574798339249124e-06,
+      "loss": 2.4478,
+      "step": 1690
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.574093741140549e-06,
+      "loss": 2.4221,
+      "step": 1691
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.573388585692525e-06,
+      "loss": 2.334,
+      "step": 1692
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.572682872990975e-06,
+      "loss": 2.3409,
+      "step": 1693
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.571976603121889e-06,
+      "loss": 2.4013,
+      "step": 1694
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.571269776171319e-06,
+      "loss": 2.4253,
+      "step": 1695
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.570562392225395e-06,
+      "loss": 2.2717,
+      "step": 1696
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.569854451370306e-06,
+      "loss": 2.322,
+      "step": 1697
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.569145953692316e-06,
+      "loss": 2.4287,
+      "step": 1698
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.56843689927775e-06,
+      "loss": 2.3066,
+      "step": 1699
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.567727288213005e-06,
+      "loss": 2.3494,
+      "step": 1700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.567017120584545e-06,
+      "loss": 2.3864,
+      "step": 1701
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.566306396478904e-06,
+      "loss": 2.4406,
+      "step": 1702
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.565595115982678e-06,
+      "loss": 2.3895,
+      "step": 1703
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.564883279182538e-06,
+      "loss": 2.3766,
+      "step": 1704
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.564170886165214e-06,
+      "loss": 2.3792,
+      "step": 1705
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.563457937017514e-06,
+      "loss": 2.3817,
+      "step": 1706
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.562744431826307e-06,
+      "loss": 2.4642,
+      "step": 1707
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.562030370678533e-06,
+      "loss": 2.42,
+      "step": 1708
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.561315753661194e-06,
+      "loss": 2.4056,
+      "step": 1709
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.560600580861366e-06,
+      "loss": 2.4011,
+      "step": 1710
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.559884852366191e-06,
+      "loss": 2.4786,
+      "step": 1711
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.55916856826288e-06,
+      "loss": 2.3902,
+      "step": 1712
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.558451728638706e-06,
+      "loss": 2.4643,
+      "step": 1713
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.557734333581019e-06,
+      "loss": 2.4237,
+      "step": 1714
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.557016383177226e-06,
+      "loss": 2.4222,
+      "step": 1715
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.556297877514812e-06,
+      "loss": 2.4856,
+      "step": 1716
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.555578816681321e-06,
+      "loss": 2.4245,
+      "step": 1717
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.554859200764371e-06,
+      "loss": 2.3968,
+      "step": 1718
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.55413902985164e-06,
+      "loss": 2.4236,
+      "step": 1719
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.553418304030886e-06,
+      "loss": 2.3597,
+      "step": 1720
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.552697023389923e-06,
+      "loss": 2.3583,
+      "step": 1721
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.551975188016638e-06,
+      "loss": 2.437,
+      "step": 1722
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.551252797998982e-06,
+      "loss": 2.4522,
+      "step": 1723
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.550529853424979e-06,
+      "loss": 2.3563,
+      "step": 1724
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.549806354382716e-06,
+      "loss": 2.4446,
+      "step": 1725
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.549082300960351e-06,
+      "loss": 2.3526,
+      "step": 1726
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.548357693246107e-06,
+      "loss": 2.4046,
+      "step": 1727
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.547632531328273e-06,
+      "loss": 2.3833,
+      "step": 1728
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.546906815295209e-06,
+      "loss": 2.4429,
+      "step": 1729
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.546180545235344e-06,
+      "loss": 2.4693,
+      "step": 1730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.545453721237167e-06,
+      "loss": 2.3382,
+      "step": 1731
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.544726343389245e-06,
+      "loss": 2.3918,
+      "step": 1732
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.543998411780202e-06,
+      "loss": 2.4076,
+      "step": 1733
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.543269926498735e-06,
+      "loss": 2.465,
+      "step": 1734
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.54254088763361e-06,
+      "loss": 2.4559,
+      "step": 1735
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.541811295273657e-06,
+      "loss": 2.4266,
+      "step": 1736
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.541081149507774e-06,
+      "loss": 2.5033,
+      "step": 1737
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.540350450424927e-06,
+      "loss": 2.4575,
+      "step": 1738
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.53961919811415e-06,
+      "loss": 2.3616,
+      "step": 1739
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.538887392664544e-06,
+      "loss": 2.4175,
+      "step": 1740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.538155034165277e-06,
+      "loss": 2.3844,
+      "step": 1741
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.537422122705585e-06,
+      "loss": 2.4728,
+      "step": 1742
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.536688658374771e-06,
+      "loss": 2.3378,
+      "step": 1743
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.535954641262206e-06,
+      "loss": 2.41,
+      "step": 1744
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.535220071457325e-06,
+      "loss": 2.4977,
+      "step": 1745
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.534484949049636e-06,
+      "loss": 2.3244,
+      "step": 1746
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.533749274128712e-06,
+      "loss": 2.2935,
+      "step": 1747
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.53301304678419e-06,
+      "loss": 2.4287,
+      "step": 1748
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.532276267105778e-06,
+      "loss": 2.3783,
+      "step": 1749
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.531538935183252e-06,
+      "loss": 2.3586,
+      "step": 1750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.530801051106449e-06,
+      "loss": 2.4452,
+      "step": 1751
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.530062614965286e-06,
+      "loss": 2.3147,
+      "step": 1752
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.529323626849731e-06,
+      "loss": 2.4005,
+      "step": 1753
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.528584086849832e-06,
+      "loss": 2.3631,
+      "step": 1754
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.527843995055698e-06,
+      "loss": 2.4089,
+      "step": 1755
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.52710335155751e-06,
+      "loss": 2.3732,
+      "step": 1756
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.526362156445508e-06,
+      "loss": 2.3591,
+      "step": 1757
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.525620409810009e-06,
+      "loss": 2.4039,
+      "step": 1758
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.524878111741388e-06,
+      "loss": 2.3812,
+      "step": 1759
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.524135262330098e-06,
+      "loss": 2.3132,
+      "step": 1760
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.523391861666649e-06,
+      "loss": 2.3612,
+      "step": 1761
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.52264790984162e-06,
+      "loss": 2.4096,
+      "step": 1762
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.521903406945666e-06,
+      "loss": 2.4515,
+      "step": 1763
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.521158353069494e-06,
+      "loss": 2.3597,
+      "step": 1764
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.520412748303894e-06,
+      "loss": 2.4011,
+      "step": 1765
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.51966659273971e-06,
+      "loss": 2.341,
+      "step": 1766
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.51891988646786e-06,
+      "loss": 2.3685,
+      "step": 1767
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.518172629579334e-06,
+      "loss": 2.3665,
+      "step": 1768
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.517424822165175e-06,
+      "loss": 2.4443,
+      "step": 1769
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.516676464316505e-06,
+      "loss": 2.4019,
+      "step": 1770
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.515927556124508e-06,
+      "loss": 2.4618,
+      "step": 1771
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.515178097680437e-06,
+      "loss": 2.3649,
+      "step": 1772
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.514428089075611e-06,
+      "loss": 2.3161,
+      "step": 1773
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.513677530401415e-06,
+      "loss": 2.3398,
+      "step": 1774
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.512926421749305e-06,
+      "loss": 2.366,
+      "step": 1775
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.512174763210798e-06,
+      "loss": 2.3735,
+      "step": 1776
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.511422554877482e-06,
+      "loss": 2.3208,
+      "step": 1777
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.510669796841014e-06,
+      "loss": 2.4034,
+      "step": 1778
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.509916489193114e-06,
+      "loss": 2.3741,
+      "step": 1779
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.50916263202557e-06,
+      "loss": 2.3938,
+      "step": 1780
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.508408225430237e-06,
+      "loss": 2.4271,
+      "step": 1781
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.507653269499035e-06,
+      "loss": 2.37,
+      "step": 1782
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.506897764323957e-06,
+      "loss": 2.4048,
+      "step": 1783
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.506141709997058e-06,
+      "loss": 2.3456,
+      "step": 1784
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.505385106610458e-06,
+      "loss": 2.3561,
+      "step": 1785
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.504627954256352e-06,
+      "loss": 2.3698,
+      "step": 1786
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.503870253026992e-06,
+      "loss": 2.3519,
+      "step": 1787
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.503112003014702e-06,
+      "loss": 2.3828,
+      "step": 1788
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.502353204311876e-06,
+      "loss": 2.4082,
+      "step": 1789
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.501593857010968e-06,
+      "loss": 2.443,
+      "step": 1790
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.500833961204504e-06,
+      "loss": 2.3706,
+      "step": 1791
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.500073516985074e-06,
+      "loss": 2.4118,
+      "step": 1792
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.499312524445337e-06,
+      "loss": 2.3698,
+      "step": 1793
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.498550983678016e-06,
+      "loss": 2.3851,
+      "step": 1794
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.497788894775903e-06,
+      "loss": 2.451,
+      "step": 1795
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.497026257831856e-06,
+      "loss": 2.4394,
+      "step": 1796
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.496263072938801e-06,
+      "loss": 2.3321,
+      "step": 1797
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.495499340189729e-06,
+      "loss": 2.3476,
+      "step": 1798
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.4947350596777e-06,
+      "loss": 2.4185,
+      "step": 1799
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.493970231495836e-06,
+      "loss": 2.3574,
+      "step": 1800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.493204855737332e-06,
+      "loss": 2.3267,
+      "step": 1801
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.492438932495444e-06,
+      "loss": 2.4522,
+      "step": 1802
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.491672461863501e-06,
+      "loss": 2.3182,
+      "step": 1803
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.490905443934892e-06,
+      "loss": 2.3503,
+      "step": 1804
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.490137878803078e-06,
+      "loss": 2.3722,
+      "step": 1805
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.489369766561584e-06,
+      "loss": 2.4032,
+      "step": 1806
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.488601107304001e-06,
+      "loss": 2.3595,
+      "step": 1807
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.487831901123989e-06,
+      "loss": 2.4491,
+      "step": 1808
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.487062148115272e-06,
+      "loss": 2.3263,
+      "step": 1809
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.486291848371642e-06,
+      "loss": 2.4665,
+      "step": 1810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.485521001986961e-06,
+      "loss": 2.3752,
+      "step": 1811
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.484749609055151e-06,
+      "loss": 2.3761,
+      "step": 1812
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.483977669670204e-06,
+      "loss": 2.4043,
+      "step": 1813
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.48320518392618e-06,
+      "loss": 2.3781,
+      "step": 1814
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.482432151917204e-06,
+      "loss": 2.467,
+      "step": 1815
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.481658573737465e-06,
+      "loss": 2.3822,
+      "step": 1816
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.480884449481224e-06,
+      "loss": 2.4205,
+      "step": 1817
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.480109779242805e-06,
+      "loss": 2.3541,
+      "step": 1818
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.4793345631166e-06,
+      "loss": 2.3849,
+      "step": 1819
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.478558801197065e-06,
+      "loss": 2.3614,
+      "step": 1820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.477782493578724e-06,
+      "loss": 2.4072,
+      "step": 1821
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.47700564035617e-06,
+      "loss": 2.4446,
+      "step": 1822
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.476228241624059e-06,
+      "loss": 2.368,
+      "step": 1823
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.475450297477113e-06,
+      "loss": 2.4,
+      "step": 1824
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.474671808010126e-06,
+      "loss": 2.4259,
+      "step": 1825
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.473892773317952e-06,
+      "loss": 2.4197,
+      "step": 1826
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.473113193495513e-06,
+      "loss": 2.4196,
+      "step": 1827
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.4723330686378e-06,
+      "loss": 2.4065,
+      "step": 1828
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.47155239883987e-06,
+      "loss": 2.4362,
+      "step": 1829
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.470771184196842e-06,
+      "loss": 2.4983,
+      "step": 1830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.469989424803907e-06,
+      "loss": 2.3879,
+      "step": 1831
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.46920712075632e-06,
+      "loss": 2.4332,
+      "step": 1832
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.468424272149402e-06,
+      "loss": 2.3816,
+      "step": 1833
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.46764087907854e-06,
+      "loss": 2.4563,
+      "step": 1834
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.46685694163919e-06,
+      "loss": 2.4074,
+      "step": 1835
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.46607245992687e-06,
+      "loss": 2.4449,
+      "step": 1836
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.465287434037167e-06,
+      "loss": 2.4379,
+      "step": 1837
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.464501864065735e-06,
+      "loss": 2.372,
+      "step": 1838
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.463715750108293e-06,
+      "loss": 2.4071,
+      "step": 1839
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.46292909226063e-06,
+      "loss": 2.4312,
+      "step": 1840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.46214189061859e-06,
+      "loss": 2.4302,
+      "step": 1841
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.461354145278098e-06,
+      "loss": 2.3294,
+      "step": 1842
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.460565856335136e-06,
+      "loss": 2.3685,
+      "step": 1843
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.459777023885754e-06,
+      "loss": 2.3567,
+      "step": 1844
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.458987648026071e-06,
+      "loss": 2.3703,
+      "step": 1845
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.458197728852268e-06,
+      "loss": 2.3929,
+      "step": 1846
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.457407266460595e-06,
+      "loss": 2.3776,
+      "step": 1847
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.456616260947367e-06,
+      "loss": 2.3699,
+      "step": 1848
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.455824712408967e-06,
+      "loss": 2.4379,
+      "step": 1849
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.45503262094184e-06,
+      "loss": 2.4111,
+      "step": 1850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.454239986642503e-06,
+      "loss": 2.4044,
+      "step": 1851
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.453446809607534e-06,
+      "loss": 2.4198,
+      "step": 1852
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.45265308993358e-06,
+      "loss": 2.3306,
+      "step": 1853
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.451858827717354e-06,
+      "loss": 2.3692,
+      "step": 1854
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.451064023055634e-06,
+      "loss": 2.4613,
+      "step": 1855
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.450268676045261e-06,
+      "loss": 2.3742,
+      "step": 1856
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.449472786783153e-06,
+      "loss": 2.3753,
+      "step": 1857
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.448676355366282e-06,
+      "loss": 2.4056,
+      "step": 1858
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.447879381891691e-06,
+      "loss": 2.4132,
+      "step": 1859
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.44708186645649e-06,
+      "loss": 2.3638,
+      "step": 1860
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.44628380915785e-06,
+      "loss": 2.4096,
+      "step": 1861
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.445485210093018e-06,
+      "loss": 2.349,
+      "step": 1862
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.444686069359294e-06,
+      "loss": 2.388,
+      "step": 1863
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.443886387054058e-06,
+      "loss": 2.4798,
+      "step": 1864
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.443086163274745e-06,
+      "loss": 2.3729,
+      "step": 1865
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.44228539811886e-06,
+      "loss": 2.4606,
+      "step": 1866
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.441484091683975e-06,
+      "loss": 2.5075,
+      "step": 1867
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.440682244067724e-06,
+      "loss": 2.4083,
+      "step": 1868
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.439879855367813e-06,
+      "loss": 2.3529,
+      "step": 1869
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.439076925682006e-06,
+      "loss": 2.4139,
+      "step": 1870
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.438273455108145e-06,
+      "loss": 2.4218,
+      "step": 1871
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.437469443744124e-06,
+      "loss": 2.3888,
+      "step": 1872
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.436664891687911e-06,
+      "loss": 2.2964,
+      "step": 1873
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.435859799037541e-06,
+      "loss": 2.4124,
+      "step": 1874
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.43505416589111e-06,
+      "loss": 2.3224,
+      "step": 1875
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.43424799234678e-06,
+      "loss": 2.3808,
+      "step": 1876
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.433441278502784e-06,
+      "loss": 2.3624,
+      "step": 1877
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.432634024457414e-06,
+      "loss": 2.3127,
+      "step": 1878
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.431826230309035e-06,
+      "loss": 2.3965,
+      "step": 1879
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.431017896156074e-06,
+      "loss": 2.4321,
+      "step": 1880
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.430209022097024e-06,
+      "loss": 2.3588,
+      "step": 1881
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.429399608230441e-06,
+      "loss": 2.4822,
+      "step": 1882
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.428589654654951e-06,
+      "loss": 2.3042,
+      "step": 1883
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.427779161469246e-06,
+      "loss": 2.4033,
+      "step": 1884
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.42696812877208e-06,
+      "loss": 2.3983,
+      "step": 1885
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.426156556662276e-06,
+      "loss": 2.4397,
+      "step": 1886
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.425344445238723e-06,
+      "loss": 2.3751,
+      "step": 1887
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.424531794600372e-06,
+      "loss": 2.4682,
+      "step": 1888
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.423718604846243e-06,
+      "loss": 2.352,
+      "step": 1889
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.42290487607542e-06,
+      "loss": 2.4419,
+      "step": 1890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.422090608387056e-06,
+      "loss": 2.4424,
+      "step": 1891
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.421275801880363e-06,
+      "loss": 2.4116,
+      "step": 1892
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.420460456654625e-06,
+      "loss": 2.408,
+      "step": 1893
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.419644572809189e-06,
+      "loss": 2.4153,
+      "step": 1894
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.418828150443469e-06,
+      "loss": 2.3801,
+      "step": 1895
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.418011189656942e-06,
+      "loss": 2.4053,
+      "step": 1896
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.417193690549151e-06,
+      "loss": 2.3612,
+      "step": 1897
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.41637565321971e-06,
+      "loss": 2.4514,
+      "step": 1898
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.41555707776829e-06,
+      "loss": 2.3958,
+      "step": 1899
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.414737964294636e-06,
+      "loss": 2.3402,
+      "step": 1900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.41391831289855e-06,
+      "loss": 2.4653,
+      "step": 1901
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.41309812367991e-06,
+      "loss": 2.3994,
+      "step": 1902
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.412277396738647e-06,
+      "loss": 2.3571,
+      "step": 1903
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.411456132174768e-06,
+      "loss": 2.4409,
+      "step": 1904
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.41063433008834e-06,
+      "loss": 2.4488,
+      "step": 1905
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.409811990579498e-06,
+      "loss": 2.3774,
+      "step": 1906
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.408989113748442e-06,
+      "loss": 2.3552,
+      "step": 1907
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.408165699695435e-06,
+      "loss": 2.4855,
+      "step": 1908
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.407341748520811e-06,
+      "loss": 2.3821,
+      "step": 1909
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.406517260324962e-06,
+      "loss": 2.3878,
+      "step": 1910
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.405692235208353e-06,
+      "loss": 2.333,
+      "step": 1911
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.404866673271506e-06,
+      "loss": 2.4125,
+      "step": 1912
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.404040574615018e-06,
+      "loss": 2.4028,
+      "step": 1913
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.403213939339546e-06,
+      "loss": 2.3321,
+      "step": 1914
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.40238676754581e-06,
+      "loss": 2.3594,
+      "step": 1915
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.401559059334601e-06,
+      "loss": 2.4233,
+      "step": 1916
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.400730814806774e-06,
+      "loss": 2.4266,
+      "step": 1917
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.399902034063244e-06,
+      "loss": 2.3957,
+      "step": 1918
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.399072717204995e-06,
+      "loss": 2.4106,
+      "step": 1919
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.398242864333084e-06,
+      "loss": 2.3608,
+      "step": 1920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.397412475548619e-06,
+      "loss": 2.3332,
+      "step": 1921
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.396581550952781e-06,
+      "loss": 2.4585,
+      "step": 1922
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.39575009064682e-06,
+      "loss": 2.4104,
+      "step": 1923
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.394918094732044e-06,
+      "loss": 2.4123,
+      "step": 1924
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.394085563309827e-06,
+      "loss": 2.3648,
+      "step": 1925
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.393252496481615e-06,
+      "loss": 2.3589,
+      "step": 1926
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.392418894348912e-06,
+      "loss": 2.3568,
+      "step": 1927
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.39158475701329e-06,
+      "loss": 2.3623,
+      "step": 1928
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.390750084576387e-06,
+      "loss": 2.4935,
+      "step": 1929
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.389914877139903e-06,
+      "loss": 2.4217,
+      "step": 1930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.38907913480561e-06,
+      "loss": 2.3405,
+      "step": 1931
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.388242857675336e-06,
+      "loss": 2.3367,
+      "step": 1932
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.38740604585098e-06,
+      "loss": 2.3726,
+      "step": 1933
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.386568699434509e-06,
+      "loss": 2.448,
+      "step": 1934
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.385730818527945e-06,
+      "loss": 2.3487,
+      "step": 1935
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.384892403233384e-06,
+      "loss": 2.3856,
+      "step": 1936
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.384053453652986e-06,
+      "loss": 2.3798,
+      "step": 1937
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.383213969888972e-06,
+      "loss": 2.3273,
+      "step": 1938
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.382373952043631e-06,
+      "loss": 2.4013,
+      "step": 1939
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.381533400219319e-06,
+      "loss": 2.3501,
+      "step": 1940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.38069231451845e-06,
+      "loss": 2.3963,
+      "step": 1941
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.379850695043513e-06,
+      "loss": 2.4384,
+      "step": 1942
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.379008541897054e-06,
+      "loss": 2.3962,
+      "step": 1943
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.378165855181687e-06,
+      "loss": 2.3623,
+      "step": 1944
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.37732263500009e-06,
+      "loss": 2.3871,
+      "step": 1945
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.376478881455008e-06,
+      "loss": 2.3211,
+      "step": 1946
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.37563459464925e-06,
+      "loss": 2.4522,
+      "step": 1947
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.37478977468569e-06,
+      "loss": 2.3807,
+      "step": 1948
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.373944421667264e-06,
+      "loss": 2.3672,
+      "step": 1949
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.37309853569698e-06,
+      "loss": 2.4377,
+      "step": 1950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.372252116877904e-06,
+      "loss": 2.2912,
+      "step": 1951
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.371405165313169e-06,
+      "loss": 2.4217,
+      "step": 1952
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.370557681105975e-06,
+      "loss": 2.3499,
+      "step": 1953
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.369709664359585e-06,
+      "loss": 2.4087,
+      "step": 1954
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.368861115177327e-06,
+      "loss": 2.4415,
+      "step": 1955
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.368012033662594e-06,
+      "loss": 2.3587,
+      "step": 1956
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.367162419918845e-06,
+      "loss": 2.3768,
+      "step": 1957
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.366312274049602e-06,
+      "loss": 2.3468,
+      "step": 1958
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.365461596158451e-06,
+      "loss": 2.3648,
+      "step": 1959
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.364610386349048e-06,
+      "loss": 2.3297,
+      "step": 1960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.36375864472511e-06,
+      "loss": 2.4271,
+      "step": 1961
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.362906371390416e-06,
+      "loss": 2.4221,
+      "step": 1962
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.362053566448816e-06,
+      "loss": 2.3453,
+      "step": 1963
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.361200230004219e-06,
+      "loss": 2.3761,
+      "step": 1964
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.360346362160604e-06,
+      "loss": 2.3856,
+      "step": 1965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.35949196302201e-06,
+      "loss": 2.4019,
+      "step": 1966
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.358637032692546e-06,
+      "loss": 2.4027,
+      "step": 1967
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.357781571276379e-06,
+      "loss": 2.3868,
+      "step": 1968
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.356925578877748e-06,
+      "loss": 2.3652,
+      "step": 1969
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.356069055600949e-06,
+      "loss": 2.437,
+      "step": 1970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.355212001550349e-06,
+      "loss": 2.3663,
+      "step": 1971
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.354354416830377e-06,
+      "loss": 2.4067,
+      "step": 1972
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.353496301545529e-06,
+      "loss": 2.4385,
+      "step": 1973
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.352637655800362e-06,
+      "loss": 2.4078,
+      "step": 1974
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.351778479699499e-06,
+      "loss": 2.4075,
+      "step": 1975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.35091877334763e-06,
+      "loss": 2.3695,
+      "step": 1976
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.350058536849505e-06,
+      "loss": 2.4245,
+      "step": 1977
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.349197770309942e-06,
+      "loss": 2.3699,
+      "step": 1978
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.348336473833824e-06,
+      "loss": 2.4108,
+      "step": 1979
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.347474647526095e-06,
+      "loss": 2.359,
+      "step": 1980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.34661229149177e-06,
+      "loss": 2.4273,
+      "step": 1981
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.34574940583592e-06,
+      "loss": 2.3903,
+      "step": 1982
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.344885990663689e-06,
+      "loss": 2.3199,
+      "step": 1983
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.344022046080277e-06,
+      "loss": 2.3945,
+      "step": 1984
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.343157572190957e-06,
+      "loss": 2.4699,
+      "step": 1985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.342292569101061e-06,
+      "loss": 2.3718,
+      "step": 1986
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.341427036915987e-06,
+      "loss": 2.3372,
+      "step": 1987
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.340560975741198e-06,
+      "loss": 2.437,
+      "step": 1988
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.339694385682219e-06,
+      "loss": 2.4106,
+      "step": 1989
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.338827266844643e-06,
+      "loss": 2.3881,
+      "step": 1990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.337959619334125e-06,
+      "loss": 2.3351,
+      "step": 1991
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.337091443256388e-06,
+      "loss": 2.4166,
+      "step": 1992
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.336222738717214e-06,
+      "loss": 2.4214,
+      "step": 1993
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.33535350582245e-06,
+      "loss": 2.4338,
+      "step": 1994
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.334483744678015e-06,
+      "loss": 2.3389,
+      "step": 1995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.333613455389883e-06,
+      "loss": 2.3934,
+      "step": 1996
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.332742638064094e-06,
+      "loss": 2.4749,
+      "step": 1997
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.33187129280676e-06,
+      "loss": 2.4325,
+      "step": 1998
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.330999419724048e-06,
+      "loss": 2.3528,
+      "step": 1999
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.330127018922195e-06,
+      "loss": 2.4259,
+      "step": 2000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.329254090507498e-06,
+      "loss": 2.3852,
+      "step": 2001
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.328380634586322e-06,
+      "loss": 2.4558,
+      "step": 2002
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.327506651265096e-06,
+      "loss": 2.3346,
+      "step": 2003
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.326632140650311e-06,
+      "loss": 2.3554,
+      "step": 2004
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.325757102848523e-06,
+      "loss": 2.3178,
+      "step": 2005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.324881537966355e-06,
+      "loss": 2.4264,
+      "step": 2006
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.32400544611049e-06,
+      "loss": 2.3666,
+      "step": 2007
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.323128827387675e-06,
+      "loss": 2.411,
+      "step": 2008
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.322251681904728e-06,
+      "loss": 2.3819,
+      "step": 2009
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.321374009768525e-06,
+      "loss": 2.434,
+      "step": 2010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.320495811086005e-06,
+      "loss": 2.4669,
+      "step": 2011
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.319617085964177e-06,
+      "loss": 2.404,
+      "step": 2012
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.31873783451011e-06,
+      "loss": 2.3969,
+      "step": 2013
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.317858056830938e-06,
+      "loss": 2.3503,
+      "step": 2014
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.316977753033858e-06,
+      "loss": 2.3565,
+      "step": 2015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.316096923226135e-06,
+      "loss": 2.4018,
+      "step": 2016
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.315215567515095e-06,
+      "loss": 2.3074,
+      "step": 2017
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.314333686008125e-06,
+      "loss": 2.448,
+      "step": 2018
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.313451278812684e-06,
+      "loss": 2.4145,
+      "step": 2019
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.312568346036288e-06,
+      "loss": 2.4144,
+      "step": 2020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.31168488778652e-06,
+      "loss": 2.4089,
+      "step": 2021
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.31080090417103e-06,
+      "loss": 2.4539,
+      "step": 2022
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.309916395297523e-06,
+      "loss": 2.4145,
+      "step": 2023
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.309031361273775e-06,
+      "loss": 2.4216,
+      "step": 2024
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.30814580220763e-06,
+      "loss": 2.3948,
+      "step": 2025
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.307259718206984e-06,
+      "loss": 2.273,
+      "step": 2026
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.30637310937981e-06,
+      "loss": 2.3761,
+      "step": 2027
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.305485975834132e-06,
+      "loss": 2.4413,
+      "step": 2028
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.30459831767805e-06,
+      "loss": 2.4336,
+      "step": 2029
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.30371013501972e-06,
+      "loss": 2.3436,
+      "step": 2030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.302821427967363e-06,
+      "loss": 2.3199,
+      "step": 2031
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.301932196629267e-06,
+      "loss": 2.402,
+      "step": 2032
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.301042441113784e-06,
+      "loss": 2.3498,
+      "step": 2033
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.300152161529325e-06,
+      "loss": 2.321,
+      "step": 2034
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.299261357984368e-06,
+      "loss": 2.3417,
+      "step": 2035
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.298370030587456e-06,
+      "loss": 2.3953,
+      "step": 2036
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.297478179447195e-06,
+      "loss": 2.3633,
+      "step": 2037
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.296585804672253e-06,
+      "loss": 2.3591,
+      "step": 2038
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.295692906371362e-06,
+      "loss": 2.3918,
+      "step": 2039
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.294799484653323e-06,
+      "loss": 2.3845,
+      "step": 2040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.293905539626992e-06,
+      "loss": 2.396,
+      "step": 2041
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.293011071401299e-06,
+      "loss": 2.3674,
+      "step": 2042
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.292116080085226e-06,
+      "loss": 2.3825,
+      "step": 2043
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.291220565787829e-06,
+      "loss": 2.3937,
+      "step": 2044
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.290324528618225e-06,
+      "loss": 2.3526,
+      "step": 2045
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.289427968685588e-06,
+      "loss": 2.3655,
+      "step": 2046
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.288530886099165e-06,
+      "loss": 2.4707,
+      "step": 2047
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.287633280968263e-06,
+      "loss": 2.4001,
+      "step": 2048
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.28673515340225e-06,
+      "loss": 2.3227,
+      "step": 2049
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.285836503510562e-06,
+      "loss": 2.368,
+      "step": 2050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.284937331402697e-06,
+      "loss": 2.4083,
+      "step": 2051
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.284037637188215e-06,
+      "loss": 2.3009,
+      "step": 2052
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.283137420976742e-06,
+      "loss": 2.3451,
+      "step": 2053
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.282236682877968e-06,
+      "loss": 2.3034,
+      "step": 2054
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.281335423001641e-06,
+      "loss": 2.4118,
+      "step": 2055
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.280433641457582e-06,
+      "loss": 2.3744,
+      "step": 2056
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.279531338355666e-06,
+      "loss": 2.3701,
+      "step": 2057
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.278628513805838e-06,
+      "loss": 2.3611,
+      "step": 2058
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.277725167918103e-06,
+      "loss": 2.4122,
+      "step": 2059
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.276821300802535e-06,
+      "loss": 2.4344,
+      "step": 2060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.275916912569261e-06,
+      "loss": 2.3419,
+      "step": 2061
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.275012003328483e-06,
+      "loss": 2.4224,
+      "step": 2062
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.27410657319046e-06,
+      "loss": 2.3446,
+      "step": 2063
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.273200622265516e-06,
+      "loss": 2.293,
+      "step": 2064
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.272294150664039e-06,
+      "loss": 2.4149,
+      "step": 2065
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.271387158496477e-06,
+      "loss": 2.4117,
+      "step": 2066
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.270479645873347e-06,
+      "loss": 2.3705,
+      "step": 2067
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.269571612905227e-06,
+      "loss": 2.3657,
+      "step": 2068
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.268663059702753e-06,
+      "loss": 2.4384,
+      "step": 2069
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.267753986376638e-06,
+      "loss": 2.3488,
+      "step": 2070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.266844393037644e-06,
+      "loss": 2.3645,
+      "step": 2071
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.265934279796602e-06,
+      "loss": 2.3504,
+      "step": 2072
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.265023646764409e-06,
+      "loss": 2.3536,
+      "step": 2073
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.264112494052022e-06,
+      "loss": 2.3886,
+      "step": 2074
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.263200821770462e-06,
+      "loss": 2.3783,
+      "step": 2075
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.262288630030814e-06,
+      "loss": 2.3443,
+      "step": 2076
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.261375918944224e-06,
+      "loss": 2.4412,
+      "step": 2077
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.260462688621906e-06,
+      "loss": 2.4043,
+      "step": 2078
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.25954893917513e-06,
+      "loss": 2.4095,
+      "step": 2079
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.25863467071524e-06,
+      "loss": 2.4203,
+      "step": 2080
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.25771988335363e-06,
+      "loss": 2.395,
+      "step": 2081
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.256804577201768e-06,
+      "loss": 2.2913,
+      "step": 2082
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.255888752371182e-06,
+      "loss": 2.3394,
+      "step": 2083
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.25497240897346e-06,
+      "loss": 2.3934,
+      "step": 2084
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.254055547120258e-06,
+      "loss": 2.3226,
+      "step": 2085
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.25313816692329e-06,
+      "loss": 2.3607,
+      "step": 2086
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.252220268494336e-06,
+      "loss": 2.3567,
+      "step": 2087
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.251301851945244e-06,
+      "loss": 2.3489,
+      "step": 2088
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.250382917387915e-06,
+      "loss": 2.4124,
+      "step": 2089
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.24946346493432e-06,
+      "loss": 2.4071,
+      "step": 2090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.248543494696493e-06,
+      "loss": 2.3235,
+      "step": 2091
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.247623006786529e-06,
+      "loss": 2.385,
+      "step": 2092
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.246702001316584e-06,
+      "loss": 2.3707,
+      "step": 2093
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.245780478398883e-06,
+      "loss": 2.3386,
+      "step": 2094
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.244858438145709e-06,
+      "loss": 2.3996,
+      "step": 2095
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.24393588066941e-06,
+      "loss": 2.3796,
+      "step": 2096
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.243012806082398e-06,
+      "loss": 2.3469,
+      "step": 2097
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.242089214497146e-06,
+      "loss": 2.3253,
+      "step": 2098
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.241165106026189e-06,
+      "loss": 2.4285,
+      "step": 2099
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.24024048078213e-06,
+      "loss": 2.4361,
+      "step": 2100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.239315338877632e-06,
+      "loss": 2.4328,
+      "step": 2101
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.238389680425417e-06,
+      "loss": 2.3705,
+      "step": 2102
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.237463505538277e-06,
+      "loss": 2.3237,
+      "step": 2103
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.236536814329062e-06,
+      "loss": 2.4178,
+      "step": 2104
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.235609606910687e-06,
+      "loss": 2.3398,
+      "step": 2105
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.234681883396129e-06,
+      "loss": 2.448,
+      "step": 2106
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.233753643898428e-06,
+      "loss": 2.3488,
+      "step": 2107
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.232824888530689e-06,
+      "loss": 2.3459,
+      "step": 2108
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.231895617406076e-06,
+      "loss": 2.3876,
+      "step": 2109
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.230965830637821e-06,
+      "loss": 2.4327,
+      "step": 2110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.230035528339212e-06,
+      "loss": 2.4216,
+      "step": 2111
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.229104710623604e-06,
+      "loss": 2.4322,
+      "step": 2112
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.228173377604417e-06,
+      "loss": 2.3503,
+      "step": 2113
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.227241529395127e-06,
+      "loss": 2.3839,
+      "step": 2114
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.226309166109281e-06,
+      "loss": 2.3737,
+      "step": 2115
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.225376287860484e-06,
+      "loss": 2.3538,
+      "step": 2116
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.224442894762401e-06,
+      "loss": 2.2659,
+      "step": 2117
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.223508986928766e-06,
+      "loss": 2.2811,
+      "step": 2118
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.222574564473372e-06,
+      "loss": 2.3359,
+      "step": 2119
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.221639627510076e-06,
+      "loss": 2.3118,
+      "step": 2120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.220704176152798e-06,
+      "loss": 2.353,
+      "step": 2121
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.219768210515518e-06,
+      "loss": 2.4108,
+      "step": 2122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.218831730712281e-06,
+      "loss": 2.3697,
+      "step": 2123
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.217894736857195e-06,
+      "loss": 2.3947,
+      "step": 2124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.21695722906443e-06,
+      "loss": 2.4968,
+      "step": 2125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.216019207448216e-06,
+      "loss": 2.3109,
+      "step": 2126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.215080672122854e-06,
+      "loss": 2.441,
+      "step": 2127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.214141623202694e-06,
+      "loss": 2.4304,
+      "step": 2128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.213202060802162e-06,
+      "loss": 2.3558,
+      "step": 2129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.21226198503574e-06,
+      "loss": 2.3689,
+      "step": 2130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.21132139601797e-06,
+      "loss": 2.3434,
+      "step": 2131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.210380293863462e-06,
+      "loss": 2.4139,
+      "step": 2132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.209438678686888e-06,
+      "loss": 2.4469,
+      "step": 2133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.208496550602979e-06,
+      "loss": 2.3562,
+      "step": 2134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.207553909726532e-06,
+      "loss": 2.3804,
+      "step": 2135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.206610756172402e-06,
+      "loss": 2.3485,
+      "step": 2136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.205667090055513e-06,
+      "loss": 2.3469,
+      "step": 2137
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.204722911490847e-06,
+      "loss": 2.3554,
+      "step": 2138
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.203778220593447e-06,
+      "loss": 2.441,
+      "step": 2139
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.202833017478421e-06,
+      "loss": 2.3782,
+      "step": 2140
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.201887302260943e-06,
+      "loss": 2.3295,
+      "step": 2141
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.200941075056242e-06,
+      "loss": 2.397,
+      "step": 2142
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.199994335979613e-06,
+      "loss": 2.4034,
+      "step": 2143
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.199047085146415e-06,
+      "loss": 2.3749,
+      "step": 2144
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.198099322672066e-06,
+      "loss": 2.4088,
+      "step": 2145
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.197151048672051e-06,
+      "loss": 2.3595,
+      "step": 2146
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.196202263261908e-06,
+      "loss": 2.3984,
+      "step": 2147
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.195252966557252e-06,
+      "loss": 2.3863,
+      "step": 2148
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.194303158673744e-06,
+      "loss": 2.3076,
+      "step": 2149
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.193352839727122e-06,
+      "loss": 2.3248,
+      "step": 2150
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.192402009833174e-06,
+      "loss": 2.3622,
+      "step": 2151
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.191450669107758e-06,
+      "loss": 2.366,
+      "step": 2152
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.190498817666793e-06,
+      "loss": 2.3214,
+      "step": 2153
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.189546455626258e-06,
+      "loss": 2.3982,
+      "step": 2154
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.188593583102195e-06,
+      "loss": 2.3784,
+      "step": 2155
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.18764020021071e-06,
+      "loss": 2.4061,
+      "step": 2156
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.186686307067968e-06,
+      "loss": 2.4193,
+      "step": 2157
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.1857319037902e-06,
+      "loss": 2.3614,
+      "step": 2158
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.184776990493696e-06,
+      "loss": 2.3488,
+      "step": 2159
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.18382156729481e-06,
+      "loss": 2.3522,
+      "step": 2160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.182865634309956e-06,
+      "loss": 2.3932,
+      "step": 2161
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.181909191655613e-06,
+      "loss": 2.3971,
+      "step": 2162
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.18095223944832e-06,
+      "loss": 2.4303,
+      "step": 2163
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.179994777804677e-06,
+      "loss": 2.2966,
+      "step": 2164
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.179036806841352e-06,
+      "loss": 2.3802,
+      "step": 2165
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.178078326675069e-06,
+      "loss": 2.318,
+      "step": 2166
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.177119337422613e-06,
+      "loss": 2.3302,
+      "step": 2167
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.176159839200838e-06,
+      "loss": 2.3429,
+      "step": 2168
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.175199832126654e-06,
+      "loss": 2.3592,
+      "step": 2169
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.174239316317034e-06,
+      "loss": 2.3891,
+      "step": 2170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.173278291889016e-06,
+      "loss": 2.294,
+      "step": 2171
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.172316758959695e-06,
+      "loss": 2.3296,
+      "step": 2172
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.171354717646238e-06,
+      "loss": 2.3963,
+      "step": 2173
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.170392168065858e-06,
+      "loss": 2.3675,
+      "step": 2174
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.169429110335842e-06,
+      "loss": 2.4405,
+      "step": 2175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.168465544573538e-06,
+      "loss": 2.3593,
+      "step": 2176
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.16750147089635e-06,
+      "loss": 2.348,
+      "step": 2177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.16653688942175e-06,
+      "loss": 2.3643,
+      "step": 2178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.165571800267267e-06,
+      "loss": 2.3027,
+      "step": 2179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.164606203550498e-06,
+      "loss": 2.4318,
+      "step": 2180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.163640099389095e-06,
+      "loss": 2.3663,
+      "step": 2181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.162673487900775e-06,
+      "loss": 2.34,
+      "step": 2182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.161706369203319e-06,
+      "loss": 2.3994,
+      "step": 2183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.160738743414564e-06,
+      "loss": 2.4099,
+      "step": 2184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.159770610652413e-06,
+      "loss": 2.4459,
+      "step": 2185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.158801971034832e-06,
+      "loss": 2.4108,
+      "step": 2186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.157832824679846e-06,
+      "loss": 2.3304,
+      "step": 2187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.156863171705543e-06,
+      "loss": 2.4193,
+      "step": 2188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.155893012230072e-06,
+      "loss": 2.4024,
+      "step": 2189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.154922346371641e-06,
+      "loss": 2.3395,
+      "step": 2190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.153951174248528e-06,
+      "loss": 2.4044,
+      "step": 2191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.152979495979064e-06,
+      "loss": 2.3677,
+      "step": 2192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.152007311681645e-06,
+      "loss": 2.328,
+      "step": 2193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.15103462147473e-06,
+      "loss": 2.3116,
+      "step": 2194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.150061425476839e-06,
+      "loss": 2.381,
+      "step": 2195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.14908772380655e-06,
+      "loss": 2.3726,
+      "step": 2196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.148113516582508e-06,
+      "loss": 2.3846,
+      "step": 2197
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.147138803923417e-06,
+      "loss": 2.3964,
+      "step": 2198
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.146163585948041e-06,
+      "loss": 2.3341,
+      "step": 2199
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.145187862775208e-06,
+      "loss": 2.2881,
+      "step": 2200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.14421163452381e-06,
+      "loss": 2.3633,
+      "step": 2201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.143234901312794e-06,
+      "loss": 2.2591,
+      "step": 2202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.142257663261173e-06,
+      "loss": 2.3172,
+      "step": 2203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.141279920488021e-06,
+      "loss": 2.2994,
+      "step": 2204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.140301673112472e-06,
+      "loss": 2.3446,
+      "step": 2205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.139322921253724e-06,
+      "loss": 2.366,
+      "step": 2206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.138343665031033e-06,
+      "loss": 2.429,
+      "step": 2207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.13736390456372e-06,
+      "loss": 2.3898,
+      "step": 2208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.136383639971166e-06,
+      "loss": 2.3233,
+      "step": 2209
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.13540287137281e-06,
+      "loss": 2.52,
+      "step": 2210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.13442159888816e-06,
+      "loss": 2.3527,
+      "step": 2211
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.133439822636779e-06,
+      "loss": 2.4339,
+      "step": 2212
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.132457542738292e-06,
+      "loss": 2.4034,
+      "step": 2213
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.13147475931239e-06,
+      "loss": 2.408,
+      "step": 2214
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.130491472478819e-06,
+      "loss": 2.4705,
+      "step": 2215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.129507682357393e-06,
+      "loss": 2.4001,
+      "step": 2216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.128523389067983e-06,
+      "loss": 2.3022,
+      "step": 2217
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.12753859273052e-06,
+      "loss": 2.3962,
+      "step": 2218
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.126553293465e-06,
+      "loss": 2.4135,
+      "step": 2219
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.125567491391476e-06,
+      "loss": 2.4181,
+      "step": 2220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.12458118663007e-06,
+      "loss": 2.368,
+      "step": 2221
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.123594379300956e-06,
+      "loss": 2.3223,
+      "step": 2222
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.122607069524377e-06,
+      "loss": 2.3573,
+      "step": 2223
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.12161925742063e-06,
+      "loss": 2.3125,
+      "step": 2224
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.120630943110078e-06,
+      "loss": 2.3649,
+      "step": 2225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.119642126713147e-06,
+      "loss": 2.3418,
+      "step": 2226
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.11865280835032e-06,
+      "loss": 2.3644,
+      "step": 2227
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.117662988142138e-06,
+      "loss": 2.3109,
+      "step": 2228
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.116672666209211e-06,
+      "loss": 2.3661,
+      "step": 2229
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.115681842672211e-06,
+      "loss": 2.3703,
+      "step": 2230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.11469051765186e-06,
+      "loss": 2.2729,
+      "step": 2231
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.11369869126895e-06,
+      "loss": 2.3947,
+      "step": 2232
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.112706363644334e-06,
+      "loss": 2.3669,
+      "step": 2233
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.111713534898923e-06,
+      "loss": 2.3808,
+      "step": 2234
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.110720205153688e-06,
+      "loss": 2.382,
+      "step": 2235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.109726374529666e-06,
+      "loss": 2.3654,
+      "step": 2236
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.108732043147952e-06,
+      "loss": 2.3688,
+      "step": 2237
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.107737211129702e-06,
+      "loss": 2.4155,
+      "step": 2238
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.106741878596132e-06,
+      "loss": 2.4204,
+      "step": 2239
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.10574604566852e-06,
+      "loss": 2.3864,
+      "step": 2240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.104749712468208e-06,
+      "loss": 2.3526,
+      "step": 2241
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.103752879116595e-06,
+      "loss": 2.3752,
+      "step": 2242
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.102755545735141e-06,
+      "loss": 2.3185,
+      "step": 2243
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.101757712445369e-06,
+      "loss": 2.4028,
+      "step": 2244
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.100759379368863e-06,
+      "loss": 2.4259,
+      "step": 2245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.099760546627262e-06,
+      "loss": 2.3801,
+      "step": 2246
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.098761214342277e-06,
+      "loss": 2.3537,
+      "step": 2247
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.09776138263567e-06,
+      "loss": 2.4273,
+      "step": 2248
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.096761051629268e-06,
+      "loss": 2.3425,
+      "step": 2249
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.09576022144496e-06,
+      "loss": 2.3681,
+      "step": 2250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.09475889220469e-06,
+      "loss": 2.3808,
+      "step": 2251
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.093757064030473e-06,
+      "loss": 2.3469,
+      "step": 2252
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.092754737044375e-06,
+      "loss": 2.2702,
+      "step": 2253
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.091751911368524e-06,
+      "loss": 2.2632,
+      "step": 2254
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.090748587125118e-06,
+      "loss": 2.3355,
+      "step": 2255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.089744764436404e-06,
+      "loss": 2.4028,
+      "step": 2256
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.088740443424695e-06,
+      "loss": 2.4346,
+      "step": 2257
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.087735624212365e-06,
+      "loss": 2.3238,
+      "step": 2258
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.08673030692185e-06,
+      "loss": 2.2797,
+      "step": 2259
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.085724491675642e-06,
+      "loss": 2.416,
+      "step": 2260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.084718178596301e-06,
+      "loss": 2.414,
+      "step": 2261
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.083711367806438e-06,
+      "loss": 2.3732,
+      "step": 2262
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.082704059428732e-06,
+      "loss": 2.4112,
+      "step": 2263
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.08169625358592e-06,
+      "loss": 2.3581,
+      "step": 2264
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.080687950400801e-06,
+      "loss": 2.3616,
+      "step": 2265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.079679149996235e-06,
+      "loss": 2.2941,
+      "step": 2266
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.078669852495138e-06,
+      "loss": 2.3181,
+      "step": 2267
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.077660058020492e-06,
+      "loss": 2.3963,
+      "step": 2268
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.076649766695336e-06,
+      "loss": 2.4067,
+      "step": 2269
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.07563897864277e-06,
+      "loss": 2.378,
+      "step": 2270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.07462769398596e-06,
+      "loss": 2.4078,
+      "step": 2271
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.073615912848126e-06,
+      "loss": 2.4297,
+      "step": 2272
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.072603635352548e-06,
+      "loss": 2.3538,
+      "step": 2273
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.07159086162257e-06,
+      "loss": 2.4408,
+      "step": 2274
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.070577591781598e-06,
+      "loss": 2.3281,
+      "step": 2275
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.069563825953092e-06,
+      "loss": 2.3526,
+      "step": 2276
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.068549564260578e-06,
+      "loss": 2.4077,
+      "step": 2277
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.06753480682764e-06,
+      "loss": 2.3319,
+      "step": 2278
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.066519553777926e-06,
+      "loss": 2.3462,
+      "step": 2279
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.065503805235139e-06,
+      "loss": 2.3406,
+      "step": 2280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.064487561323046e-06,
+      "loss": 2.3635,
+      "step": 2281
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.06347082216547e-06,
+      "loss": 2.4368,
+      "step": 2282
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.062453587886302e-06,
+      "loss": 2.3764,
+      "step": 2283
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.061435858609486e-06,
+      "loss": 2.3587,
+      "step": 2284
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.060417634459032e-06,
+      "loss": 2.3139,
+      "step": 2285
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.059398915559005e-06,
+      "loss": 2.3849,
+      "step": 2286
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.058379702033533e-06,
+      "loss": 2.3976,
+      "step": 2287
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.057359994006806e-06,
+      "loss": 2.3315,
+      "step": 2288
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.056339791603069e-06,
+      "loss": 2.3495,
+      "step": 2289
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.055319094946633e-06,
+      "loss": 2.3373,
+      "step": 2290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.054297904161868e-06,
+      "loss": 2.3449,
+      "step": 2291
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.0532762193732e-06,
+      "loss": 2.388,
+      "step": 2292
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.052254040705121e-06,
+      "loss": 2.3519,
+      "step": 2293
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.051231368282177e-06,
+      "loss": 2.377,
+      "step": 2294
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.050208202228981e-06,
+      "loss": 2.4316,
+      "step": 2295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.0491845426702e-06,
+      "loss": 2.3146,
+      "step": 2296
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.048160389730565e-06,
+      "loss": 2.3728,
+      "step": 2297
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.047135743534866e-06,
+      "loss": 2.3364,
+      "step": 2298
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.046110604207955e-06,
+      "loss": 2.356,
+      "step": 2299
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.045084971874738e-06,
+      "loss": 2.3098,
+      "step": 2300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.044058846660187e-06,
+      "loss": 2.2794,
+      "step": 2301
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.043032228689333e-06,
+      "loss": 2.3651,
+      "step": 2302
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.042005118087267e-06,
+      "loss": 2.3074,
+      "step": 2303
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.040977514979136e-06,
+      "loss": 2.3437,
+      "step": 2304
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.039949419490152e-06,
+      "loss": 2.3945,
+      "step": 2305
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.038920831745587e-06,
+      "loss": 2.3966,
+      "step": 2306
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.037891751870772e-06,
+      "loss": 2.4001,
+      "step": 2307
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.036862179991092e-06,
+      "loss": 2.3311,
+      "step": 2308
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.035832116232002e-06,
+      "loss": 2.3865,
+      "step": 2309
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.03480156071901e-06,
+      "loss": 2.3428,
+      "step": 2310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.033770513577688e-06,
+      "loss": 2.3465,
+      "step": 2311
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.032738974933663e-06,
+      "loss": 2.3997,
+      "step": 2312
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.031706944912627e-06,
+      "loss": 2.3342,
+      "step": 2313
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.03067442364033e-06,
+      "loss": 2.4441,
+      "step": 2314
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.02964141124258e-06,
+      "loss": 2.3473,
+      "step": 2315
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.028607907845247e-06,
+      "loss": 2.3464,
+      "step": 2316
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.02757391357426e-06,
+      "loss": 2.3208,
+      "step": 2317
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.026539428555609e-06,
+      "loss": 2.3757,
+      "step": 2318
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.025504452915345e-06,
+      "loss": 2.366,
+      "step": 2319
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.02446898677957e-06,
+      "loss": 2.3568,
+      "step": 2320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.02343303027446e-06,
+      "loss": 2.3465,
+      "step": 2321
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.022396583526238e-06,
+      "loss": 2.3064,
+      "step": 2322
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.021359646661194e-06,
+      "loss": 2.3355,
+      "step": 2323
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.020322219805674e-06,
+      "loss": 2.3671,
+      "step": 2324
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.019284303086086e-06,
+      "loss": 2.3559,
+      "step": 2325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.0182458966289e-06,
+      "loss": 2.3896,
+      "step": 2326
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.017207000560639e-06,
+      "loss": 2.3894,
+      "step": 2327
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.01616761500789e-06,
+      "loss": 2.3081,
+      "step": 2328
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.015127740097301e-06,
+      "loss": 2.3488,
+      "step": 2329
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.014087375955574e-06,
+      "loss": 2.3383,
+      "step": 2330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.013046522709477e-06,
+      "loss": 2.3762,
+      "step": 2331
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.012005180485834e-06,
+      "loss": 2.3753,
+      "step": 2332
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.010963349411529e-06,
+      "loss": 2.3584,
+      "step": 2333
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.009921029613506e-06,
+      "loss": 2.3595,
+      "step": 2334
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.00887822121877e-06,
+      "loss": 2.4016,
+      "step": 2335
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.007834924354384e-06,
+      "loss": 2.3828,
+      "step": 2336
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.006791139147468e-06,
+      "loss": 2.3105,
+      "step": 2337
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.005746865725206e-06,
+      "loss": 2.3431,
+      "step": 2338
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.00470210421484e-06,
+      "loss": 2.2942,
+      "step": 2339
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.003656854743667e-06,
+      "loss": 2.3473,
+      "step": 2340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.002611117439054e-06,
+      "loss": 2.3068,
+      "step": 2341
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.001564892428416e-06,
+      "loss": 2.3632,
+      "step": 2342
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.000518179839236e-06,
+      "loss": 2.3584,
+      "step": 2343
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.999470979799048e-06,
+      "loss": 2.2905,
+      "step": 2344
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.998423292435455e-06,
+      "loss": 2.2982,
+      "step": 2345
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.99737511787611e-06,
+      "loss": 2.351,
+      "step": 2346
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.996326456248732e-06,
+      "loss": 2.3139,
+      "step": 2347
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.9952773076811e-06,
+      "loss": 2.3389,
+      "step": 2348
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.994227672301046e-06,
+      "loss": 2.3933,
+      "step": 2349
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.993177550236464e-06,
+      "loss": 2.3712,
+      "step": 2350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.992126941615314e-06,
+      "loss": 2.3166,
+      "step": 2351
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.991075846565603e-06,
+      "loss": 2.3823,
+      "step": 2352
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.990024265215405e-06,
+      "loss": 2.3508,
+      "step": 2353
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.988972197692857e-06,
+      "loss": 2.3553,
+      "step": 2354
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.987919644126145e-06,
+      "loss": 2.4348,
+      "step": 2355
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.986866604643518e-06,
+      "loss": 2.3783,
+      "step": 2356
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.985813079373293e-06,
+      "loss": 2.3128,
+      "step": 2357
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.984759068443832e-06,
+      "loss": 2.4309,
+      "step": 2358
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.983704571983568e-06,
+      "loss": 2.4014,
+      "step": 2359
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.982649590120982e-06,
+      "loss": 2.3918,
+      "step": 2360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.981594122984628e-06,
+      "loss": 2.3876,
+      "step": 2361
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.980538170703104e-06,
+      "loss": 2.4133,
+      "step": 2362
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.97948173340508e-06,
+      "loss": 2.2512,
+      "step": 2363
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.978424811219277e-06,
+      "loss": 2.413,
+      "step": 2364
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.97736740427448e-06,
+      "loss": 2.3113,
+      "step": 2365
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.97630951269953e-06,
+      "loss": 2.3314,
+      "step": 2366
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.975251136623326e-06,
+      "loss": 2.372,
+      "step": 2367
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.97419227617483e-06,
+      "loss": 2.3331,
+      "step": 2368
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.973132931483057e-06,
+      "loss": 2.3156,
+      "step": 2369
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.972073102677091e-06,
+      "loss": 2.3968,
+      "step": 2370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.971012789886066e-06,
+      "loss": 2.3994,
+      "step": 2371
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.969951993239177e-06,
+      "loss": 2.3333,
+      "step": 2372
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.96889071286568e-06,
+      "loss": 2.3223,
+      "step": 2373
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.96782894889489e-06,
+      "loss": 2.3175,
+      "step": 2374
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.966766701456177e-06,
+      "loss": 2.3248,
+      "step": 2375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.965703970678974e-06,
+      "loss": 2.2949,
+      "step": 2376
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.96464075669277e-06,
+      "loss": 2.2976,
+      "step": 2377
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.963577059627117e-06,
+      "loss": 2.3432,
+      "step": 2378
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.962512879611624e-06,
+      "loss": 2.4087,
+      "step": 2379
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.961448216775955e-06,
+      "loss": 2.3366,
+      "step": 2380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.960383071249837e-06,
+      "loss": 2.3347,
+      "step": 2381
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.959317443163054e-06,
+      "loss": 2.3703,
+      "step": 2382
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.95825133264545e-06,
+      "loss": 2.4417,
+      "step": 2383
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.957184739826929e-06,
+      "loss": 2.3191,
+      "step": 2384
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.956117664837452e-06,
+      "loss": 2.3267,
+      "step": 2385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.955050107807035e-06,
+      "loss": 2.3655,
+      "step": 2386
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.95398206886576e-06,
+      "loss": 2.3795,
+      "step": 2387
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.952913548143766e-06,
+      "loss": 2.3534,
+      "step": 2388
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.951844545771244e-06,
+      "loss": 2.3474,
+      "step": 2389
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.950775061878453e-06,
+      "loss": 2.3225,
+      "step": 2390
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.949705096595704e-06,
+      "loss": 2.3541,
+      "step": 2391
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.94863465005337e-06,
+      "loss": 2.312,
+      "step": 2392
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.94756372238188e-06,
+      "loss": 2.3658,
+      "step": 2393
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.946492313711725e-06,
+      "loss": 2.3681,
+      "step": 2394
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.945420424173455e-06,
+      "loss": 2.3495,
+      "step": 2395
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.944348053897672e-06,
+      "loss": 2.3298,
+      "step": 2396
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.943275203015042e-06,
+      "loss": 2.369,
+      "step": 2397
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.942201871656292e-06,
+      "loss": 2.2441,
+      "step": 2398
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.9411280599522e-06,
+      "loss": 2.3447,
+      "step": 2399
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.94005376803361e-06,
+      "loss": 2.3574,
+      "step": 2400
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.93897899603142e-06,
+      "loss": 2.3365,
+      "step": 2401
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.937903744076587e-06,
+      "loss": 2.3684,
+      "step": 2402
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.936828012300127e-06,
+      "loss": 2.3742,
+      "step": 2403
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.935751800833117e-06,
+      "loss": 2.3256,
+      "step": 2404
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.934675109806688e-06,
+      "loss": 2.3749,
+      "step": 2405
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.933597939352031e-06,
+      "loss": 2.2932,
+      "step": 2406
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.932520289600396e-06,
+      "loss": 2.3765,
+      "step": 2407
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.931442160683094e-06,
+      "loss": 2.4431,
+      "step": 2408
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.930363552731491e-06,
+      "loss": 2.3795,
+      "step": 2409
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.92928446587701e-06,
+      "loss": 2.2582,
+      "step": 2410
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.928204900251136e-06,
+      "loss": 2.3849,
+      "step": 2411
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.92712485598541e-06,
+      "loss": 2.3643,
+      "step": 2412
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.926044333211433e-06,
+      "loss": 2.3215,
+      "step": 2413
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.924963332060863e-06,
+      "loss": 2.3607,
+      "step": 2414
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.923881852665416e-06,
+      "loss": 2.4022,
+      "step": 2415
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.922799895156868e-06,
+      "loss": 2.3596,
+      "step": 2416
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.921717459667052e-06,
+      "loss": 2.3077,
+      "step": 2417
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.920634546327857e-06,
+      "loss": 2.3374,
+      "step": 2418
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.919551155271239e-06,
+      "loss": 2.4082,
+      "step": 2419
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.9184672866292e-06,
+      "loss": 2.2832,
+      "step": 2420
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.917382940533809e-06,
+      "loss": 2.316,
+      "step": 2421
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.916298117117188e-06,
+      "loss": 2.3216,
+      "step": 2422
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.915212816511521e-06,
+      "loss": 2.3863,
+      "step": 2423
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.91412703884905e-06,
+      "loss": 2.3413,
+      "step": 2424
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.91304078426207e-06,
+      "loss": 2.5047,
+      "step": 2425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.911954052882941e-06,
+      "loss": 2.344,
+      "step": 2426
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.910866844844077e-06,
+      "loss": 2.2951,
+      "step": 2427
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.909779160277951e-06,
+      "loss": 2.3764,
+      "step": 2428
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.908690999317094e-06,
+      "loss": 2.2939,
+      "step": 2429
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.907602362094094e-06,
+      "loss": 2.3611,
+      "step": 2430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.9065132487416e-06,
+      "loss": 2.3828,
+      "step": 2431
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.905423659392316e-06,
+      "loss": 2.3301,
+      "step": 2432
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.904333594179007e-06,
+      "loss": 2.329,
+      "step": 2433
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.903243053234492e-06,
+      "loss": 2.3743,
+      "step": 2434
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.902152036691649e-06,
+      "loss": 2.321,
+      "step": 2435
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.90106054468342e-06,
+      "loss": 2.2918,
+      "step": 2436
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.899968577342794e-06,
+      "loss": 2.34,
+      "step": 2437
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.898876134802827e-06,
+      "loss": 2.3367,
+      "step": 2438
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.897783217196629e-06,
+      "loss": 2.3076,
+      "step": 2439
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.896689824657371e-06,
+      "loss": 2.2867,
+      "step": 2440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.895595957318277e-06,
+      "loss": 2.3175,
+      "step": 2441
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.894501615312633e-06,
+      "loss": 2.3086,
+      "step": 2442
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.89340679877378e-06,
+      "loss": 2.4127,
+      "step": 2443
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.892311507835118e-06,
+      "loss": 2.3991,
+      "step": 2444
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.891215742630106e-06,
+      "loss": 2.3849,
+      "step": 2445
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.890119503292258e-06,
+      "loss": 2.3628,
+      "step": 2446
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.88902278995515e-06,
+      "loss": 2.3502,
+      "step": 2447
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.887925602752411e-06,
+      "loss": 2.3555,
+      "step": 2448
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.886827941817731e-06,
+      "loss": 2.3499,
+      "step": 2449
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.885729807284855e-06,
+      "loss": 2.2444,
+      "step": 2450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.88463119928759e-06,
+      "loss": 2.3718,
+      "step": 2451
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.883532117959797e-06,
+      "loss": 2.4212,
+      "step": 2452
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.882432563435394e-06,
+      "loss": 2.2995,
+      "step": 2453
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.88133253584836e-06,
+      "loss": 2.3619,
+      "step": 2454
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.88023203533273e-06,
+      "loss": 2.3797,
+      "step": 2455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.879131062022598e-06,
+      "loss": 2.31,
+      "step": 2456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.87802961605211e-06,
+      "loss": 2.3686,
+      "step": 2457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.87692769755548e-06,
+      "loss": 2.4397,
+      "step": 2458
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.875825306666968e-06,
+      "loss": 2.3933,
+      "step": 2459
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.874722443520898e-06,
+      "loss": 2.3543,
+      "step": 2460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.873619108251654e-06,
+      "loss": 2.3472,
+      "step": 2461
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.872515300993669e-06,
+      "loss": 2.4183,
+      "step": 2462
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.871411021881444e-06,
+      "loss": 2.2953,
+      "step": 2463
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.870306271049527e-06,
+      "loss": 2.4065,
+      "step": 2464
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.869201048632531e-06,
+      "loss": 2.3628,
+      "step": 2465
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.868095354765125e-06,
+      "loss": 2.3374,
+      "step": 2466
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.866989189582033e-06,
+      "loss": 2.3639,
+      "step": 2467
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.865882553218036e-06,
+      "loss": 2.3239,
+      "step": 2468
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.864775445807979e-06,
+      "loss": 2.3715,
+      "step": 2469
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.863667867486756e-06,
+      "loss": 2.3391,
+      "step": 2470
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.862559818389322e-06,
+      "loss": 2.2802,
+      "step": 2471
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.861451298650692e-06,
+      "loss": 2.381,
+      "step": 2472
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.860342308405933e-06,
+      "loss": 2.3719,
+      "step": 2473
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.859232847790175e-06,
+      "loss": 2.3334,
+      "step": 2474
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.858122916938601e-06,
+      "loss": 2.3227,
+      "step": 2475
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.857012515986452e-06,
+      "loss": 2.3238,
+      "step": 2476
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.855901645069026e-06,
+      "loss": 2.3647,
+      "step": 2477
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.854790304321682e-06,
+      "loss": 2.3733,
+      "step": 2478
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.853678493879832e-06,
+      "loss": 2.3268,
+      "step": 2479
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.852566213878947e-06,
+      "loss": 2.2896,
+      "step": 2480
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.851453464454555e-06,
+      "loss": 2.3798,
+      "step": 2481
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.85034024574224e-06,
+      "loss": 2.3174,
+      "step": 2482
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.849226557877647e-06,
+      "loss": 2.3016,
+      "step": 2483
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.848112400996473e-06,
+      "loss": 2.4,
+      "step": 2484
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.846997775234476e-06,
+      "loss": 2.2909,
+      "step": 2485
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.84588268072747e-06,
+      "loss": 2.3254,
+      "step": 2486
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.844767117611324e-06,
+      "loss": 2.3401,
+      "step": 2487
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.843651086021966e-06,
+      "loss": 2.3699,
+      "step": 2488
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.842534586095383e-06,
+      "loss": 2.2892,
+      "step": 2489
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.841417617967618e-06,
+      "loss": 2.3511,
+      "step": 2490
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.840300181774767e-06,
+      "loss": 2.3438,
+      "step": 2491
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.83918227765299e-06,
+      "loss": 2.379,
+      "step": 2492
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.838063905738495e-06,
+      "loss": 2.3012,
+      "step": 2493
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.836945066167556e-06,
+      "loss": 2.3322,
+      "step": 2494
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.8358257590765e-06,
+      "loss": 2.2249,
+      "step": 2495
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.834705984601708e-06,
+      "loss": 2.3134,
+      "step": 2496
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.833585742879627e-06,
+      "loss": 2.3041,
+      "step": 2497
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.83246503404675e-06,
+      "loss": 2.3425,
+      "step": 2498
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.831343858239634e-06,
+      "loss": 2.3197,
+      "step": 2499
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.83022221559489e-06,
+      "loss": 2.3319,
+      "step": 2500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.829100106249189e-06,
+      "loss": 2.2551,
+      "step": 2501
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.827977530339254e-06,
+      "loss": 2.3421,
+      "step": 2502
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.826854488001869e-06,
+      "loss": 2.2845,
+      "step": 2503
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.825730979373873e-06,
+      "loss": 2.3095,
+      "step": 2504
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.824607004592161e-06,
+      "loss": 2.3544,
+      "step": 2505
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.823482563793687e-06,
+      "loss": 2.3279,
+      "step": 2506
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.82235765711546e-06,
+      "loss": 2.3303,
+      "step": 2507
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.821232284694545e-06,
+      "loss": 2.3059,
+      "step": 2508
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.820106446668071e-06,
+      "loss": 2.3639,
+      "step": 2509
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.818980143173212e-06,
+      "loss": 2.2859,
+      "step": 2510
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.817853374347208e-06,
+      "loss": 2.3562,
+      "step": 2511
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.81672614032735e-06,
+      "loss": 2.3614,
+      "step": 2512
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.815598441250987e-06,
+      "loss": 2.3038,
+      "step": 2513
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.814470277255532e-06,
+      "loss": 2.3468,
+      "step": 2514
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.813341648478443e-06,
+      "loss": 2.3123,
+      "step": 2515
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.81221255505724e-06,
+      "loss": 2.38,
+      "step": 2516
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.811082997129501e-06,
+      "loss": 2.289,
+      "step": 2517
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.80995297483286e-06,
+      "loss": 2.2733,
+      "step": 2518
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.808822488305005e-06,
+      "loss": 2.3584,
+      "step": 2519
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.807691537683685e-06,
+      "loss": 2.2975,
+      "step": 2520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.8065601231067e-06,
+      "loss": 2.4128,
+      "step": 2521
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.80542824471191e-06,
+      "loss": 2.309,
+      "step": 2522
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.804295902637233e-06,
+      "loss": 2.3441,
+      "step": 2523
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.803163097020637e-06,
+      "loss": 2.4029,
+      "step": 2524
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.802029828000157e-06,
+      "loss": 2.3703,
+      "step": 2525
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.80089609571387e-06,
+      "loss": 2.3092,
+      "step": 2526
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.799761900299929e-06,
+      "loss": 2.2972,
+      "step": 2527
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.798627241896524e-06,
+      "loss": 2.2982,
+      "step": 2528
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.797492120641913e-06,
+      "loss": 2.3053,
+      "step": 2529
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.796356536674404e-06,
+      "loss": 2.3811,
+      "step": 2530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.795220490132369e-06,
+      "loss": 2.4129,
+      "step": 2531
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.794083981154229e-06,
+      "loss": 2.3321,
+      "step": 2532
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.792947009878463e-06,
+      "loss": 2.3404,
+      "step": 2533
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.791809576443611e-06,
+      "loss": 2.4667,
+      "step": 2534
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.790671680988261e-06,
+      "loss": 2.2724,
+      "step": 2535
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.789533323651067e-06,
+      "loss": 2.3319,
+      "step": 2536
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.788394504570732e-06,
+      "loss": 2.3726,
+      "step": 2537
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.78725522388602e-06,
+      "loss": 2.3488,
+      "step": 2538
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.786115481735745e-06,
+      "loss": 2.3699,
+      "step": 2539
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.784975278258783e-06,
+      "loss": 2.3487,
+      "step": 2540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.783834613594064e-06,
+      "loss": 2.3937,
+      "step": 2541
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.782693487880575e-06,
+      "loss": 2.34,
+      "step": 2542
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.78155190125736e-06,
+      "loss": 2.3734,
+      "step": 2543
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.780409853863517e-06,
+      "loss": 2.3359,
+      "step": 2544
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.779267345838198e-06,
+      "loss": 2.2712,
+      "step": 2545
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.778124377320619e-06,
+      "loss": 2.3337,
+      "step": 2546
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.776980948450043e-06,
+      "loss": 2.3609,
+      "step": 2547
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.775837059365796e-06,
+      "loss": 2.2416,
+      "step": 2548
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.774692710207257e-06,
+      "loss": 2.2553,
+      "step": 2549
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.773547901113862e-06,
+      "loss": 2.3701,
+      "step": 2550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.772402632225098e-06,
+      "loss": 2.2254,
+      "step": 2551
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.77125690368052e-06,
+      "loss": 2.4131,
+      "step": 2552
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.770110715619726e-06,
+      "loss": 2.3252,
+      "step": 2553
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.768964068182378e-06,
+      "loss": 2.342,
+      "step": 2554
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.767816961508191e-06,
+      "loss": 2.318,
+      "step": 2555
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.766669395736936e-06,
+      "loss": 2.4172,
+      "step": 2556
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.765521371008439e-06,
+      "loss": 2.3116,
+      "step": 2557
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.764372887462587e-06,
+      "loss": 2.2788,
+      "step": 2558
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.763223945239317e-06,
+      "loss": 2.3197,
+      "step": 2559
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.762074544478622e-06,
+      "loss": 2.3252,
+      "step": 2560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.760924685320558e-06,
+      "loss": 2.354,
+      "step": 2561
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.759774367905228e-06,
+      "loss": 2.3252,
+      "step": 2562
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.758623592372797e-06,
+      "loss": 2.3113,
+      "step": 2563
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.757472358863481e-06,
+      "loss": 2.3739,
+      "step": 2564
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.756320667517557e-06,
+      "loss": 2.3316,
+      "step": 2565
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.755168518475351e-06,
+      "loss": 2.333,
+      "step": 2566
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.754015911877255e-06,
+      "loss": 2.3659,
+      "step": 2567
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.752862847863707e-06,
+      "loss": 2.3383,
+      "step": 2568
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.751709326575204e-06,
+      "loss": 2.3133,
+      "step": 2569
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.750555348152299e-06,
+      "loss": 2.3619,
+      "step": 2570
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.749400912735602e-06,
+      "loss": 2.3582,
+      "step": 2571
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.748246020465776e-06,
+      "loss": 2.3538,
+      "step": 2572
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.747090671483542e-06,
+      "loss": 2.3235,
+      "step": 2573
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.745934865929676e-06,
+      "loss": 2.3554,
+      "step": 2574
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.744778603945013e-06,
+      "loss": 2.295,
+      "step": 2575
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.743621885670431e-06,
+      "loss": 2.3514,
+      "step": 2576
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.74246471124688e-06,
+      "loss": 2.3342,
+      "step": 2577
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.741307080815357e-06,
+      "loss": 2.3574,
+      "step": 2578
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.740148994516912e-06,
+      "loss": 2.4055,
+      "step": 2579
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.73899045249266e-06,
+      "loss": 2.3082,
+      "step": 2580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.737831454883762e-06,
+      "loss": 2.3282,
+      "step": 2581
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.736672001831438e-06,
+      "loss": 2.3223,
+      "step": 2582
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.735512093476968e-06,
+      "loss": 2.3911,
+      "step": 2583
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.73435172996168e-06,
+      "loss": 2.3661,
+      "step": 2584
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.733190911426957e-06,
+      "loss": 2.3631,
+      "step": 2585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.732029638014249e-06,
+      "loss": 2.3525,
+      "step": 2586
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.730867909865048e-06,
+      "loss": 2.3899,
+      "step": 2587
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.729705727120911e-06,
+      "loss": 2.3634,
+      "step": 2588
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.728543089923444e-06,
+      "loss": 2.3465,
+      "step": 2589
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.727379998414311e-06,
+      "loss": 2.3433,
+      "step": 2590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.726216452735233e-06,
+      "loss": 2.3328,
+      "step": 2591
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.725052453027982e-06,
+      "loss": 2.394,
+      "step": 2592
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.723887999434389e-06,
+      "loss": 2.3583,
+      "step": 2593
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.722723092096337e-06,
+      "loss": 2.3658,
+      "step": 2594
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.72155773115577e-06,
+      "loss": 2.3137,
+      "step": 2595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.720391916754683e-06,
+      "loss": 2.3545,
+      "step": 2596
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.719225649035126e-06,
+      "loss": 2.2535,
+      "step": 2597
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.718058928139205e-06,
+      "loss": 2.3618,
+      "step": 2598
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.716891754209081e-06,
+      "loss": 2.3746,
+      "step": 2599
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.715724127386971e-06,
+      "loss": 2.3283,
+      "step": 2600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.714556047815148e-06,
+      "loss": 2.2878,
+      "step": 2601
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.713387515635938e-06,
+      "loss": 2.3157,
+      "step": 2602
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.712218530991723e-06,
+      "loss": 2.3505,
+      "step": 2603
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.711049094024942e-06,
+      "loss": 2.3705,
+      "step": 2604
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.709879204878082e-06,
+      "loss": 2.3847,
+      "step": 2605
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.708708863693696e-06,
+      "loss": 2.3381,
+      "step": 2606
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.707538070614385e-06,
+      "loss": 2.2994,
+      "step": 2607
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.706366825782805e-06,
+      "loss": 2.4249,
+      "step": 2608
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.705195129341672e-06,
+      "loss": 2.3034,
+      "step": 2609
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.70402298143375e-06,
+      "loss": 2.2915,
+      "step": 2610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.702850382201863e-06,
+      "loss": 2.3612,
+      "step": 2611
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.701677331788891e-06,
+      "loss": 2.3457,
+      "step": 2612
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.700503830337763e-06,
+      "loss": 2.3344,
+      "step": 2613
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.699329877991469e-06,
+      "loss": 2.3489,
+      "step": 2614
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.69815547489305e-06,
+      "loss": 2.36,
+      "step": 2615
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.696980621185602e-06,
+      "loss": 2.2989,
+      "step": 2616
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.695805317012283e-06,
+      "loss": 2.386,
+      "step": 2617
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.694629562516295e-06,
+      "loss": 2.3198,
+      "step": 2618
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.6934533578409e-06,
+      "loss": 2.2867,
+      "step": 2619
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.692276703129421e-06,
+      "loss": 2.3059,
+      "step": 2620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.691099598525222e-06,
+      "loss": 2.3367,
+      "step": 2621
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.689922044171735e-06,
+      "loss": 2.2317,
+      "step": 2622
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.688744040212438e-06,
+      "loss": 2.3149,
+      "step": 2623
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.68756558679087e-06,
+      "loss": 2.3098,
+      "step": 2624
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.68638668405062e-06,
+      "loss": 2.3103,
+      "step": 2625
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.685207332135337e-06,
+      "loss": 2.3633,
+      "step": 2626
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.684027531188717e-06,
+      "loss": 2.3914,
+      "step": 2627
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.682847281354517e-06,
+      "loss": 2.2749,
+      "step": 2628
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.681666582776547e-06,
+      "loss": 2.3898,
+      "step": 2629
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.680485435598674e-06,
+      "loss": 2.238,
+      "step": 2630
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.679303839964811e-06,
+      "loss": 2.3803,
+      "step": 2631
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.678121796018938e-06,
+      "loss": 2.3567,
+      "step": 2632
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.67693930390508e-06,
+      "loss": 2.3226,
+      "step": 2633
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.675756363767322e-06,
+      "loss": 2.301,
+      "step": 2634
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.6745729757498e-06,
+      "loss": 2.3725,
+      "step": 2635
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.673389139996708e-06,
+      "loss": 2.2522,
+      "step": 2636
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.672204856652291e-06,
+      "loss": 2.2112,
+      "step": 2637
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.671020125860851e-06,
+      "loss": 2.3389,
+      "step": 2638
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.669834947766746e-06,
+      "loss": 2.3101,
+      "step": 2639
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.668649322514382e-06,
+      "loss": 2.3327,
+      "step": 2640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.667463250248229e-06,
+      "loss": 2.3084,
+      "step": 2641
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.666276731112802e-06,
+      "loss": 2.3741,
+      "step": 2642
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.665089765252674e-06,
+      "loss": 2.3841,
+      "step": 2643
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.66390235281248e-06,
+      "loss": 2.3506,
+      "step": 2644
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.662714493936895e-06,
+      "loss": 2.4059,
+      "step": 2645
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.66152618877066e-06,
+      "loss": 2.2906,
+      "step": 2646
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.660337437458565e-06,
+      "loss": 2.3219,
+      "step": 2647
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.659148240145456e-06,
+      "loss": 2.3512,
+      "step": 2648
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.657958596976235e-06,
+      "loss": 2.3033,
+      "step": 2649
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.656768508095853e-06,
+      "loss": 2.3017,
+      "step": 2650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.655577973649322e-06,
+      "loss": 2.3305,
+      "step": 2651
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.654386993781703e-06,
+      "loss": 2.2947,
+      "step": 2652
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.653195568638114e-06,
+      "loss": 2.3015,
+      "step": 2653
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.652003698363724e-06,
+      "loss": 2.4233,
+      "step": 2654
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.650811383103763e-06,
+      "loss": 2.2914,
+      "step": 2655
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.649618623003509e-06,
+      "loss": 2.4175,
+      "step": 2656
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.648425418208294e-06,
+      "loss": 2.333,
+      "step": 2657
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.647231768863513e-06,
+      "loss": 2.4235,
+      "step": 2658
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.6460376751146e-06,
+      "loss": 2.2854,
+      "step": 2659
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.644843137107058e-06,
+      "loss": 2.4026,
+      "step": 2660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.643648154986436e-06,
+      "loss": 2.3263,
+      "step": 2661
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.642452728898339e-06,
+      "loss": 2.3378,
+      "step": 2662
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.641256858988424e-06,
+      "loss": 2.3047,
+      "step": 2663
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.640060545402407e-06,
+      "loss": 2.3113,
+      "step": 2664
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.638863788286054e-06,
+      "loss": 2.2918,
+      "step": 2665
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.637666587785185e-06,
+      "loss": 2.3672,
+      "step": 2666
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.636468944045677e-06,
+      "loss": 2.3155,
+      "step": 2667
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.63527085721346e-06,
+      "loss": 2.3999,
+      "step": 2668
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.634072327434515e-06,
+      "loss": 2.3244,
+      "step": 2669
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.632873354854881e-06,
+      "loss": 2.3008,
+      "step": 2670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.631673939620647e-06,
+      "loss": 2.3406,
+      "step": 2671
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.630474081877959e-06,
+      "loss": 2.3057,
+      "step": 2672
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.62927378177302e-06,
+      "loss": 2.3074,
+      "step": 2673
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.628073039452076e-06,
+      "loss": 2.3089,
+      "step": 2674
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.626871855061438e-06,
+      "loss": 2.3595,
+      "step": 2675
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.625670228747467e-06,
+      "loss": 2.4221,
+      "step": 2676
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.624468160656576e-06,
+      "loss": 2.3001,
+      "step": 2677
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.623265650935233e-06,
+      "loss": 2.3068,
+      "step": 2678
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.622062699729963e-06,
+      "loss": 2.2787,
+      "step": 2679
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.620859307187339e-06,
+      "loss": 2.3308,
+      "step": 2680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.61965547345399e-06,
+      "loss": 2.3488,
+      "step": 2681
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.618451198676602e-06,
+      "loss": 2.3022,
+      "step": 2682
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.617246483001914e-06,
+      "loss": 2.3606,
+      "step": 2683
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.616041326576711e-06,
+      "loss": 2.3335,
+      "step": 2684
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.614835729547841e-06,
+      "loss": 2.3549,
+      "step": 2685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.613629692062204e-06,
+      "loss": 2.3748,
+      "step": 2686
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.612423214266749e-06,
+      "loss": 2.2627,
+      "step": 2687
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.611216296308485e-06,
+      "loss": 2.292,
+      "step": 2688
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.610008938334467e-06,
+      "loss": 2.3923,
+      "step": 2689
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.608801140491811e-06,
+      "loss": 2.2919,
+      "step": 2690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.607592902927684e-06,
+      "loss": 2.3426,
+      "step": 2691
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.606384225789304e-06,
+      "loss": 2.3285,
+      "step": 2692
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.605175109223945e-06,
+      "loss": 2.3296,
+      "step": 2693
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.603965553378934e-06,
+      "loss": 2.2913,
+      "step": 2694
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.602755558401653e-06,
+      "loss": 2.3519,
+      "step": 2695
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.601545124439535e-06,
+      "loss": 2.3193,
+      "step": 2696
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.60033425164007e-06,
+      "loss": 2.3692,
+      "step": 2697
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.599122940150795e-06,
+      "loss": 2.2839,
+      "step": 2698
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.597911190119308e-06,
+      "loss": 2.3072,
+      "step": 2699
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.596699001693257e-06,
+      "loss": 2.3492,
+      "step": 2700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.59548637502034e-06,
+      "loss": 2.3913,
+      "step": 2701
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.594273310248317e-06,
+      "loss": 2.332,
+      "step": 2702
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.593059807524993e-06,
+      "loss": 2.3728,
+      "step": 2703
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.591845866998231e-06,
+      "loss": 2.3675,
+      "step": 2704
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.590631488815945e-06,
+      "loss": 2.3424,
+      "step": 2705
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.589416673126104e-06,
+      "loss": 2.2531,
+      "step": 2706
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.588201420076727e-06,
+      "loss": 2.3253,
+      "step": 2707
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.586985729815895e-06,
+      "loss": 2.2733,
+      "step": 2708
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.585769602491729e-06,
+      "loss": 2.2545,
+      "step": 2709
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.584553038252415e-06,
+      "loss": 2.3584,
+      "step": 2710
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.583336037246187e-06,
+      "loss": 2.3063,
+      "step": 2711
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.58211859962133e-06,
+      "loss": 2.3875,
+      "step": 2712
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.580900725526189e-06,
+      "loss": 2.3187,
+      "step": 2713
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.579682415109156e-06,
+      "loss": 2.3496,
+      "step": 2714
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.57846366851868e-06,
+      "loss": 2.3882,
+      "step": 2715
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.57724448590326e-06,
+      "loss": 2.3272,
+      "step": 2716
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.576024867411452e-06,
+      "loss": 2.3497,
+      "step": 2717
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.574804813191859e-06,
+      "loss": 2.2817,
+      "step": 2718
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.573584323393142e-06,
+      "loss": 2.3694,
+      "step": 2719
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.572363398164017e-06,
+      "loss": 2.3319,
+      "step": 2720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.571142037653249e-06,
+      "loss": 2.3215,
+      "step": 2721
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.569920242009655e-06,
+      "loss": 2.3224,
+      "step": 2722
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.568698011382108e-06,
+      "loss": 2.2807,
+      "step": 2723
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.567475345919532e-06,
+      "loss": 2.3729,
+      "step": 2724
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.56625224577091e-06,
+      "loss": 2.3361,
+      "step": 2725
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.565028711085266e-06,
+      "loss": 2.3031,
+      "step": 2726
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.563804742011689e-06,
+      "loss": 2.3089,
+      "step": 2727
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.562580338699313e-06,
+      "loss": 2.3191,
+      "step": 2728
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.56135550129733e-06,
+      "loss": 2.4013,
+      "step": 2729
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.560130229954985e-06,
+      "loss": 2.314,
+      "step": 2730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.558904524821565e-06,
+      "loss": 2.3541,
+      "step": 2731
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.557678386046429e-06,
+      "loss": 2.3388,
+      "step": 2732
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.55645181377897e-06,
+      "loss": 2.3411,
+      "step": 2733
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.555224808168644e-06,
+      "loss": 2.3375,
+      "step": 2734
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.553997369364964e-06,
+      "loss": 2.3375,
+      "step": 2735
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.55276949751748e-06,
+      "loss": 2.3295,
+      "step": 2736
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.551541192775813e-06,
+      "loss": 2.2908,
+      "step": 2737
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.550312455289624e-06,
+      "loss": 2.3618,
+      "step": 2738
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.549083285208632e-06,
+      "loss": 2.3581,
+      "step": 2739
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.547853682682605e-06,
+      "loss": 2.2965,
+      "step": 2740
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.54662364786137e-06,
+      "loss": 2.3494,
+      "step": 2741
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.545393180894801e-06,
+      "loss": 2.3149,
+      "step": 2742
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.544162281932829e-06,
+      "loss": 2.3864,
+      "step": 2743
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.542930951125432e-06,
+      "loss": 2.3414,
+      "step": 2744
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.541699188622645e-06,
+      "loss": 2.3561,
+      "step": 2745
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.540466994574556e-06,
+      "loss": 2.3503,
+      "step": 2746
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.539234369131301e-06,
+      "loss": 2.3199,
+      "step": 2747
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.538001312443078e-06,
+      "loss": 2.3127,
+      "step": 2748
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.536767824660124e-06,
+      "loss": 2.3131,
+      "step": 2749
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.535533905932739e-06,
+      "loss": 2.307,
+      "step": 2750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.534299556411272e-06,
+      "loss": 2.3033,
+      "step": 2751
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.533064776246126e-06,
+      "loss": 2.3134,
+      "step": 2752
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.531829565587751e-06,
+      "loss": 2.2964,
+      "step": 2753
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.530593924586659e-06,
+      "loss": 2.2424,
+      "step": 2754
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.529357853393406e-06,
+      "loss": 2.3108,
+      "step": 2755
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.528121352158604e-06,
+      "loss": 2.341,
+      "step": 2756
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.526884421032916e-06,
+      "loss": 2.2804,
+      "step": 2757
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.525647060167063e-06,
+      "loss": 2.2672,
+      "step": 2758
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.524409269711808e-06,
+      "loss": 2.2841,
+      "step": 2759
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.523171049817974e-06,
+      "loss": 2.3385,
+      "step": 2760
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.521932400636435e-06,
+      "loss": 2.3076,
+      "step": 2761
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.520693322318116e-06,
+      "loss": 2.2829,
+      "step": 2762
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.519453815013996e-06,
+      "loss": 2.3448,
+      "step": 2763
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.518213878875103e-06,
+      "loss": 2.2863,
+      "step": 2764
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.51697351405252e-06,
+      "loss": 2.312,
+      "step": 2765
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.515732720697383e-06,
+      "loss": 2.3424,
+      "step": 2766
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.51449149896088e-06,
+      "loss": 2.3662,
+      "step": 2767
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.513249848994248e-06,
+      "loss": 2.3274,
+      "step": 2768
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.512007770948775e-06,
+      "loss": 2.3066,
+      "step": 2769
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.510765264975813e-06,
+      "loss": 2.2733,
+      "step": 2770
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.509522331226751e-06,
+      "loss": 2.334,
+      "step": 2771
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.508278969853037e-06,
+      "loss": 2.2729,
+      "step": 2772
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.507035181006175e-06,
+      "loss": 2.3591,
+      "step": 2773
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.505790964837712e-06,
+      "loss": 2.3795,
+      "step": 2774
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.504546321499255e-06,
+      "loss": 2.277,
+      "step": 2775
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.50330125114246e-06,
+      "loss": 2.3289,
+      "step": 2776
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.502055753919033e-06,
+      "loss": 2.332,
+      "step": 2777
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.500809829980734e-06,
+      "loss": 2.3074,
+      "step": 2778
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.499563479479378e-06,
+      "loss": 2.3172,
+      "step": 2779
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.498316702566828e-06,
+      "loss": 2.2976,
+      "step": 2780
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.497069499394998e-06,
+      "loss": 2.3032,
+      "step": 2781
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.495821870115857e-06,
+      "loss": 2.342,
+      "step": 2782
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.494573814881426e-06,
+      "loss": 2.3124,
+      "step": 2783
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.493325333843776e-06,
+      "loss": 2.3494,
+      "step": 2784
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.492076427155031e-06,
+      "loss": 2.3429,
+      "step": 2785
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.490827094967364e-06,
+      "loss": 2.315,
+      "step": 2786
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.489577337433006e-06,
+      "loss": 2.3244,
+      "step": 2787
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.488327154704232e-06,
+      "loss": 2.3471,
+      "step": 2788
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.487076546933378e-06,
+      "loss": 2.2648,
+      "step": 2789
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.485825514272824e-06,
+      "loss": 2.339,
+      "step": 2790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.484574056875004e-06,
+      "loss": 2.372,
+      "step": 2791
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.483322174892404e-06,
+      "loss": 2.287,
+      "step": 2792
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.482069868477565e-06,
+      "loss": 2.2904,
+      "step": 2793
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.480817137783073e-06,
+      "loss": 2.3538,
+      "step": 2794
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.479563982961572e-06,
+      "loss": 2.3332,
+      "step": 2795
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.478310404165756e-06,
+      "loss": 2.2551,
+      "step": 2796
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.477056401548364e-06,
+      "loss": 2.3353,
+      "step": 2797
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.4758019752622e-06,
+      "loss": 2.3312,
+      "step": 2798
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.474547125460108e-06,
+      "loss": 2.301,
+      "step": 2799
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.473291852294986e-06,
+      "loss": 2.2306,
+      "step": 2800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.47203615591979e-06,
+      "loss": 2.3054,
+      "step": 2801
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.47078003648752e-06,
+      "loss": 2.3151,
+      "step": 2802
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.469523494151229e-06,
+      "loss": 2.3325,
+      "step": 2803
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.468266529064025e-06,
+      "loss": 2.3829,
+      "step": 2804
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.467009141379065e-06,
+      "loss": 2.2797,
+      "step": 2805
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.465751331249558e-06,
+      "loss": 2.3089,
+      "step": 2806
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.464493098828763e-06,
+      "loss": 2.354,
+      "step": 2807
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.463234444269994e-06,
+      "loss": 2.3449,
+      "step": 2808
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.461975367726614e-06,
+      "loss": 2.3389,
+      "step": 2809
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.460715869352035e-06,
+      "loss": 2.361,
+      "step": 2810
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.459455949299725e-06,
+      "loss": 2.2684,
+      "step": 2811
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.458195607723201e-06,
+      "loss": 2.2045,
+      "step": 2812
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.456934844776033e-06,
+      "loss": 2.2908,
+      "step": 2813
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.45567366061184e-06,
+      "loss": 2.2651,
+      "step": 2814
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.454412055384292e-06,
+      "loss": 2.3138,
+      "step": 2815
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.453150029247115e-06,
+      "loss": 2.3096,
+      "step": 2816
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.451887582354081e-06,
+      "loss": 2.2855,
+      "step": 2817
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.450624714859016e-06,
+      "loss": 2.3745,
+      "step": 2818
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.449361426915797e-06,
+      "loss": 2.2923,
+      "step": 2819
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.44809771867835e-06,
+      "loss": 2.3425,
+      "step": 2820
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.446833590300656e-06,
+      "loss": 2.3476,
+      "step": 2821
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.445569041936743e-06,
+      "loss": 2.3159,
+      "step": 2822
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.444304073740695e-06,
+      "loss": 2.3604,
+      "step": 2823
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.443038685866643e-06,
+      "loss": 2.3347,
+      "step": 2824
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.44177287846877e-06,
+      "loss": 2.2997,
+      "step": 2825
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.440506651701315e-06,
+      "loss": 2.3637,
+      "step": 2826
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.439240005718556e-06,
+      "loss": 2.326,
+      "step": 2827
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.437972940674838e-06,
+      "loss": 2.316,
+      "step": 2828
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.436705456724545e-06,
+      "loss": 2.3121,
+      "step": 2829
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.435437554022116e-06,
+      "loss": 2.3025,
+      "step": 2830
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.434169232722043e-06,
+      "loss": 2.3437,
+      "step": 2831
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.432900492978864e-06,
+      "loss": 2.2926,
+      "step": 2832
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.431631334947175e-06,
+      "loss": 2.4155,
+      "step": 2833
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.430361758781616e-06,
+      "loss": 2.3004,
+      "step": 2834
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.429091764636883e-06,
+      "loss": 2.3766,
+      "step": 2835
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.427821352667719e-06,
+      "loss": 2.2686,
+      "step": 2836
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.42655052302892e-06,
+      "loss": 2.3105,
+      "step": 2837
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.425279275875336e-06,
+      "loss": 2.2965,
+      "step": 2838
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.424007611361861e-06,
+      "loss": 2.3746,
+      "step": 2839
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.422735529643445e-06,
+      "loss": 2.2491,
+      "step": 2840
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.421463030875086e-06,
+      "loss": 2.3355,
+      "step": 2841
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.420190115211835e-06,
+      "loss": 2.3484,
+      "step": 2842
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.418916782808795e-06,
+      "loss": 2.321,
+      "step": 2843
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.417643033821114e-06,
+      "loss": 2.3668,
+      "step": 2844
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.416368868403997e-06,
+      "loss": 2.2882,
+      "step": 2845
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.415094286712694e-06,
+      "loss": 2.355,
+      "step": 2846
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.413819288902514e-06,
+      "loss": 2.2991,
+      "step": 2847
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.412543875128809e-06,
+      "loss": 2.269,
+      "step": 2848
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.411268045546984e-06,
+      "loss": 2.2881,
+      "step": 2849
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.409991800312493e-06,
+      "loss": 2.2652,
+      "step": 2850
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.408715139580846e-06,
+      "loss": 2.3458,
+      "step": 2851
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.4074380635076e-06,
+      "loss": 2.3999,
+      "step": 2852
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.406160572248361e-06,
+      "loss": 2.3285,
+      "step": 2853
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.404882665958788e-06,
+      "loss": 2.3144,
+      "step": 2854
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.40360434479459e-06,
+      "loss": 2.3381,
+      "step": 2855
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.402325608911527e-06,
+      "loss": 2.3165,
+      "step": 2856
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.40104645846541e-06,
+      "loss": 2.2274,
+      "step": 2857
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.399766893612096e-06,
+      "loss": 2.3341,
+      "step": 2858
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.398486914507501e-06,
+      "loss": 2.3553,
+      "step": 2859
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.397206521307584e-06,
+      "loss": 2.3318,
+      "step": 2860
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.395925714168356e-06,
+      "loss": 2.3066,
+      "step": 2861
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.394644493245882e-06,
+      "loss": 2.3066,
+      "step": 2862
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.393362858696272e-06,
+      "loss": 2.3886,
+      "step": 2863
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.392080810675692e-06,
+      "loss": 2.3347,
+      "step": 2864
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.390798349340354e-06,
+      "loss": 2.317,
+      "step": 2865
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.389515474846522e-06,
+      "loss": 2.23,
+      "step": 2866
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.388232187350513e-06,
+      "loss": 2.2987,
+      "step": 2867
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.386948487008687e-06,
+      "loss": 2.2813,
+      "step": 2868
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.385664373977462e-06,
+      "loss": 2.3758,
+      "step": 2869
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.384379848413304e-06,
+      "loss": 2.2368,
+      "step": 2870
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.383094910472728e-06,
+      "loss": 2.2233,
+      "step": 2871
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.381809560312298e-06,
+      "loss": 2.2783,
+      "step": 2872
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.38052379808863e-06,
+      "loss": 2.3132,
+      "step": 2873
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.379237623958393e-06,
+      "loss": 2.3736,
+      "step": 2874
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.377951038078303e-06,
+      "loss": 2.3208,
+      "step": 2875
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.376664040605122e-06,
+      "loss": 2.3279,
+      "step": 2876
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.375376631695673e-06,
+      "loss": 2.3185,
+      "step": 2877
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.374088811506819e-06,
+      "loss": 2.2969,
+      "step": 2878
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.372800580195479e-06,
+      "loss": 2.2909,
+      "step": 2879
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.371511937918616e-06,
+      "loss": 2.313,
+      "step": 2880
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.370222884833255e-06,
+      "loss": 2.3594,
+      "step": 2881
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.368933421096454e-06,
+      "loss": 2.2847,
+      "step": 2882
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.367643546865339e-06,
+      "loss": 2.3464,
+      "step": 2883
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.366353262297069e-06,
+      "loss": 2.3548,
+      "step": 2884
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.365062567548868e-06,
+      "loss": 2.322,
+      "step": 2885
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.363771462778e-06,
+      "loss": 2.3327,
+      "step": 2886
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.36247994814178e-06,
+      "loss": 2.2839,
+      "step": 2887
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.361188023797581e-06,
+      "loss": 2.3577,
+      "step": 2888
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.359895689902815e-06,
+      "loss": 2.2811,
+      "step": 2889
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.358602946614952e-06,
+      "loss": 2.2922,
+      "step": 2890
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.357309794091508e-06,
+      "loss": 2.3284,
+      "step": 2891
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.356016232490047e-06,
+      "loss": 2.291,
+      "step": 2892
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.35472226196819e-06,
+      "loss": 2.3154,
+      "step": 2893
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.353427882683601e-06,
+      "loss": 2.2974,
+      "step": 2894
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.352133094793996e-06,
+      "loss": 2.2834,
+      "step": 2895
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.350837898457142e-06,
+      "loss": 2.3447,
+      "step": 2896
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.349542293830854e-06,
+      "loss": 2.3229,
+      "step": 2897
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.348246281072998e-06,
+      "loss": 2.3226,
+      "step": 2898
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.346949860341489e-06,
+      "loss": 2.2928,
+      "step": 2899
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.345653031794292e-06,
+      "loss": 2.314,
+      "step": 2900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.34435579558942e-06,
+      "loss": 2.3468,
+      "step": 2901
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.343058151884942e-06,
+      "loss": 2.2387,
+      "step": 2902
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.341760100838967e-06,
+      "loss": 2.369,
+      "step": 2903
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.34046164260966e-06,
+      "loss": 2.4054,
+      "step": 2904
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.339162777355233e-06,
+      "loss": 2.3004,
+      "step": 2905
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.337863505233954e-06,
+      "loss": 2.3488,
+      "step": 2906
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.336563826404129e-06,
+      "loss": 2.3223,
+      "step": 2907
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.335263741024123e-06,
+      "loss": 2.2852,
+      "step": 2908
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.333963249252347e-06,
+      "loss": 2.3177,
+      "step": 2909
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.332662351247262e-06,
+      "loss": 2.2877,
+      "step": 2910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.33136104716738e-06,
+      "loss": 2.3104,
+      "step": 2911
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.33005933717126e-06,
+      "loss": 2.3286,
+      "step": 2912
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.328757221417507e-06,
+      "loss": 2.3229,
+      "step": 2913
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.327454700064788e-06,
+      "loss": 2.3472,
+      "step": 2914
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.326151773271805e-06,
+      "loss": 2.2967,
+      "step": 2915
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.324848441197317e-06,
+      "loss": 2.2942,
+      "step": 2916
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.323544704000134e-06,
+      "loss": 2.3333,
+      "step": 2917
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.32224056183911e-06,
+      "loss": 2.3452,
+      "step": 2918
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.320936014873148e-06,
+      "loss": 2.3806,
+      "step": 2919
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.319631063261209e-06,
+      "loss": 2.3002,
+      "step": 2920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.318325707162293e-06,
+      "loss": 2.3417,
+      "step": 2921
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.317019946735456e-06,
+      "loss": 2.2706,
+      "step": 2922
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.315713782139801e-06,
+      "loss": 2.358,
+      "step": 2923
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.314407213534477e-06,
+      "loss": 2.2871,
+      "step": 2924
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.313100241078689e-06,
+      "loss": 2.3216,
+      "step": 2925
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.311792864931686e-06,
+      "loss": 2.2736,
+      "step": 2926
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.310485085252767e-06,
+      "loss": 2.2465,
+      "step": 2927
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.309176902201283e-06,
+      "loss": 2.3954,
+      "step": 2928
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.307868315936631e-06,
+      "loss": 2.3517,
+      "step": 2929
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.30655932661826e-06,
+      "loss": 2.3166,
+      "step": 2930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.305249934405663e-06,
+      "loss": 2.2827,
+      "step": 2931
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.303940139458389e-06,
+      "loss": 2.3329,
+      "step": 2932
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.302629941936032e-06,
+      "loss": 2.2638,
+      "step": 2933
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.301319341998231e-06,
+      "loss": 2.304,
+      "step": 2934
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.300008339804686e-06,
+      "loss": 2.3033,
+      "step": 2935
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.298696935515132e-06,
+      "loss": 2.2676,
+      "step": 2936
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.297385129289361e-06,
+      "loss": 2.2693,
+      "step": 2937
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.296072921287217e-06,
+      "loss": 2.2508,
+      "step": 2938
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.294760311668586e-06,
+      "loss": 2.2852,
+      "step": 2939
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.293447300593402e-06,
+      "loss": 2.2848,
+      "step": 2940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.292133888221659e-06,
+      "loss": 2.2916,
+      "step": 2941
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.290820074713383e-06,
+      "loss": 2.3879,
+      "step": 2942
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.289505860228666e-06,
+      "loss": 2.2845,
+      "step": 2943
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.288191244927637e-06,
+      "loss": 2.3704,
+      "step": 2944
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.28687622897048e-06,
+      "loss": 2.3979,
+      "step": 2945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.285560812517423e-06,
+      "loss": 2.2921,
+      "step": 2946
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.284244995728749e-06,
+      "loss": 2.3194,
+      "step": 2947
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.282928778764783e-06,
+      "loss": 2.3324,
+      "step": 2948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.281612161785903e-06,
+      "loss": 2.2653,
+      "step": 2949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.280295144952537e-06,
+      "loss": 2.3079,
+      "step": 2950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.278977728425157e-06,
+      "loss": 2.2848,
+      "step": 2951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.277659912364288e-06,
+      "loss": 2.3717,
+      "step": 2952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.276341696930499e-06,
+      "loss": 2.3037,
+      "step": 2953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.275023082284413e-06,
+      "loss": 2.3482,
+      "step": 2954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.273704068586698e-06,
+      "loss": 2.3275,
+      "step": 2955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.272384655998075e-06,
+      "loss": 2.375,
+      "step": 2956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.271064844679306e-06,
+      "loss": 2.3384,
+      "step": 2957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.269744634791207e-06,
+      "loss": 2.3179,
+      "step": 2958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.268424026494646e-06,
+      "loss": 2.2706,
+      "step": 2959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.267103019950529e-06,
+      "loss": 2.1833,
+      "step": 2960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.26578161531982e-06,
+      "loss": 2.3068,
+      "step": 2961
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.264459812763525e-06,
+      "loss": 2.264,
+      "step": 2962
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.263137612442705e-06,
+      "loss": 2.3297,
+      "step": 2963
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.261815014518465e-06,
+      "loss": 2.329,
+      "step": 2964
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.260492019151962e-06,
+      "loss": 2.3143,
+      "step": 2965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.259168626504395e-06,
+      "loss": 2.3689,
+      "step": 2966
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.257844836737017e-06,
+      "loss": 2.2705,
+      "step": 2967
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.256520650011126e-06,
+      "loss": 2.2967,
+      "step": 2968
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.255196066488074e-06,
+      "loss": 2.3883,
+      "step": 2969
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.253871086329255e-06,
+      "loss": 2.3181,
+      "step": 2970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.252545709696115e-06,
+      "loss": 2.3103,
+      "step": 2971
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.251219936750145e-06,
+      "loss": 2.2563,
+      "step": 2972
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.249893767652888e-06,
+      "loss": 2.2715,
+      "step": 2973
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.248567202565934e-06,
+      "loss": 2.2671,
+      "step": 2974
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.247240241650918e-06,
+      "loss": 2.3653,
+      "step": 2975
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.24591288506953e-06,
+      "loss": 2.2034,
+      "step": 2976
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.244585132983505e-06,
+      "loss": 2.2684,
+      "step": 2977
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.243256985554622e-06,
+      "loss": 2.3368,
+      "step": 2978
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.24192844294471e-06,
+      "loss": 2.3291,
+      "step": 2979
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.240599505315656e-06,
+      "loss": 2.293,
+      "step": 2980
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.239270172829379e-06,
+      "loss": 2.3371,
+      "step": 2981
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.237940445647858e-06,
+      "loss": 2.2815,
+      "step": 2982
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.236610323933115e-06,
+      "loss": 2.2527,
+      "step": 2983
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.235279807847223e-06,
+      "loss": 2.2909,
+      "step": 2984
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.2339488975523e-06,
+      "loss": 2.2987,
+      "step": 2985
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.232617593210512e-06,
+      "loss": 2.2541,
+      "step": 2986
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.231285894984076e-06,
+      "loss": 2.3434,
+      "step": 2987
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.229953803035256e-06,
+      "loss": 2.315,
+      "step": 2988
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.228621317526362e-06,
+      "loss": 2.3513,
+      "step": 2989
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.227288438619754e-06,
+      "loss": 2.3136,
+      "step": 2990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.22595516647784e-06,
+      "loss": 2.325,
+      "step": 2991
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.224621501263073e-06,
+      "loss": 2.3196,
+      "step": 2992
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.223287443137957e-06,
+      "loss": 2.4052,
+      "step": 2993
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.221952992265046e-06,
+      "loss": 2.3137,
+      "step": 2994
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.220618148806934e-06,
+      "loss": 2.387,
+      "step": 2995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.21928291292627e-06,
+      "loss": 2.3247,
+      "step": 2996
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.217947284785748e-06,
+      "loss": 2.259,
+      "step": 2997
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.21661126454811e-06,
+      "loss": 2.3555,
+      "step": 2998
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.215274852376148e-06,
+      "loss": 2.2975,
+      "step": 2999
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.213938048432697e-06,
+      "loss": 2.3187,
+      "step": 3000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.212600852880644e-06,
+      "loss": 2.2504,
+      "step": 3001
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.211263265882923e-06,
+      "loss": 2.2722,
+      "step": 3002
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.209925287602513e-06,
+      "loss": 2.2957,
+      "step": 3003
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.208586918202444e-06,
+      "loss": 2.3131,
+      "step": 3004
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.20724815784579e-06,
+      "loss": 2.3265,
+      "step": 3005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.205909006695679e-06,
+      "loss": 2.328,
+      "step": 3006
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.204569464915278e-06,
+      "loss": 2.2542,
+      "step": 3007
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.203229532667808e-06,
+      "loss": 2.3591,
+      "step": 3008
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.201889210116536e-06,
+      "loss": 2.2917,
+      "step": 3009
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.200548497424779e-06,
+      "loss": 2.3055,
+      "step": 3010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.199207394755892e-06,
+      "loss": 2.3388,
+      "step": 3011
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.197865902273291e-06,
+      "loss": 2.3221,
+      "step": 3012
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.196524020140428e-06,
+      "loss": 2.3171,
+      "step": 3013
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.19518174852081e-06,
+      "loss": 2.2868,
+      "step": 3014
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.19383908757799e-06,
+      "loss": 2.2829,
+      "step": 3015
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.192496037475562e-06,
+      "loss": 2.2393,
+      "step": 3016
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.191152598377179e-06,
+      "loss": 2.3629,
+      "step": 3017
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.189808770446528e-06,
+      "loss": 2.3029,
+      "step": 3018
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.188464553847356e-06,
+      "loss": 2.3042,
+      "step": 3019
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.18711994874345e-06,
+      "loss": 2.3722,
+      "step": 3020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.185774955298645e-06,
+      "loss": 2.2299,
+      "step": 3021
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.184429573676825e-06,
+      "loss": 2.3227,
+      "step": 3022
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.183083804041922e-06,
+      "loss": 2.3459,
+      "step": 3023
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.181737646557912e-06,
+      "loss": 2.2904,
+      "step": 3024
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.18039110138882e-06,
+      "loss": 2.3735,
+      "step": 3025
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.179044168698722e-06,
+      "loss": 2.3371,
+      "step": 3026
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.177696848651733e-06,
+      "loss": 2.2744,
+      "step": 3027
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.176349141412022e-06,
+      "loss": 2.3261,
+      "step": 3028
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.175001047143804e-06,
+      "loss": 2.222,
+      "step": 3029
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.173652566011339e-06,
+      "loss": 2.3417,
+      "step": 3030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.172303698178934e-06,
+      "loss": 2.3448,
+      "step": 3031
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.170954443810947e-06,
+      "loss": 2.3117,
+      "step": 3032
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.169604803071783e-06,
+      "loss": 2.3883,
+      "step": 3033
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.168254776125883e-06,
+      "loss": 2.3263,
+      "step": 3034
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.166904363137752e-06,
+      "loss": 2.3244,
+      "step": 3035
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.165553564271928e-06,
+      "loss": 2.303,
+      "step": 3036
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.164202379693008e-06,
+      "loss": 2.3739,
+      "step": 3037
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.162850809565623e-06,
+      "loss": 2.2041,
+      "step": 3038
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.161498854054462e-06,
+      "loss": 2.2998,
+      "step": 3039
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.160146513324256e-06,
+      "loss": 2.2733,
+      "step": 3040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.158793787539782e-06,
+      "loss": 2.2817,
+      "step": 3041
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.157440676865866e-06,
+      "loss": 2.2949,
+      "step": 3042
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.156087181467382e-06,
+      "loss": 2.3383,
+      "step": 3043
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.154733301509249e-06,
+      "loss": 2.3358,
+      "step": 3044
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.153379037156433e-06,
+      "loss": 2.2757,
+      "step": 3045
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.152024388573945e-06,
+      "loss": 2.3084,
+      "step": 3046
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.150669355926848e-06,
+      "loss": 2.302,
+      "step": 3047
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.149313939380244e-06,
+      "loss": 2.3019,
+      "step": 3048
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.147958139099292e-06,
+      "loss": 2.3098,
+      "step": 3049
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.146601955249187e-06,
+      "loss": 2.3445,
+      "step": 3050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.14524538799518e-06,
+      "loss": 2.3014,
+      "step": 3051
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.143888437502565e-06,
+      "loss": 2.3183,
+      "step": 3052
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.142531103936677e-06,
+      "loss": 2.2831,
+      "step": 3053
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.141173387462908e-06,
+      "loss": 2.2773,
+      "step": 3054
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.139815288246692e-06,
+      "loss": 2.2237,
+      "step": 3055
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.138456806453503e-06,
+      "loss": 2.2973,
+      "step": 3056
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.137097942248875e-06,
+      "loss": 2.2853,
+      "step": 3057
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.135738695798377e-06,
+      "loss": 2.2755,
+      "step": 3058
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.13437906726763e-06,
+      "loss": 2.3297,
+      "step": 3059
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.133019056822303e-06,
+      "loss": 2.2924,
+      "step": 3060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.131658664628108e-06,
+      "loss": 2.2841,
+      "step": 3061
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.130297890850803e-06,
+      "loss": 2.3774,
+      "step": 3062
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.128936735656195e-06,
+      "loss": 2.3515,
+      "step": 3063
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.127575199210136e-06,
+      "loss": 2.3045,
+      "step": 3064
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.126213281678527e-06,
+      "loss": 2.2908,
+      "step": 3065
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.124850983227313e-06,
+      "loss": 2.2898,
+      "step": 3066
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.123488304022485e-06,
+      "loss": 2.3355,
+      "step": 3067
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.12212524423008e-06,
+      "loss": 2.3286,
+      "step": 3068
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.120761804016186e-06,
+      "loss": 2.2988,
+      "step": 3069
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.119397983546932e-06,
+      "loss": 2.3081,
+      "step": 3070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.118033782988496e-06,
+      "loss": 2.28,
+      "step": 3071
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.116669202507102e-06,
+      "loss": 2.3764,
+      "step": 3072
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.115304242269017e-06,
+      "loss": 2.2955,
+      "step": 3073
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.113938902440563e-06,
+      "loss": 2.3288,
+      "step": 3074
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.112573183188099e-06,
+      "loss": 2.3191,
+      "step": 3075
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.111207084678033e-06,
+      "loss": 2.3327,
+      "step": 3076
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.10984060707682e-06,
+      "loss": 2.2907,
+      "step": 3077
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.108473750550965e-06,
+      "loss": 2.3281,
+      "step": 3078
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.10710651526701e-06,
+      "loss": 2.3004,
+      "step": 3079
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.105738901391553e-06,
+      "loss": 2.3254,
+      "step": 3080
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.10437090909123e-06,
+      "loss": 2.2516,
+      "step": 3081
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.103002538532729e-06,
+      "loss": 2.307,
+      "step": 3082
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.101633789882781e-06,
+      "loss": 2.2756,
+      "step": 3083
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.100264663308165e-06,
+      "loss": 2.3328,
+      "step": 3084
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.098895158975702e-06,
+      "loss": 2.3336,
+      "step": 3085
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.097525277052265e-06,
+      "loss": 2.3065,
+      "step": 3086
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.096155017704768e-06,
+      "loss": 2.3277,
+      "step": 3087
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.094784381100174e-06,
+      "loss": 2.2811,
+      "step": 3088
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.09341336740549e-06,
+      "loss": 2.2916,
+      "step": 3089
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.092041976787772e-06,
+      "loss": 2.2502,
+      "step": 3090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.090670209414117e-06,
+      "loss": 2.3385,
+      "step": 3091
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.089298065451673e-06,
+      "loss": 2.3007,
+      "step": 3092
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.08792554506763e-06,
+      "loss": 2.3384,
+      "step": 3093
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.086552648429225e-06,
+      "loss": 2.2795,
+      "step": 3094
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.085179375703745e-06,
+      "loss": 2.3671,
+      "step": 3095
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.083805727058514e-06,
+      "loss": 2.2954,
+      "step": 3096
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.08243170266091e-06,
+      "loss": 2.2761,
+      "step": 3097
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.081057302678352e-06,
+      "loss": 2.2595,
+      "step": 3098
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.07968252727831e-06,
+      "loss": 2.3261,
+      "step": 3099
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.078307376628292e-06,
+      "loss": 2.3339,
+      "step": 3100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.076931850895858e-06,
+      "loss": 2.3643,
+      "step": 3101
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.075555950248613e-06,
+      "loss": 2.276,
+      "step": 3102
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.0741796748542e-06,
+      "loss": 2.3168,
+      "step": 3103
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.072803024880322e-06,
+      "loss": 2.2727,
+      "step": 3104
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.071426000494716e-06,
+      "loss": 2.2937,
+      "step": 3105
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.07004860186517e-06,
+      "loss": 2.2226,
+      "step": 3106
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.068670829159511e-06,
+      "loss": 2.2375,
+      "step": 3107
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.067292682545622e-06,
+      "loss": 2.278,
+      "step": 3108
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.065914162191424e-06,
+      "loss": 2.2664,
+      "step": 3109
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.064535268264883e-06,
+      "loss": 2.2924,
+      "step": 3110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.063156000934017e-06,
+      "loss": 2.2773,
+      "step": 3111
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.061776360366883e-06,
+      "loss": 2.3535,
+      "step": 3112
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.060396346731587e-06,
+      "loss": 2.324,
+      "step": 3113
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.05901596019628e-06,
+      "loss": 2.3006,
+      "step": 3114
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.057635200929157e-06,
+      "loss": 2.3111,
+      "step": 3115
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.05625406909846e-06,
+      "loss": 2.3026,
+      "step": 3116
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.054872564872474e-06,
+      "loss": 2.3363,
+      "step": 3117
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.053490688419532e-06,
+      "loss": 2.2255,
+      "step": 3118
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.052108439908014e-06,
+      "loss": 2.3503,
+      "step": 3119
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.05072581950634e-06,
+      "loss": 2.2499,
+      "step": 3120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.049342827382978e-06,
+      "loss": 2.2941,
+      "step": 3121
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.047959463706441e-06,
+      "loss": 2.3473,
+      "step": 3122
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.04657572864529e-06,
+      "loss": 2.2949,
+      "step": 3123
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.045191622368128e-06,
+      "loss": 2.277,
+      "step": 3124
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.043807145043604e-06,
+      "loss": 2.2811,
+      "step": 3125
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.04242229684041e-06,
+      "loss": 2.3038,
+      "step": 3126
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.041037077927291e-06,
+      "loss": 2.3701,
+      "step": 3127
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.039651488473028e-06,
+      "loss": 2.3136,
+      "step": 3128
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.03826552864645e-06,
+      "loss": 2.2988,
+      "step": 3129
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.036879198616434e-06,
+      "loss": 2.2684,
+      "step": 3130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.0354924985519e-06,
+      "loss": 2.3736,
+      "step": 3131
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.034105428621812e-06,
+      "loss": 2.252,
+      "step": 3132
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.03271798899518e-06,
+      "loss": 2.2406,
+      "step": 3133
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.031330179841062e-06,
+      "loss": 2.2767,
+      "step": 3134
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.029942001328555e-06,
+      "loss": 2.3027,
+      "step": 3135
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.028553453626809e-06,
+      "loss": 2.1908,
+      "step": 3136
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.027164536905008e-06,
+      "loss": 2.2663,
+      "step": 3137
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.02577525133239e-06,
+      "loss": 2.2658,
+      "step": 3138
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.024385597078239e-06,
+      "loss": 2.2538,
+      "step": 3139
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.022995574311876e-06,
+      "loss": 2.3514,
+      "step": 3140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.021605183202669e-06,
+      "loss": 2.2484,
+      "step": 3141
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.020214423920039e-06,
+      "loss": 2.3045,
+      "step": 3142
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.018823296633442e-06,
+      "loss": 2.2703,
+      "step": 3143
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.017431801512384e-06,
+      "loss": 2.294,
+      "step": 3144
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.016039938726413e-06,
+      "loss": 2.2851,
+      "step": 3145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.014647708445124e-06,
+      "loss": 2.2856,
+      "step": 3146
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.013255110838156e-06,
+      "loss": 2.3004,
+      "step": 3147
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.011862146075194e-06,
+      "loss": 2.3547,
+      "step": 3148
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.010468814325964e-06,
+      "loss": 2.2538,
+      "step": 3149
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.009075115760243e-06,
+      "loss": 2.3196,
+      "step": 3150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.007681050547844e-06,
+      "loss": 2.3547,
+      "step": 3151
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.006286618858634e-06,
+      "loss": 2.2617,
+      "step": 3152
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.004891820862516e-06,
+      "loss": 2.318,
+      "step": 3153
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.003496656729448e-06,
+      "loss": 2.3163,
+      "step": 3154
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.002101126629422e-06,
+      "loss": 2.3676,
+      "step": 3155
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.000705230732478e-06,
+      "loss": 2.2502,
+      "step": 3156
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.999308969208705e-06,
+      "loss": 2.277,
+      "step": 3157
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.997912342228232e-06,
+      "loss": 2.2772,
+      "step": 3158
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.996515349961233e-06,
+      "loss": 2.2234,
+      "step": 3159
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.99511799257793e-06,
+      "loss": 2.3313,
+      "step": 3160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.993720270248583e-06,
+      "loss": 2.3177,
+      "step": 3161
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.992322183143504e-06,
+      "loss": 2.2348,
+      "step": 3162
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.990923731433043e-06,
+      "loss": 2.2908,
+      "step": 3163
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.989524915287595e-06,
+      "loss": 2.2942,
+      "step": 3164
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.988125734877607e-06,
+      "loss": 2.2823,
+      "step": 3165
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.986726190373562e-06,
+      "loss": 2.2985,
+      "step": 3166
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.985326281945988e-06,
+      "loss": 2.3227,
+      "step": 3167
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.983926009765464e-06,
+      "loss": 2.3029,
+      "step": 3168
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.982525374002607e-06,
+      "loss": 2.2828,
+      "step": 3169
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.981124374828079e-06,
+      "loss": 2.3947,
+      "step": 3170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.97972301241259e-06,
+      "loss": 2.3348,
+      "step": 3171
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.978321286926892e-06,
+      "loss": 2.2587,
+      "step": 3172
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.976919198541775e-06,
+      "loss": 2.3266,
+      "step": 3173
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.975516747428087e-06,
+      "loss": 2.2829,
+      "step": 3174
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.974113933756708e-06,
+      "loss": 2.2844,
+      "step": 3175
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.972710757698567e-06,
+      "loss": 2.3185,
+      "step": 3176
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.971307219424637e-06,
+      "loss": 2.31,
+      "step": 3177
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.969903319105935e-06,
+      "loss": 2.361,
+      "step": 3178
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.968499056913525e-06,
+      "loss": 2.3618,
+      "step": 3179
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.967094433018508e-06,
+      "loss": 2.2723,
+      "step": 3180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.965689447592034e-06,
+      "loss": 2.3156,
+      "step": 3181
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.964284100805297e-06,
+      "loss": 2.2873,
+      "step": 3182
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.962878392829533e-06,
+      "loss": 2.3168,
+      "step": 3183
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.961472323836025e-06,
+      "loss": 2.3801,
+      "step": 3184
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.960065893996099e-06,
+      "loss": 2.3344,
+      "step": 3185
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.95865910348112e-06,
+      "loss": 2.3149,
+      "step": 3186
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.957251952462506e-06,
+      "loss": 2.3428,
+      "step": 3187
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.95584444111171e-06,
+      "loss": 2.2678,
+      "step": 3188
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.954436569600238e-06,
+      "loss": 2.2444,
+      "step": 3189
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.953028338099628e-06,
+      "loss": 2.3142,
+      "step": 3190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.951619746781474e-06,
+      "loss": 2.3207,
+      "step": 3191
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.950210795817406e-06,
+      "loss": 2.2807,
+      "step": 3192
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.948801485379103e-06,
+      "loss": 2.2095,
+      "step": 3193
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.947391815638284e-06,
+      "loss": 2.3355,
+      "step": 3194
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.945981786766712e-06,
+      "loss": 2.2698,
+      "step": 3195
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.944571398936193e-06,
+      "loss": 2.2966,
+      "step": 3196
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.943160652318585e-06,
+      "loss": 2.3186,
+      "step": 3197
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.941749547085778e-06,
+      "loss": 2.3378,
+      "step": 3198
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.94033808340971e-06,
+      "loss": 2.2581,
+      "step": 3199
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.938926261462366e-06,
+      "loss": 2.2837,
+      "step": 3200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.937514081415773e-06,
+      "loss": 2.3273,
+      "step": 3201
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.936101543441998e-06,
+      "loss": 2.2958,
+      "step": 3202
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.934688647713158e-06,
+      "loss": 2.3164,
+      "step": 3203
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.933275394401407e-06,
+      "loss": 2.3029,
+      "step": 3204
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.931861783678946e-06,
+      "loss": 2.245,
+      "step": 3205
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.930447815718022e-06,
+      "loss": 2.1961,
+      "step": 3206
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.929033490690921e-06,
+      "loss": 2.3019,
+      "step": 3207
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.927618808769971e-06,
+      "loss": 2.3417,
+      "step": 3208
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.926203770127552e-06,
+      "loss": 2.2824,
+      "step": 3209
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.92478837493608e-06,
+      "loss": 2.3449,
+      "step": 3210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.923372623368014e-06,
+      "loss": 2.2593,
+      "step": 3211
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.921956515595861e-06,
+      "loss": 2.3006,
+      "step": 3212
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.920540051792171e-06,
+      "loss": 2.301,
+      "step": 3213
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.919123232129535e-06,
+      "loss": 2.2768,
+      "step": 3214
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.917706056780588e-06,
+      "loss": 2.2854,
+      "step": 3215
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.916288525918008e-06,
+      "loss": 2.2514,
+      "step": 3216
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.914870639714517e-06,
+      "loss": 2.3022,
+      "step": 3217
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.913452398342882e-06,
+      "loss": 2.2903,
+      "step": 3218
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.912033801975907e-06,
+      "loss": 2.3321,
+      "step": 3219
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.910614850786448e-06,
+      "loss": 2.2531,
+      "step": 3220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.909195544947398e-06,
+      "loss": 2.3402,
+      "step": 3221
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.907775884631694e-06,
+      "loss": 2.2136,
+      "step": 3222
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.90635587001232e-06,
+      "loss": 2.2511,
+      "step": 3223
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.904935501262301e-06,
+      "loss": 2.2918,
+      "step": 3224
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.903514778554699e-06,
+      "loss": 2.3265,
+      "step": 3225
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.90209370206263e-06,
+      "loss": 2.3568,
+      "step": 3226
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.900672271959247e-06,
+      "loss": 2.2732,
+      "step": 3227
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.899250488417746e-06,
+      "loss": 2.3045,
+      "step": 3228
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.897828351611368e-06,
+      "loss": 2.3487,
+      "step": 3229
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.896405861713393e-06,
+      "loss": 2.301,
+      "step": 3230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.894983018897153e-06,
+      "loss": 2.2621,
+      "step": 3231
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.893559823336013e-06,
+      "loss": 2.2929,
+      "step": 3232
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.892136275203383e-06,
+      "loss": 2.3512,
+      "step": 3233
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.890712374672724e-06,
+      "loss": 2.3048,
+      "step": 3234
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.889288121917528e-06,
+      "loss": 2.2958,
+      "step": 3235
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.887863517111337e-06,
+      "loss": 2.3358,
+      "step": 3236
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.88643856042774e-06,
+      "loss": 2.3344,
+      "step": 3237
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.88501325204036e-06,
+      "loss": 2.2787,
+      "step": 3238
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.883587592122864e-06,
+      "loss": 2.2441,
+      "step": 3239
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.882161580848966e-06,
+      "loss": 2.2284,
+      "step": 3240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.880735218392424e-06,
+      "loss": 2.2812,
+      "step": 3241
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.879308504927034e-06,
+      "loss": 2.3359,
+      "step": 3242
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.877881440626635e-06,
+      "loss": 2.2859,
+      "step": 3243
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.876454025665114e-06,
+      "loss": 2.2768,
+      "step": 3244
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.875026260216395e-06,
+      "loss": 2.3007,
+      "step": 3245
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.873598144454444e-06,
+      "loss": 2.2851,
+      "step": 3246
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.872169678553279e-06,
+      "loss": 2.2543,
+      "step": 3247
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.87074086268695e-06,
+      "loss": 2.2748,
+      "step": 3248
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.869311697029553e-06,
+      "loss": 2.2888,
+      "step": 3249
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.86788218175523e-06,
+      "loss": 2.302,
+      "step": 3250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.866452317038164e-06,
+      "loss": 2.2214,
+      "step": 3251
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.865022103052578e-06,
+      "loss": 2.3471,
+      "step": 3252
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.863591539972739e-06,
+      "loss": 2.258,
+      "step": 3253
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.862160627972956e-06,
+      "loss": 2.3255,
+      "step": 3254
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.860729367227582e-06,
+      "loss": 2.3367,
+      "step": 3255
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.859297757911013e-06,
+      "loss": 2.2139,
+      "step": 3256
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.857865800197684e-06,
+      "loss": 2.2938,
+      "step": 3257
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.856433494262078e-06,
+      "loss": 2.3119,
+      "step": 3258
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.855000840278715e-06,
+      "loss": 2.2886,
+      "step": 3259
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.85356783842216e-06,
+      "loss": 2.1962,
+      "step": 3260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.852134488867017e-06,
+      "loss": 2.2504,
+      "step": 3261
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.850700791787941e-06,
+      "loss": 2.2836,
+      "step": 3262
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.849266747359619e-06,
+      "loss": 2.2538,
+      "step": 3263
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.847832355756788e-06,
+      "loss": 2.2377,
+      "step": 3264
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.846397617154223e-06,
+      "loss": 2.3419,
+      "step": 3265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.844962531726742e-06,
+      "loss": 2.1857,
+      "step": 3266
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.843527099649204e-06,
+      "loss": 2.2963,
+      "step": 3267
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.842091321096515e-06,
+      "loss": 2.1963,
+      "step": 3268
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.84065519624362e-06,
+      "loss": 2.3223,
+      "step": 3269
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.839218725265507e-06,
+      "loss": 2.325,
+      "step": 3270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.837781908337204e-06,
+      "loss": 2.3153,
+      "step": 3271
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.836344745633785e-06,
+      "loss": 2.3876,
+      "step": 3272
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.834907237330359e-06,
+      "loss": 2.2407,
+      "step": 3273
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.833469383602086e-06,
+      "loss": 2.2816,
+      "step": 3274
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.832031184624165e-06,
+      "loss": 2.2842,
+      "step": 3275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.830592640571833e-06,
+      "loss": 2.316,
+      "step": 3276
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.829153751620375e-06,
+      "loss": 2.2285,
+      "step": 3277
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.827714517945116e-06,
+      "loss": 2.2465,
+      "step": 3278
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.826274939721417e-06,
+      "loss": 2.2979,
+      "step": 3279
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.82483501712469e-06,
+      "loss": 2.346,
+      "step": 3280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.823394750330386e-06,
+      "loss": 2.2868,
+      "step": 3281
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.821954139513997e-06,
+      "loss": 2.2717,
+      "step": 3282
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.820513184851052e-06,
+      "loss": 2.2919,
+      "step": 3283
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.819071886517134e-06,
+      "loss": 2.2494,
+      "step": 3284
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.817630244687857e-06,
+      "loss": 2.2817,
+      "step": 3285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.816188259538885e-06,
+      "loss": 2.3041,
+      "step": 3286
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.814745931245911e-06,
+      "loss": 2.2958,
+      "step": 3287
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.813303259984685e-06,
+      "loss": 2.3237,
+      "step": 3288
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.81186024593099e-06,
+      "loss": 2.2998,
+      "step": 3289
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.810416889260653e-06,
+      "loss": 2.2764,
+      "step": 3290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.808973190149544e-06,
+      "loss": 2.3209,
+      "step": 3291
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.807529148773572e-06,
+      "loss": 2.3072,
+      "step": 3292
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.80608476530869e-06,
+      "loss": 2.3007,
+      "step": 3293
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.80464003993089e-06,
+      "loss": 2.2906,
+      "step": 3294
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.80319497281621e-06,
+      "loss": 2.2838,
+      "step": 3295
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.801749564140724e-06,
+      "loss": 2.2695,
+      "step": 3296
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.800303814080552e-06,
+      "loss": 2.4093,
+      "step": 3297
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.798857722811857e-06,
+      "loss": 2.2917,
+      "step": 3298
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.797411290510836e-06,
+      "loss": 2.3117,
+      "step": 3299
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.795964517353734e-06,
+      "loss": 2.3245,
+      "step": 3300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.79451740351684e-06,
+      "loss": 2.1834,
+      "step": 3301
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.793069949176474e-06,
+      "loss": 2.344,
+      "step": 3302
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.791622154509008e-06,
+      "loss": 2.2702,
+      "step": 3303
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.79017401969085e-06,
+      "loss": 2.3042,
+      "step": 3304
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.788725544898452e-06,
+      "loss": 2.3219,
+      "step": 3305
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.787276730308304e-06,
+      "loss": 2.2508,
+      "step": 3306
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.785827576096943e-06,
+      "loss": 2.321,
+      "step": 3307
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.78437808244094e-06,
+      "loss": 2.3212,
+      "step": 3308
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.782928249516915e-06,
+      "loss": 2.3055,
+      "step": 3309
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.781478077501526e-06,
+      "loss": 2.2417,
+      "step": 3310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.780027566571467e-06,
+      "loss": 2.3274,
+      "step": 3311
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.778576716903484e-06,
+      "loss": 2.3119,
+      "step": 3312
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.777125528674356e-06,
+      "loss": 2.3057,
+      "step": 3313
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.775674002060905e-06,
+      "loss": 2.28,
+      "step": 3314
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.774222137239998e-06,
+      "loss": 2.3189,
+      "step": 3315
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.772769934388537e-06,
+      "loss": 2.2039,
+      "step": 3316
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.771317393683471e-06,
+      "loss": 2.3106,
+      "step": 3317
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.769864515301787e-06,
+      "loss": 2.3115,
+      "step": 3318
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.768411299420513e-06,
+      "loss": 2.3313,
+      "step": 3319
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.76695774621672e-06,
+      "loss": 2.2125,
+      "step": 3320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.76550385586752e-06,
+      "loss": 2.322,
+      "step": 3321
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.764049628550063e-06,
+      "loss": 2.3435,
+      "step": 3322
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.762595064441542e-06,
+      "loss": 2.3427,
+      "step": 3323
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.761140163719194e-06,
+      "loss": 2.3174,
+      "step": 3324
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.759684926560292e-06,
+      "loss": 2.2713,
+      "step": 3325
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.758229353142153e-06,
+      "loss": 2.2817,
+      "step": 3326
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.756773443642132e-06,
+      "loss": 2.3183,
+      "step": 3327
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.755317198237631e-06,
+      "loss": 2.3116,
+      "step": 3328
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.753860617106085e-06,
+      "loss": 2.279,
+      "step": 3329
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.752403700424978e-06,
+      "loss": 2.3515,
+      "step": 3330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.75094644837183e-06,
+      "loss": 2.2755,
+      "step": 3331
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.7494888611242e-06,
+      "loss": 2.2816,
+      "step": 3332
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.748030938859692e-06,
+      "loss": 2.2795,
+      "step": 3333
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.74657268175595e-06,
+      "loss": 2.3202,
+      "step": 3334
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.74511408999066e-06,
+      "loss": 2.3429,
+      "step": 3335
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.743655163741544e-06,
+      "loss": 2.2609,
+      "step": 3336
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.742195903186366e-06,
+      "loss": 2.294,
+      "step": 3337
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.740736308502939e-06,
+      "loss": 2.323,
+      "step": 3338
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.739276379869105e-06,
+      "loss": 2.2824,
+      "step": 3339
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.737816117462752e-06,
+      "loss": 2.3059,
+      "step": 3340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.736355521461812e-06,
+      "loss": 2.3017,
+      "step": 3341
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.734894592044249e-06,
+      "loss": 2.2869,
+      "step": 3342
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.733433329388079e-06,
+      "loss": 2.3101,
+      "step": 3343
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.731971733671347e-06,
+      "loss": 2.2851,
+      "step": 3344
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.730509805072146e-06,
+      "loss": 2.2858,
+      "step": 3345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.729047543768608e-06,
+      "loss": 2.2663,
+      "step": 3346
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.727584949938907e-06,
+      "loss": 2.3476,
+      "step": 3347
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.726122023761252e-06,
+      "loss": 2.3162,
+      "step": 3348
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.724658765413897e-06,
+      "loss": 2.2905,
+      "step": 3349
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.723195175075136e-06,
+      "loss": 2.3289,
+      "step": 3350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.721731252923305e-06,
+      "loss": 2.2005,
+      "step": 3351
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.720266999136774e-06,
+      "loss": 2.3277,
+      "step": 3352
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.718802413893963e-06,
+      "loss": 2.3238,
+      "step": 3353
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.717337497373324e-06,
+      "loss": 2.3287,
+      "step": 3354
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.715872249753353e-06,
+      "loss": 2.3124,
+      "step": 3355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.714406671212589e-06,
+      "loss": 2.362,
+      "step": 3356
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.712940761929604e-06,
+      "loss": 2.3797,
+      "step": 3357
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.711474522083015e-06,
+      "loss": 2.3798,
+      "step": 3358
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.710007951851482e-06,
+      "loss": 2.2814,
+      "step": 3359
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.7085410514137e-06,
+      "loss": 2.327,
+      "step": 3360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.707073820948407e-06,
+      "loss": 2.3114,
+      "step": 3361
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.70560626063438e-06,
+      "loss": 2.3021,
+      "step": 3362
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.704138370650437e-06,
+      "loss": 2.2848,
+      "step": 3363
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.702670151175435e-06,
+      "loss": 2.2319,
+      "step": 3364
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.701201602388276e-06,
+      "loss": 2.2878,
+      "step": 3365
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.699732724467894e-06,
+      "loss": 2.2896,
+      "step": 3366
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.698263517593268e-06,
+      "loss": 2.313,
+      "step": 3367
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.696793981943418e-06,
+      "loss": 2.3408,
+      "step": 3368
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.695324117697401e-06,
+      "loss": 2.2288,
+      "step": 3369
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.693853925034316e-06,
+      "loss": 2.3178,
+      "step": 3370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.692383404133302e-06,
+      "loss": 2.3108,
+      "step": 3371
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.690912555173536e-06,
+      "loss": 2.1773,
+      "step": 3372
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.689441378334239e-06,
+      "loss": 2.3373,
+      "step": 3373
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.687969873794667e-06,
+      "loss": 2.2941,
+      "step": 3374
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.686498041734121e-06,
+      "loss": 2.2634,
+      "step": 3375
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.685025882331936e-06,
+      "loss": 2.279,
+      "step": 3376
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.683553395767492e-06,
+      "loss": 2.3241,
+      "step": 3377
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.682080582220206e-06,
+      "loss": 2.2203,
+      "step": 3378
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.680607441869538e-06,
+      "loss": 2.3367,
+      "step": 3379
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.679133974894984e-06,
+      "loss": 2.2475,
+      "step": 3380
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.67766018147608e-06,
+      "loss": 2.3068,
+      "step": 3381
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.676186061792408e-06,
+      "loss": 2.3327,
+      "step": 3382
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.67471161602358e-06,
+      "loss": 2.3009,
+      "step": 3383
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.673236844349257e-06,
+      "loss": 2.2773,
+      "step": 3384
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.671761746949133e-06,
+      "loss": 2.2475,
+      "step": 3385
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.670286324002943e-06,
+      "loss": 2.2405,
+      "step": 3386
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.668810575690465e-06,
+      "loss": 2.2252,
+      "step": 3387
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.667334502191514e-06,
+      "loss": 2.3048,
+      "step": 3388
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.665858103685944e-06,
+      "loss": 2.2881,
+      "step": 3389
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.66438138035365e-06,
+      "loss": 2.2411,
+      "step": 3390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.662904332374568e-06,
+      "loss": 2.3225,
+      "step": 3391
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.66142695992867e-06,
+      "loss": 2.2839,
+      "step": 3392
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.659949263195971e-06,
+      "loss": 2.3298,
+      "step": 3393
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.658471242356521e-06,
+      "loss": 2.3036,
+      "step": 3394
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.656992897590416e-06,
+      "loss": 2.2284,
+      "step": 3395
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.655514229077784e-06,
+      "loss": 2.3778,
+      "step": 3396
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.6540352369988e-06,
+      "loss": 2.2466,
+      "step": 3397
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.652555921533671e-06,
+      "loss": 2.2669,
+      "step": 3398
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.65107628286265e-06,
+      "loss": 2.3605,
+      "step": 3399
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.649596321166024e-06,
+      "loss": 2.2569,
+      "step": 3400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.648116036624125e-06,
+      "loss": 2.2796,
+      "step": 3401
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.646635429417322e-06,
+      "loss": 2.3226,
+      "step": 3402
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.645154499726017e-06,
+      "loss": 2.2923,
+      "step": 3403
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.64367324773066e-06,
+      "loss": 2.2804,
+      "step": 3404
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.642191673611737e-06,
+      "loss": 2.3094,
+      "step": 3405
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.640709777549773e-06,
+      "loss": 2.2471,
+      "step": 3406
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.639227559725333e-06,
+      "loss": 2.2509,
+      "step": 3407
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.637745020319019e-06,
+      "loss": 2.2647,
+      "step": 3408
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.636262159511479e-06,
+      "loss": 2.2803,
+      "step": 3409
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.634778977483389e-06,
+      "loss": 2.2894,
+      "step": 3410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.633295474415473e-06,
+      "loss": 2.3767,
+      "step": 3411
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.63181165048849e-06,
+      "loss": 2.2454,
+      "step": 3412
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.630327505883243e-06,
+      "loss": 2.2638,
+      "step": 3413
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.628843040780567e-06,
+      "loss": 2.3366,
+      "step": 3414
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.627358255361339e-06,
+      "loss": 2.3066,
+      "step": 3415
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.6258731498064796e-06,
+      "loss": 2.3195,
+      "step": 3416
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.624387724296941e-06,
+      "loss": 2.2613,
+      "step": 3417
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.622901979013717e-06,
+      "loss": 2.2879,
+      "step": 3418
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.6214159141378465e-06,
+      "loss": 2.2704,
+      "step": 3419
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.619929529850397e-06,
+      "loss": 2.3103,
+      "step": 3420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.6184428263324815e-06,
+      "loss": 2.3203,
+      "step": 3421
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.616955803765249e-06,
+      "loss": 2.2709,
+      "step": 3422
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.61546846232989e-06,
+      "loss": 2.2162,
+      "step": 3423
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.613980802207633e-06,
+      "loss": 2.2859,
+      "step": 3424
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.612492823579744e-06,
+      "loss": 2.2952,
+      "step": 3425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.6110045266275305e-06,
+      "loss": 2.2464,
+      "step": 3426
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.6095159115323335e-06,
+      "loss": 2.269,
+      "step": 3427
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.6080269784755405e-06,
+      "loss": 2.2794,
+      "step": 3428
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.60653772763857e-06,
+      "loss": 2.2657,
+      "step": 3429
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.605048159202884e-06,
+      "loss": 2.2983,
+      "step": 3430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.6035582733499805e-06,
+      "loss": 2.3033,
+      "step": 3431
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.6020680702613995e-06,
+      "loss": 2.2844,
+      "step": 3432
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.6005775501187165e-06,
+      "loss": 2.2861,
+      "step": 3433
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.5990867131035474e-06,
+      "loss": 2.3191,
+      "step": 3434
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.597595559397548e-06,
+      "loss": 2.2608,
+      "step": 3435
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.596104089182408e-06,
+      "loss": 2.2887,
+      "step": 3436
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.594612302639859e-06,
+      "loss": 2.2679,
+      "step": 3437
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.5931201999516715e-06,
+      "loss": 2.2645,
+      "step": 3438
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.591627781299654e-06,
+      "loss": 2.3323,
+      "step": 3439
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.590135046865652e-06,
+      "loss": 2.2551,
+      "step": 3440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.588641996831551e-06,
+      "loss": 2.3063,
+      "step": 3441
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.587148631379276e-06,
+      "loss": 2.3254,
+      "step": 3442
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.585654950690786e-06,
+      "loss": 2.3346,
+      "step": 3443
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.5841609549480854e-06,
+      "loss": 2.3213,
+      "step": 3444
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.58266664433321e-06,
+      "loss": 2.3773,
+      "step": 3445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.581172019028238e-06,
+      "loss": 2.3332,
+      "step": 3446
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.579677079215286e-06,
+      "loss": 2.268,
+      "step": 3447
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.578181825076506e-06,
+      "loss": 2.2948,
+      "step": 3448
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.576686256794092e-06,
+      "loss": 2.2764,
+      "step": 3449
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.575190374550272e-06,
+      "loss": 2.2919,
+      "step": 3450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.573694178527317e-06,
+      "loss": 2.2485,
+      "step": 3451
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.572197668907533e-06,
+      "loss": 2.2744,
+      "step": 3452
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.570700845873265e-06,
+      "loss": 2.2944,
+      "step": 3453
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.569203709606898e-06,
+      "loss": 2.2838,
+      "step": 3454
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.5677062602908515e-06,
+      "loss": 2.3535,
+      "step": 3455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.566208498107586e-06,
+      "loss": 2.3991,
+      "step": 3456
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.5647104232395985e-06,
+      "loss": 2.3041,
+      "step": 3457
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.563212035869426e-06,
+      "loss": 2.2329,
+      "step": 3458
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.561713336179642e-06,
+      "loss": 2.3643,
+      "step": 3459
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.560214324352858e-06,
+      "loss": 2.2841,
+      "step": 3460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.5587150005717256e-06,
+      "loss": 2.3056,
+      "step": 3461
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.55721536501893e-06,
+      "loss": 2.2928,
+      "step": 3462
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.555715417877201e-06,
+      "loss": 2.3675,
+      "step": 3463
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.5542151593293e-06,
+      "loss": 2.3348,
+      "step": 3464
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.55271458955803e-06,
+      "loss": 2.2908,
+      "step": 3465
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.55121370874623e-06,
+      "loss": 2.3736,
+      "step": 3466
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.549712517076777e-06,
+      "loss": 2.2824,
+      "step": 3467
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.548211014732589e-06,
+      "loss": 2.1671,
+      "step": 3468
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.546709201896619e-06,
+      "loss": 2.2061,
+      "step": 3469
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.545207078751858e-06,
+      "loss": 2.2305,
+      "step": 3470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.543704645481333e-06,
+      "loss": 2.3064,
+      "step": 3471
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.542201902268115e-06,
+      "loss": 2.2466,
+      "step": 3472
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5406988492953045e-06,
+      "loss": 2.2695,
+      "step": 3473
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.539195486746047e-06,
+      "loss": 2.2621,
+      "step": 3474
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.537691814803522e-06,
+      "loss": 2.171,
+      "step": 3475
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.536187833650947e-06,
+      "loss": 2.2997,
+      "step": 3476
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.534683543471577e-06,
+      "loss": 2.321,
+      "step": 3477
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.533178944448705e-06,
+      "loss": 2.2362,
+      "step": 3478
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.531674036765662e-06,
+      "loss": 2.2698,
+      "step": 3479
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.530168820605819e-06,
+      "loss": 2.2578,
+      "step": 3480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5286632961525806e-06,
+      "loss": 2.2759,
+      "step": 3481
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.527157463589389e-06,
+      "loss": 2.302,
+      "step": 3482
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.525651323099725e-06,
+      "loss": 2.265,
+      "step": 3483
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.52414487486711e-06,
+      "loss": 2.2826,
+      "step": 3484
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.522638119075097e-06,
+      "loss": 2.251,
+      "step": 3485
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.521131055907283e-06,
+      "loss": 2.2599,
+      "step": 3486
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5196236855472945e-06,
+      "loss": 2.3227,
+      "step": 3487
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.518116008178805e-06,
+      "loss": 2.3305,
+      "step": 3488
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.516608023985516e-06,
+      "loss": 2.2446,
+      "step": 3489
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.515099733151177e-06,
+      "loss": 2.2392,
+      "step": 3490
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5135911358595615e-06,
+      "loss": 2.3074,
+      "step": 3491
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.512082232294491e-06,
+      "loss": 2.3144,
+      "step": 3492
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.51057302263982e-06,
+      "loss": 2.3211,
+      "step": 3493
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.509063507079443e-06,
+      "loss": 2.301,
+      "step": 3494
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.507553685797288e-06,
+      "loss": 2.3147,
+      "step": 3495
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5060435589773215e-06,
+      "loss": 2.3279,
+      "step": 3496
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5045331268035505e-06,
+      "loss": 2.2298,
+      "step": 3497
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.503022389460014e-06,
+      "loss": 2.2335,
+      "step": 3498
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.501511347130792e-06,
+      "loss": 2.2965,
+      "step": 3499
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 2.3654,
+      "step": 3500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.498488348251793e-06,
+      "loss": 2.3493,
+      "step": 3501
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.496976392070358e-06,
+      "loss": 2.2335,
+      "step": 3502
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.495464131639923e-06,
+      "loss": 2.2274,
+      "step": 3503
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.493951567144755e-06,
+      "loss": 2.235,
+      "step": 3504
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.4924386987691525e-06,
+      "loss": 2.3188,
+      "step": 3505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.490925526697455e-06,
+      "loss": 2.2508,
+      "step": 3506
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.489412051114038e-06,
+      "loss": 2.3246,
+      "step": 3507
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.487898272203314e-06,
+      "loss": 2.2549,
+      "step": 3508
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.486384190149731e-06,
+      "loss": 2.2399,
+      "step": 3509
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.484869805137778e-06,
+      "loss": 2.2251,
+      "step": 3510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.483355117351976e-06,
+      "loss": 2.2824,
+      "step": 3511
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.481840126976885e-06,
+      "loss": 2.285,
+      "step": 3512
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.480324834197103e-06,
+      "loss": 2.2782,
+      "step": 3513
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.478809239197264e-06,
+      "loss": 2.3316,
+      "step": 3514
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.477293342162038e-06,
+      "loss": 2.2236,
+      "step": 3515
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.475777143276133e-06,
+      "loss": 2.3057,
+      "step": 3516
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.4742606427242935e-06,
+      "loss": 2.2928,
+      "step": 3517
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.4727438406912986e-06,
+      "loss": 2.2426,
+      "step": 3518
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.471226737361968e-06,
+      "loss": 2.2559,
+      "step": 3519
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.469709332921155e-06,
+      "loss": 2.2924,
+      "step": 3520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.468191627553752e-06,
+      "loss": 2.3194,
+      "step": 3521
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.4666736214446855e-06,
+      "loss": 2.2971,
+      "step": 3522
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.46515531477892e-06,
+      "loss": 2.2607,
+      "step": 3523
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.463636707741458e-06,
+      "loss": 2.283,
+      "step": 3524
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.462117800517337e-06,
+      "loss": 2.2865,
+      "step": 3525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.460598593291628e-06,
+      "loss": 2.2521,
+      "step": 3526
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.459079086249445e-06,
+      "loss": 2.245,
+      "step": 3527
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.4575592795759356e-06,
+      "loss": 2.3044,
+      "step": 3528
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.456039173456282e-06,
+      "loss": 2.2541,
+      "step": 3529
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.454518768075705e-06,
+      "loss": 2.2638,
+      "step": 3530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.452998063619461e-06,
+      "loss": 2.2068,
+      "step": 3531
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.451477060272844e-06,
+      "loss": 2.1861,
+      "step": 3532
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.449955758221184e-06,
+      "loss": 2.2806,
+      "step": 3533
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.448434157649846e-06,
+      "loss": 2.2412,
+      "step": 3534
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.446912258744232e-06,
+      "loss": 2.2666,
+      "step": 3535
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.445390061689782e-06,
+      "loss": 2.2915,
+      "step": 3536
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.443867566671971e-06,
+      "loss": 2.272,
+      "step": 3537
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.44234477387631e-06,
+      "loss": 2.3464,
+      "step": 3538
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.440821683488346e-06,
+      "loss": 2.3159,
+      "step": 3539
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.4392982956936644e-06,
+      "loss": 2.3263,
+      "step": 3540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.437774610677884e-06,
+      "loss": 2.232,
+      "step": 3541
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.436250628626662e-06,
+      "loss": 2.2561,
+      "step": 3542
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.434726349725692e-06,
+      "loss": 2.256,
+      "step": 3543
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.433201774160701e-06,
+      "loss": 2.3568,
+      "step": 3544
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.431676902117453e-06,
+      "loss": 2.3057,
+      "step": 3545
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.430151733781752e-06,
+      "loss": 2.27,
+      "step": 3546
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.428626269339433e-06,
+      "loss": 2.3146,
+      "step": 3547
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.42710050897637e-06,
+      "loss": 2.2441,
+      "step": 3548
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.425574452878474e-06,
+      "loss": 2.256,
+      "step": 3549
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.424048101231687e-06,
+      "loss": 2.3131,
+      "step": 3550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.42252145422199e-06,
+      "loss": 2.2683,
+      "step": 3551
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.4209945120354045e-06,
+      "loss": 2.3369,
+      "step": 3552
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.419467274857981e-06,
+      "loss": 2.3037,
+      "step": 3553
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.4179397428758085e-06,
+      "loss": 2.2148,
+      "step": 3554
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.416411916275012e-06,
+      "loss": 2.255,
+      "step": 3555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.414883795241754e-06,
+      "loss": 2.2794,
+      "step": 3556
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.413355379962231e-06,
+      "loss": 2.189,
+      "step": 3557
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.411826670622676e-06,
+      "loss": 2.2068,
+      "step": 3558
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.410297667409356e-06,
+      "loss": 2.3013,
+      "step": 3559
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.408768370508577e-06,
+      "loss": 2.2557,
+      "step": 3560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.407238780106679e-06,
+      "loss": 2.2519,
+      "step": 3561
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.405708896390037e-06,
+      "loss": 2.3184,
+      "step": 3562
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.404178719545063e-06,
+      "loss": 2.2616,
+      "step": 3563
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.402648249758204e-06,
+      "loss": 2.2464,
+      "step": 3564
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.401117487215945e-06,
+      "loss": 2.2534,
+      "step": 3565
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3995864321048036e-06,
+      "loss": 2.3326,
+      "step": 3566
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.398055084611333e-06,
+      "loss": 2.2517,
+      "step": 3567
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.396523444922126e-06,
+      "loss": 2.3336,
+      "step": 3568
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.394991513223806e-06,
+      "loss": 2.3215,
+      "step": 3569
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.393459289703035e-06,
+      "loss": 2.3196,
+      "step": 3570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.391926774546509e-06,
+      "loss": 2.2013,
+      "step": 3571
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.390393967940962e-06,
+      "loss": 2.3534,
+      "step": 3572
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.38886087007316e-06,
+      "loss": 2.2673,
+      "step": 3573
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3873274811299065e-06,
+      "loss": 2.2515,
+      "step": 3574
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3857938012980425e-06,
+      "loss": 2.2754,
+      "step": 3575
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3842598307644396e-06,
+      "loss": 2.2874,
+      "step": 3576
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.382725569716009e-06,
+      "loss": 2.2608,
+      "step": 3577
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.381191018339697e-06,
+      "loss": 2.2413,
+      "step": 3578
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.379656176822481e-06,
+      "loss": 2.2769,
+      "step": 3579
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.378121045351378e-06,
+      "loss": 2.324,
+      "step": 3580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.376585624113438e-06,
+      "loss": 2.2641,
+      "step": 3581
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.37504991329575e-06,
+      "loss": 2.3131,
+      "step": 3582
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.373513913085434e-06,
+      "loss": 2.2353,
+      "step": 3583
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.371977623669646e-06,
+      "loss": 2.2795,
+      "step": 3584
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.370441045235581e-06,
+      "loss": 2.3212,
+      "step": 3585
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.368904177970466e-06,
+      "loss": 2.3247,
+      "step": 3586
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3673670220615615e-06,
+      "loss": 2.2955,
+      "step": 3587
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.365829577696166e-06,
+      "loss": 2.263,
+      "step": 3588
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.364291845061613e-06,
+      "loss": 2.3319,
+      "step": 3589
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.362753824345271e-06,
+      "loss": 2.2584,
+      "step": 3590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.361215515734542e-06,
+      "loss": 2.2814,
+      "step": 3591
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3596769194168646e-06,
+      "loss": 2.296,
+      "step": 3592
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.358138035579712e-06,
+      "loss": 2.325,
+      "step": 3593
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3565988644105926e-06,
+      "loss": 2.2739,
+      "step": 3594
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.35505940609705e-06,
+      "loss": 2.295,
+      "step": 3595
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.353519660826665e-06,
+      "loss": 2.3075,
+      "step": 3596
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.351979628787045e-06,
+      "loss": 2.2558,
+      "step": 3597
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.350439310165842e-06,
+      "loss": 2.2766,
+      "step": 3598
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.34889870515074e-06,
+      "loss": 2.2551,
+      "step": 3599
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.347357813929455e-06,
+      "loss": 2.2956,
+      "step": 3600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.345816636689741e-06,
+      "loss": 2.3132,
+      "step": 3601
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.344275173619385e-06,
+      "loss": 2.3507,
+      "step": 3602
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3427334249062085e-06,
+      "loss": 2.2453,
+      "step": 3603
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.341191390738073e-06,
+      "loss": 2.3407,
+      "step": 3604
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3396490713028674e-06,
+      "loss": 2.2692,
+      "step": 3605
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.33810646678852e-06,
+      "loss": 2.2932,
+      "step": 3606
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.33656357738299e-06,
+      "loss": 2.2593,
+      "step": 3607
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.335020403274277e-06,
+      "loss": 2.2594,
+      "step": 3608
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.333476944650411e-06,
+      "loss": 2.3381,
+      "step": 3609
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3319332016994575e-06,
+      "loss": 2.2921,
+      "step": 3610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.330389174609516e-06,
+      "loss": 2.2881,
+      "step": 3611
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3288448635687215e-06,
+      "loss": 2.2742,
+      "step": 3612
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.327300268765246e-06,
+      "loss": 2.3449,
+      "step": 3613
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.325755390387293e-06,
+      "loss": 2.2083,
+      "step": 3614
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3242102286231e-06,
+      "loss": 2.2355,
+      "step": 3615
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.32266478366094e-06,
+      "loss": 2.2477,
+      "step": 3616
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.321119055689121e-06,
+      "loss": 2.3046,
+      "step": 3617
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.319573044895986e-06,
+      "loss": 2.2411,
+      "step": 3618
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.318026751469912e-06,
+      "loss": 2.3058,
+      "step": 3619
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.31648017559931e-06,
+      "loss": 2.2693,
+      "step": 3620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3149333174726246e-06,
+      "loss": 2.2665,
+      "step": 3621
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.313386177278335e-06,
+      "loss": 2.2816,
+      "step": 3622
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.31183875520496e-06,
+      "loss": 2.2654,
+      "step": 3623
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.310291051441044e-06,
+      "loss": 2.2848,
+      "step": 3624
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.308743066175172e-06,
+      "loss": 2.1909,
+      "step": 3625
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.307194799595958e-06,
+      "loss": 2.2243,
+      "step": 3626
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.305646251892058e-06,
+      "loss": 2.2769,
+      "step": 3627
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3040974232521555e-06,
+      "loss": 2.2712,
+      "step": 3628
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.302548313864971e-06,
+      "loss": 2.2825,
+      "step": 3629
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.300998923919259e-06,
+      "loss": 2.2614,
+      "step": 3630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.299449253603807e-06,
+      "loss": 2.2931,
+      "step": 3631
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.297899303107441e-06,
+      "loss": 2.3297,
+      "step": 3632
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.2963490726190134e-06,
+      "loss": 2.2456,
+      "step": 3633
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.294798562327417e-06,
+      "loss": 2.202,
+      "step": 3634
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.293247772421577e-06,
+      "loss": 2.2524,
+      "step": 3635
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.291696703090449e-06,
+      "loss": 2.2298,
+      "step": 3636
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.2901453545230325e-06,
+      "loss": 2.2757,
+      "step": 3637
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.288593726908351e-06,
+      "loss": 2.2304,
+      "step": 3638
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.287041820435465e-06,
+      "loss": 2.2607,
+      "step": 3639
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.285489635293472e-06,
+      "loss": 2.2841,
+      "step": 3640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.283937171671498e-06,
+      "loss": 2.2505,
+      "step": 3641
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.282384429758709e-06,
+      "loss": 2.2326,
+      "step": 3642
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.280831409744299e-06,
+      "loss": 2.2743,
+      "step": 3643
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.279278111817502e-06,
+      "loss": 2.2872,
+      "step": 3644
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.2777245361675786e-06,
+      "loss": 2.2954,
+      "step": 3645
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.27617068298383e-06,
+      "loss": 2.3318,
+      "step": 3646
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.274616552455589e-06,
+      "loss": 2.3159,
+      "step": 3647
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.27306214477222e-06,
+      "loss": 2.3741,
+      "step": 3648
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.271507460123124e-06,
+      "loss": 2.2744,
+      "step": 3649
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.269952498697734e-06,
+      "loss": 2.2525,
+      "step": 3650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.268397260685518e-06,
+      "loss": 2.228,
+      "step": 3651
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.266841746275977e-06,
+      "loss": 2.2843,
+      "step": 3652
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.265285955658645e-06,
+      "loss": 2.3133,
+      "step": 3653
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.26372988902309e-06,
+      "loss": 2.278,
+      "step": 3654
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.262173546558914e-06,
+      "loss": 2.2987,
+      "step": 3655
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.260616928455754e-06,
+      "loss": 2.2585,
+      "step": 3656
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.259060034903278e-06,
+      "loss": 2.1647,
+      "step": 3657
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.257502866091192e-06,
+      "loss": 2.2575,
+      "step": 3658
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2559454222092265e-06,
+      "loss": 2.2739,
+      "step": 3659
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.254387703447154e-06,
+      "loss": 2.3304,
+      "step": 3660
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2528297099947796e-06,
+      "loss": 2.2188,
+      "step": 3661
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.251271442041938e-06,
+      "loss": 2.24,
+      "step": 3662
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2497128997785e-06,
+      "loss": 2.3238,
+      "step": 3663
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.24815408339437e-06,
+      "loss": 2.2888,
+      "step": 3664
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.246594993079483e-06,
+      "loss": 2.2285,
+      "step": 3665
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.245035629023812e-06,
+      "loss": 2.2466,
+      "step": 3666
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.24347599141736e-06,
+      "loss": 2.318,
+      "step": 3667
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.241916080450163e-06,
+      "loss": 2.272,
+      "step": 3668
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.24035589631229e-06,
+      "loss": 2.293,
+      "step": 3669
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.238795439193849e-06,
+      "loss": 2.2736,
+      "step": 3670
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2372347092849744e-06,
+      "loss": 2.2584,
+      "step": 3671
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.235673706775837e-06,
+      "loss": 2.315,
+      "step": 3672
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.234112431856639e-06,
+      "loss": 2.2376,
+      "step": 3673
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2325508847176175e-06,
+      "loss": 2.3319,
+      "step": 3674
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2309890655490446e-06,
+      "loss": 2.2575,
+      "step": 3675
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2294269745412214e-06,
+      "loss": 2.1885,
+      "step": 3676
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.227864611884483e-06,
+      "loss": 2.3445,
+      "step": 3677
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.226301977769199e-06,
+      "loss": 2.3019,
+      "step": 3678
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.224739072385773e-06,
+      "loss": 2.3105,
+      "step": 3679
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.223175895924638e-06,
+      "loss": 2.2195,
+      "step": 3680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.221612448576265e-06,
+      "loss": 2.2566,
+      "step": 3681
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.220048730531154e-06,
+      "loss": 2.3326,
+      "step": 3682
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2184847419798384e-06,
+      "loss": 2.2741,
+      "step": 3683
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.216920483112886e-06,
+      "loss": 2.3306,
+      "step": 3684
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.215355954120899e-06,
+      "loss": 2.2474,
+      "step": 3685
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.21379115519451e-06,
+      "loss": 2.3122,
+      "step": 3686
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.212226086524381e-06,
+      "loss": 2.319,
+      "step": 3687
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.210660748301214e-06,
+      "loss": 2.252,
+      "step": 3688
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.209095140715742e-06,
+      "loss": 2.328,
+      "step": 3689
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.207529263958727e-06,
+      "loss": 2.3374,
+      "step": 3690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2059631182209676e-06,
+      "loss": 2.3366,
+      "step": 3691
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2043967036932935e-06,
+      "loss": 2.2599,
+      "step": 3692
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.202830020566567e-06,
+      "loss": 2.2983,
+      "step": 3693
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.201263069031686e-06,
+      "loss": 2.3121,
+      "step": 3694
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.199695849279576e-06,
+      "loss": 2.315,
+      "step": 3695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.1981283615012e-06,
+      "loss": 2.2068,
+      "step": 3696
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.196560605887551e-06,
+      "loss": 2.311,
+      "step": 3697
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.194992582629654e-06,
+      "loss": 2.3344,
+      "step": 3698
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.19342429191857e-06,
+      "loss": 2.3003,
+      "step": 3699
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.191855733945388e-06,
+      "loss": 2.3008,
+      "step": 3700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.190286908901234e-06,
+      "loss": 2.3847,
+      "step": 3701
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.188717816977264e-06,
+      "loss": 2.2727,
+      "step": 3702
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.187148458364668e-06,
+      "loss": 2.3204,
+      "step": 3703
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.185578833254665e-06,
+      "loss": 2.297,
+      "step": 3704
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.184008941838512e-06,
+      "loss": 2.288,
+      "step": 3705
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.182438784307495e-06,
+      "loss": 2.2869,
+      "step": 3706
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.18086836085293e-06,
+      "loss": 2.3064,
+      "step": 3707
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.179297671666171e-06,
+      "loss": 2.3211,
+      "step": 3708
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.177726716938602e-06,
+      "loss": 2.3468,
+      "step": 3709
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.176155496861639e-06,
+      "loss": 2.2806,
+      "step": 3710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.174584011626729e-06,
+      "loss": 2.2975,
+      "step": 3711
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.173012261425352e-06,
+      "loss": 2.2525,
+      "step": 3712
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.171440246449024e-06,
+      "loss": 2.2363,
+      "step": 3713
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.169867966889288e-06,
+      "loss": 2.2891,
+      "step": 3714
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.168295422937723e-06,
+      "loss": 2.2475,
+      "step": 3715
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.166722614785937e-06,
+      "loss": 2.2082,
+      "step": 3716
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.1651495426255725e-06,
+      "loss": 2.2006,
+      "step": 3717
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.1635762066483035e-06,
+      "loss": 2.2692,
+      "step": 3718
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.162002607045838e-06,
+      "loss": 2.2647,
+      "step": 3719
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.160428744009913e-06,
+      "loss": 2.2671,
+      "step": 3720
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.1588546177322975e-06,
+      "loss": 2.2862,
+      "step": 3721
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.157280228404796e-06,
+      "loss": 2.2315,
+      "step": 3722
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.155705576219242e-06,
+      "loss": 2.339,
+      "step": 3723
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.154130661367503e-06,
+      "loss": 2.1743,
+      "step": 3724
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.1525554840414765e-06,
+      "loss": 2.2443,
+      "step": 3725
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.150980044433094e-06,
+      "loss": 2.2112,
+      "step": 3726
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.149404342734317e-06,
+      "loss": 2.2805,
+      "step": 3727
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.1478283791371415e-06,
+      "loss": 2.263,
+      "step": 3728
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.146252153833594e-06,
+      "loss": 2.3412,
+      "step": 3729
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.1446756670157306e-06,
+      "loss": 2.2356,
+      "step": 3730
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.143098918875643e-06,
+      "loss": 2.3042,
+      "step": 3731
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.141521909605452e-06,
+      "loss": 2.3659,
+      "step": 3732
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.139944639397313e-06,
+      "loss": 2.316,
+      "step": 3733
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.138367108443411e-06,
+      "loss": 2.2601,
+      "step": 3734
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.1367893169359636e-06,
+      "loss": 2.1893,
+      "step": 3735
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.135211265067217e-06,
+      "loss": 2.2795,
+      "step": 3736
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.133632953029457e-06,
+      "loss": 2.2324,
+      "step": 3737
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.1320543810149945e-06,
+      "loss": 2.2199,
+      "step": 3738
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.130475549216171e-06,
+      "loss": 2.2666,
+      "step": 3739
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.128896457825364e-06,
+      "loss": 2.2227,
+      "step": 3740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.127317107034982e-06,
+      "loss": 2.2186,
+      "step": 3741
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.125737497037464e-06,
+      "loss": 2.1813,
+      "step": 3742
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.124157628025279e-06,
+      "loss": 2.2076,
+      "step": 3743
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.12257750019093e-06,
+      "loss": 2.2663,
+      "step": 3744
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.120997113726951e-06,
+      "loss": 2.2196,
+      "step": 3745
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.119416468825908e-06,
+      "loss": 2.3029,
+      "step": 3746
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.117835565680399e-06,
+      "loss": 2.2778,
+      "step": 3747
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.116254404483049e-06,
+      "loss": 2.1782,
+      "step": 3748
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.114672985426516e-06,
+      "loss": 2.3033,
+      "step": 3749
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.113091308703498e-06,
+      "loss": 2.3329,
+      "step": 3750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.111509374506712e-06,
+      "loss": 2.2976,
+      "step": 3751
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.1099271830289155e-06,
+      "loss": 2.3012,
+      "step": 3752
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.10834473446289e-06,
+      "loss": 2.2917,
+      "step": 3753
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.106762029001455e-06,
+      "loss": 2.2853,
+      "step": 3754
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.105179066837456e-06,
+      "loss": 2.2735,
+      "step": 3755
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.103595848163775e-06,
+      "loss": 2.2229,
+      "step": 3756
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.102012373173319e-06,
+      "loss": 2.2449,
+      "step": 3757
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.100428642059033e-06,
+      "loss": 2.2905,
+      "step": 3758
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.098844655013886e-06,
+      "loss": 2.2901,
+      "step": 3759
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0972604122308865e-06,
+      "loss": 2.2773,
+      "step": 3760
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.095675913903067e-06,
+      "loss": 2.219,
+      "step": 3761
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.094091160223493e-06,
+      "loss": 2.2201,
+      "step": 3762
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.092506151385265e-06,
+      "loss": 2.3007,
+      "step": 3763
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.090920887581507e-06,
+      "loss": 2.2111,
+      "step": 3764
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0893353690053845e-06,
+      "loss": 2.3521,
+      "step": 3765
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.087749595850084e-06,
+      "loss": 2.2453,
+      "step": 3766
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.086163568308828e-06,
+      "loss": 2.2732,
+      "step": 3767
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0845772865748684e-06,
+      "loss": 2.2708,
+      "step": 3768
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0829907508414894e-06,
+      "loss": 2.3121,
+      "step": 3769
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.081403961302007e-06,
+      "loss": 2.2416,
+      "step": 3770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.079816918149765e-06,
+      "loss": 2.2448,
+      "step": 3771
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.07822962157814e-06,
+      "loss": 2.2525,
+      "step": 3772
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.076642071780539e-06,
+      "loss": 2.3013,
+      "step": 3773
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.075054268950402e-06,
+      "loss": 2.201,
+      "step": 3774
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.073466213281196e-06,
+      "loss": 2.2782,
+      "step": 3775
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.071877904966422e-06,
+      "loss": 2.239,
+      "step": 3776
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.070289344199609e-06,
+      "loss": 2.2359,
+      "step": 3777
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0687005311743195e-06,
+      "loss": 2.3027,
+      "step": 3778
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0671114660841455e-06,
+      "loss": 2.2554,
+      "step": 3779
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.06552214912271e-06,
+      "loss": 2.3241,
+      "step": 3780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.063932580483665e-06,
+      "loss": 2.2543,
+      "step": 3781
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0623427603606965e-06,
+      "loss": 2.224,
+      "step": 3782
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.060752688947519e-06,
+      "loss": 2.2778,
+      "step": 3783
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.059162366437875e-06,
+      "loss": 2.2463,
+      "step": 3784
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.057571793025545e-06,
+      "loss": 2.2923,
+      "step": 3785
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0559809689043325e-06,
+      "loss": 2.3043,
+      "step": 3786
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0543898942680755e-06,
+      "loss": 2.2184,
+      "step": 3787
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.052798569310641e-06,
+      "loss": 2.2464,
+      "step": 3788
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0512069942259275e-06,
+      "loss": 2.257,
+      "step": 3789
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.049615169207864e-06,
+      "loss": 2.311,
+      "step": 3790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.048023094450412e-06,
+      "loss": 2.2025,
+      "step": 3791
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0464307701475544e-06,
+      "loss": 2.232,
+      "step": 3792
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.044838196493315e-06,
+      "loss": 2.2901,
+      "step": 3793
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.043245373681746e-06,
+      "loss": 2.2755,
+      "step": 3794
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.041652301906925e-06,
+      "loss": 2.2412,
+      "step": 3795
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0400589813629645e-06,
+      "loss": 2.2444,
+      "step": 3796
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.038465412244005e-06,
+      "loss": 2.2846,
+      "step": 3797
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.036871594744218e-06,
+      "loss": 2.2093,
+      "step": 3798
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.035277529057807e-06,
+      "loss": 2.25,
+      "step": 3799
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.033683215379002e-06,
+      "loss": 2.2825,
+      "step": 3800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0320886539020674e-06,
+      "loss": 2.3583,
+      "step": 3801
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.030493844821291e-06,
+      "loss": 2.2455,
+      "step": 3802
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.028898788331e-06,
+      "loss": 2.2991,
+      "step": 3803
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.027303484625547e-06,
+      "loss": 2.2626,
+      "step": 3804
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.025707933899314e-06,
+      "loss": 2.279,
+      "step": 3805
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.024112136346713e-06,
+      "loss": 2.2222,
+      "step": 3806
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.022516092162187e-06,
+      "loss": 2.2397,
+      "step": 3807
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0209198015402115e-06,
+      "loss": 2.2852,
+      "step": 3808
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.01932326467529e-06,
+      "loss": 2.2635,
+      "step": 3809
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0177264817619514e-06,
+      "loss": 2.2906,
+      "step": 3810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.016129452994761e-06,
+      "loss": 2.2045,
+      "step": 3811
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.014532178568314e-06,
+      "loss": 2.2916,
+      "step": 3812
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.01293465867723e-06,
+      "loss": 2.31,
+      "step": 3813
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.011336893516167e-06,
+      "loss": 2.279,
+      "step": 3814
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.009738883279802e-06,
+      "loss": 2.2436,
+      "step": 3815
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.008140628162851e-06,
+      "loss": 2.2622,
+      "step": 3816
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.006542128360054e-06,
+      "loss": 2.2158,
+      "step": 3817
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0049433840661875e-06,
+      "loss": 2.2549,
+      "step": 3818
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.003344395476051e-06,
+      "loss": 2.2676,
+      "step": 3819
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0017451627844765e-06,
+      "loss": 2.2952,
+      "step": 3820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0001456861863236e-06,
+      "loss": 2.2959,
+      "step": 3821
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.998545965876489e-06,
+      "loss": 2.2297,
+      "step": 3822
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.996946002049889e-06,
+      "loss": 2.276,
+      "step": 3823
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.995345794901477e-06,
+      "loss": 2.276,
+      "step": 3824
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.993745344626232e-06,
+      "loss": 2.2262,
+      "step": 3825
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.992144651419163e-06,
+      "loss": 2.2586,
+      "step": 3826
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.990543715475314e-06,
+      "loss": 2.2364,
+      "step": 3827
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.98894253698975e-06,
+      "loss": 2.2138,
+      "step": 3828
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.987341116157571e-06,
+      "loss": 2.3192,
+      "step": 3829
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.985739453173903e-06,
+      "loss": 2.2743,
+      "step": 3830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.984137548233909e-06,
+      "loss": 2.2802,
+      "step": 3831
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.9825354015327715e-06,
+      "loss": 2.318,
+      "step": 3832
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.98093301326571e-06,
+      "loss": 2.2955,
+      "step": 3833
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.979330383627969e-06,
+      "loss": 2.2878,
+      "step": 3834
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.977727512814826e-06,
+      "loss": 2.2233,
+      "step": 3835
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.976124401021583e-06,
+      "loss": 2.292,
+      "step": 3836
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.974521048443577e-06,
+      "loss": 2.2236,
+      "step": 3837
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.97291745527617e-06,
+      "loss": 2.2387,
+      "step": 3838
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.971313621714756e-06,
+      "loss": 2.1901,
+      "step": 3839
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.9697095479547564e-06,
+      "loss": 2.2366,
+      "step": 3840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.968105234191623e-06,
+      "loss": 2.3268,
+      "step": 3841
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.966500680620837e-06,
+      "loss": 2.2285,
+      "step": 3842
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.9648958874379084e-06,
+      "loss": 2.245,
+      "step": 3843
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.963290854838376e-06,
+      "loss": 2.2516,
+      "step": 3844
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.961685583017808e-06,
+      "loss": 2.1817,
+      "step": 3845
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.960080072171802e-06,
+      "loss": 2.2474,
+      "step": 3846
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.9584743224959846e-06,
+      "loss": 2.2351,
+      "step": 3847
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9568683341860135e-06,
+      "loss": 2.3035,
+      "step": 3848
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.95526210743757e-06,
+      "loss": 2.2841,
+      "step": 3849
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.953655642446368e-06,
+      "loss": 2.2696,
+      "step": 3850
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.952048939408156e-06,
+      "loss": 2.305,
+      "step": 3851
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.950441998518699e-06,
+      "loss": 2.3159,
+      "step": 3852
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.948834819973803e-06,
+      "loss": 2.2432,
+      "step": 3853
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.947227403969293e-06,
+      "loss": 2.2881,
+      "step": 3854
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9456197507010315e-06,
+      "loss": 2.2834,
+      "step": 3855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.944011860364905e-06,
+      "loss": 2.2716,
+      "step": 3856
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.942403733156832e-06,
+      "loss": 2.3237,
+      "step": 3857
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.940795369272754e-06,
+      "loss": 2.2287,
+      "step": 3858
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.939186768908647e-06,
+      "loss": 2.2786,
+      "step": 3859
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9375779322605154e-06,
+      "loss": 2.286,
+      "step": 3860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.935968859524389e-06,
+      "loss": 2.3482,
+      "step": 3861
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.934359550896332e-06,
+      "loss": 2.2733,
+      "step": 3862
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.932750006572428e-06,
+      "loss": 2.2438,
+      "step": 3863
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9311402267488004e-06,
+      "loss": 2.2783,
+      "step": 3864
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.929530211621593e-06,
+      "loss": 2.2799,
+      "step": 3865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.927919961386984e-06,
+      "loss": 2.291,
+      "step": 3866
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.926309476241174e-06,
+      "loss": 2.2827,
+      "step": 3867
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.924698756380398e-06,
+      "loss": 2.2239,
+      "step": 3868
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.923087802000916e-06,
+      "loss": 2.2902,
+      "step": 3869
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.921476613299018e-06,
+      "loss": 2.2409,
+      "step": 3870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.919865190471027e-06,
+      "loss": 2.2346,
+      "step": 3871
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9182535337132824e-06,
+      "loss": 2.2404,
+      "step": 3872
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9166416432221636e-06,
+      "loss": 2.2234,
+      "step": 3873
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.915029519194076e-06,
+      "loss": 2.2653,
+      "step": 3874
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.913417161825449e-06,
+      "loss": 2.3183,
+      "step": 3875
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.911804571312746e-06,
+      "loss": 2.2623,
+      "step": 3876
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.910191747852455e-06,
+      "loss": 2.2529,
+      "step": 3877
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.908578691641092e-06,
+      "loss": 2.2657,
+      "step": 3878
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.906965402875207e-06,
+      "loss": 2.269,
+      "step": 3879
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.905351881751372e-06,
+      "loss": 2.2897,
+      "step": 3880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.903738128466189e-06,
+      "loss": 2.2839,
+      "step": 3881
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9021241432162886e-06,
+      "loss": 2.2634,
+      "step": 3882
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.900509926198332e-06,
+      "loss": 2.2736,
+      "step": 3883
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.898895477609007e-06,
+      "loss": 2.2709,
+      "step": 3884
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.897280797645026e-06,
+      "loss": 2.2643,
+      "step": 3885
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.895665886503136e-06,
+      "loss": 2.253,
+      "step": 3886
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.8940507443801076e-06,
+      "loss": 2.2596,
+      "step": 3887
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.892435371472741e-06,
+      "loss": 2.1956,
+      "step": 3888
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.890819767977865e-06,
+      "loss": 2.213,
+      "step": 3889
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.889203934092337e-06,
+      "loss": 2.3278,
+      "step": 3890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.887587870013039e-06,
+      "loss": 2.3409,
+      "step": 3891
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.885971575936884e-06,
+      "loss": 2.232,
+      "step": 3892
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.884355052060814e-06,
+      "loss": 2.2219,
+      "step": 3893
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.882738298581797e-06,
+      "loss": 2.1914,
+      "step": 3894
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.881121315696828e-06,
+      "loss": 2.3599,
+      "step": 3895
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.879504103602934e-06,
+      "loss": 2.273,
+      "step": 3896
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.877886662497165e-06,
+      "loss": 2.2917,
+      "step": 3897
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.876268992576605e-06,
+      "loss": 2.2393,
+      "step": 3898
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.874651094038358e-06,
+      "loss": 2.2287,
+      "step": 3899
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.873032967079562e-06,
+      "loss": 2.3259,
+      "step": 3900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.87141461189738e-06,
+      "loss": 2.2939,
+      "step": 3901
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.869796028689002e-06,
+      "loss": 2.1955,
+      "step": 3902
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.8681772176516525e-06,
+      "loss": 2.2703,
+      "step": 3903
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.866558178982575e-06,
+      "loss": 2.1975,
+      "step": 3904
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.8649389128790455e-06,
+      "loss": 2.2453,
+      "step": 3905
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.863319419538366e-06,
+      "loss": 2.3177,
+      "step": 3906
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.861699699157868e-06,
+      "loss": 2.2589,
+      "step": 3907
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.860079751934908e-06,
+      "loss": 2.2816,
+      "step": 3908
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.858459578066873e-06,
+      "loss": 2.2213,
+      "step": 3909
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.856839177751175e-06,
+      "loss": 2.2923,
+      "step": 3910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.8552185511852555e-06,
+      "loss": 2.2885,
+      "step": 3911
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.853597698566583e-06,
+      "loss": 2.2753,
+      "step": 3912
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.851976620092655e-06,
+      "loss": 2.2491,
+      "step": 3913
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.850355315960992e-06,
+      "loss": 2.2374,
+      "step": 3914
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.848733786369147e-06,
+      "loss": 2.2274,
+      "step": 3915
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.847112031514698e-06,
+      "loss": 2.303,
+      "step": 3916
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.845490051595252e-06,
+      "loss": 2.3122,
+      "step": 3917
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.843867846808438e-06,
+      "loss": 2.3137,
+      "step": 3918
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.842245417351923e-06,
+      "loss": 2.2953,
+      "step": 3919
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.840622763423391e-06,
+      "loss": 2.2891,
+      "step": 3920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.838999885220558e-06,
+      "loss": 2.2688,
+      "step": 3921
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.837376782941168e-06,
+      "loss": 2.2981,
+      "step": 3922
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.83575345678299e-06,
+      "loss": 2.2565,
+      "step": 3923
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.834129906943822e-06,
+      "loss": 2.2077,
+      "step": 3924
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.832506133621487e-06,
+      "loss": 2.3081,
+      "step": 3925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.830882137013839e-06,
+      "loss": 2.2939,
+      "step": 3926
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.829257917318757e-06,
+      "loss": 2.2579,
+      "step": 3927
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.827633474734145e-06,
+      "loss": 2.3314,
+      "step": 3928
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.82600880945794e-06,
+      "loss": 2.2441,
+      "step": 3929
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.824383921688098e-06,
+      "loss": 2.2371,
+      "step": 3930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.822758811622611e-06,
+      "loss": 2.2979,
+      "step": 3931
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.821133479459492e-06,
+      "loss": 2.2763,
+      "step": 3932
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.819507925396782e-06,
+      "loss": 2.2357,
+      "step": 3933
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.81788214963255e-06,
+      "loss": 2.2997,
+      "step": 3934
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.816256152364893e-06,
+      "loss": 2.2208,
+      "step": 3935
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.814629933791932e-06,
+      "loss": 2.1886,
+      "step": 3936
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.8130034941118185e-06,
+      "loss": 2.2563,
+      "step": 3937
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.811376833522729e-06,
+      "loss": 2.2534,
+      "step": 3938
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.809749952222867e-06,
+      "loss": 2.2367,
+      "step": 3939
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.808122850410461e-06,
+      "loss": 2.2811,
+      "step": 3940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.806495528283772e-06,
+      "loss": 2.1884,
+      "step": 3941
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.804867986041084e-06,
+      "loss": 2.2019,
+      "step": 3942
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.803240223880705e-06,
+      "loss": 2.2628,
+      "step": 3943
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8016122420009745e-06,
+      "loss": 2.2739,
+      "step": 3944
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.799984040600257e-06,
+      "loss": 2.2822,
+      "step": 3945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.798355619876944e-06,
+      "loss": 2.2406,
+      "step": 3946
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.796726980029454e-06,
+      "loss": 2.2548,
+      "step": 3947
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7950981212562315e-06,
+      "loss": 2.197,
+      "step": 3948
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.793469043755747e-06,
+      "loss": 2.2781,
+      "step": 3949
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7918397477265e-06,
+      "loss": 2.2228,
+      "step": 3950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7902102333670185e-06,
+      "loss": 2.243,
+      "step": 3951
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.788580500875848e-06,
+      "loss": 2.2369,
+      "step": 3952
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.786950550451568e-06,
+      "loss": 2.2801,
+      "step": 3953
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.785320382292783e-06,
+      "loss": 2.3169,
+      "step": 3954
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.783689996598126e-06,
+      "loss": 2.2603,
+      "step": 3955
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.782059393566254e-06,
+      "loss": 2.3303,
+      "step": 3956
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7804285733958495e-06,
+      "loss": 2.2518,
+      "step": 3957
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.778797536285625e-06,
+      "loss": 2.252,
+      "step": 3958
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.777166282434316e-06,
+      "loss": 2.2655,
+      "step": 3959
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.775534812040686e-06,
+      "loss": 2.2451,
+      "step": 3960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.773903125303525e-06,
+      "loss": 2.2626,
+      "step": 3961
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.772271222421649e-06,
+      "loss": 2.202,
+      "step": 3962
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7706391035939e-06,
+      "loss": 2.2716,
+      "step": 3963
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.769006769019147e-06,
+      "loss": 2.2691,
+      "step": 3964
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.767374218896286e-06,
+      "loss": 2.2145,
+      "step": 3965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.765741453424237e-06,
+      "loss": 2.2386,
+      "step": 3966
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.764108472801949e-06,
+      "loss": 2.3122,
+      "step": 3967
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.762475277228393e-06,
+      "loss": 2.2127,
+      "step": 3968
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.760841866902572e-06,
+      "loss": 2.238,
+      "step": 3969
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.759208242023509e-06,
+      "loss": 2.3142,
+      "step": 3970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.75757440279026e-06,
+      "loss": 2.3178,
+      "step": 3971
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.755940349401901e-06,
+      "loss": 2.2644,
+      "step": 3972
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.754306082057534e-06,
+      "loss": 2.2421,
+      "step": 3973
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.752671600956295e-06,
+      "loss": 2.233,
+      "step": 3974
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.751036906297338e-06,
+      "loss": 2.2546,
+      "step": 3975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.749401998279845e-06,
+      "loss": 2.3127,
+      "step": 3976
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.747766877103025e-06,
+      "loss": 2.2637,
+      "step": 3977
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.746131542966112e-06,
+      "loss": 2.3057,
+      "step": 3978
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.744495996068367e-06,
+      "loss": 2.2606,
+      "step": 3979
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7428602366090764e-06,
+      "loss": 2.3256,
+      "step": 3980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.741224264787553e-06,
+      "loss": 2.2968,
+      "step": 3981
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.739588080803134e-06,
+      "loss": 2.1907,
+      "step": 3982
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.737951684855185e-06,
+      "loss": 2.2347,
+      "step": 3983
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.736315077143095e-06,
+      "loss": 2.2142,
+      "step": 3984
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7346782578662795e-06,
+      "loss": 2.2985,
+      "step": 3985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.733041227224182e-06,
+      "loss": 2.3339,
+      "step": 3986
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.731403985416265e-06,
+      "loss": 2.2547,
+      "step": 3987
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.729766532642024e-06,
+      "loss": 2.251,
+      "step": 3988
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7281288691009795e-06,
+      "loss": 2.2221,
+      "step": 3989
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7264909949926735e-06,
+      "loss": 2.3102,
+      "step": 3990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7248529105166785e-06,
+      "loss": 2.2867,
+      "step": 3991
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.723214615872585e-06,
+      "loss": 2.2495,
+      "step": 3992
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.72157611126002e-06,
+      "loss": 2.2297,
+      "step": 3993
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.719937396878628e-06,
+      "loss": 2.2157,
+      "step": 3994
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.718298472928082e-06,
+      "loss": 2.3177,
+      "step": 3995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.716659339608077e-06,
+      "loss": 2.2314,
+      "step": 3996
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7150199971183395e-06,
+      "loss": 2.3101,
+      "step": 3997
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.713380445658618e-06,
+      "loss": 2.3031,
+      "step": 3998
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.711740685428687e-06,
+      "loss": 2.268,
+      "step": 3999
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.710100716628345e-06,
+      "loss": 2.2178,
+      "step": 4000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.708460539457418e-06,
+      "loss": 2.3294,
+      "step": 4001
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7068201541157555e-06,
+      "loss": 2.229,
+      "step": 4002
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.705179560803236e-06,
+      "loss": 2.1875,
+      "step": 4003
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.70353875971976e-06,
+      "loss": 2.2299,
+      "step": 4004
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.701897751065251e-06,
+      "loss": 2.2067,
+      "step": 4005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.700256535039665e-06,
+      "loss": 2.225,
+      "step": 4006
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.698615111842977e-06,
+      "loss": 2.2724,
+      "step": 4007
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.6969734816751906e-06,
+      "loss": 2.2919,
+      "step": 4008
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.6953316447363335e-06,
+      "loss": 2.2843,
+      "step": 4009
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.693689601226458e-06,
+      "loss": 2.2393,
+      "step": 4010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.692047351345641e-06,
+      "loss": 2.2662,
+      "step": 4011
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.690404895293987e-06,
+      "loss": 2.2934,
+      "step": 4012
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.688762233271625e-06,
+      "loss": 2.2792,
+      "step": 4013
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.687119365478707e-06,
+      "loss": 2.2311,
+      "step": 4014
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.685476292115411e-06,
+      "loss": 2.2253,
+      "step": 4015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.683833013381942e-06,
+      "loss": 2.2467,
+      "step": 4016
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.682189529478528e-06,
+      "loss": 2.2848,
+      "step": 4017
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.680545840605423e-06,
+      "loss": 2.2807,
+      "step": 4018
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.6789019469629034e-06,
+      "loss": 2.2272,
+      "step": 4019
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.677257848751276e-06,
+      "loss": 2.2695,
+      "step": 4020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.675613546170866e-06,
+      "loss": 2.2232,
+      "step": 4021
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.673969039422029e-06,
+      "loss": 2.2543,
+      "step": 4022
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.672324328705142e-06,
+      "loss": 2.2568,
+      "step": 4023
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.6706794142206085e-06,
+      "loss": 2.2142,
+      "step": 4024
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.669034296168855e-06,
+      "loss": 2.2279,
+      "step": 4025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.6673889747503364e-06,
+      "loss": 2.3494,
+      "step": 4026
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.665743450165528e-06,
+      "loss": 2.2166,
+      "step": 4027
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.664097722614934e-06,
+      "loss": 2.2416,
+      "step": 4028
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.6624517922990795e-06,
+      "loss": 2.2435,
+      "step": 4029
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.6608056594185166e-06,
+      "loss": 2.2309,
+      "step": 4030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.659159324173823e-06,
+      "loss": 2.3133,
+      "step": 4031
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.657512786765599e-06,
+      "loss": 2.2443,
+      "step": 4032
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.655866047394468e-06,
+      "loss": 2.2629,
+      "step": 4033
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.654219106261082e-06,
+      "loss": 2.2186,
+      "step": 4034
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.652571963566116e-06,
+      "loss": 2.1825,
+      "step": 4035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.6509246195102685e-06,
+      "loss": 2.1977,
+      "step": 4036
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.649277074294265e-06,
+      "loss": 2.2762,
+      "step": 4037
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.647629328118852e-06,
+      "loss": 2.1585,
+      "step": 4038
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.645981381184804e-06,
+      "loss": 2.3294,
+      "step": 4039
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.644333233692917e-06,
+      "loss": 2.2285,
+      "step": 4040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.642684885844013e-06,
+      "loss": 2.2115,
+      "step": 4041
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.64103633783894e-06,
+      "loss": 2.2298,
+      "step": 4042
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6393875898785655e-06,
+      "loss": 2.2199,
+      "step": 4043
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.637738642163785e-06,
+      "loss": 2.2386,
+      "step": 4044
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.63608949489552e-06,
+      "loss": 2.2974,
+      "step": 4045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.634440148274712e-06,
+      "loss": 2.2737,
+      "step": 4046
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.632790602502331e-06,
+      "loss": 2.2044,
+      "step": 4047
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.631140857779368e-06,
+      "loss": 2.2709,
+      "step": 4048
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.629490914306839e-06,
+      "loss": 2.2772,
+      "step": 4049
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.627840772285784e-06,
+      "loss": 2.1614,
+      "step": 4050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.62619043191727e-06,
+      "loss": 2.2061,
+      "step": 4051
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.624539893402383e-06,
+      "loss": 2.2912,
+      "step": 4052
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.622889156942239e-06,
+      "loss": 2.2185,
+      "step": 4053
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6212382227379726e-06,
+      "loss": 2.2498,
+      "step": 4054
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.619587090990748e-06,
+      "loss": 2.2944,
+      "step": 4055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.617935761901748e-06,
+      "loss": 2.2014,
+      "step": 4056
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.616284235672184e-06,
+      "loss": 2.2069,
+      "step": 4057
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.614632512503289e-06,
+      "loss": 2.251,
+      "step": 4058
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.612980592596319e-06,
+      "loss": 2.3306,
+      "step": 4059
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.611328476152557e-06,
+      "loss": 2.2527,
+      "step": 4060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6096761633733065e-06,
+      "loss": 2.2792,
+      "step": 4061
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6080236544599e-06,
+      "loss": 2.253,
+      "step": 4062
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.606370949613688e-06,
+      "loss": 2.2182,
+      "step": 4063
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.604718049036047e-06,
+      "loss": 2.2921,
+      "step": 4064
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.603064952928382e-06,
+      "loss": 2.3016,
+      "step": 4065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.601411661492114e-06,
+      "loss": 2.3084,
+      "step": 4066
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.599758174928692e-06,
+      "loss": 2.2341,
+      "step": 4067
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.59810449343959e-06,
+      "loss": 2.22,
+      "step": 4068
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.596450617226303e-06,
+      "loss": 2.2779,
+      "step": 4069
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.594796546490351e-06,
+      "loss": 2.3244,
+      "step": 4070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.593142281433277e-06,
+      "loss": 2.2746,
+      "step": 4071
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.591487822256648e-06,
+      "loss": 2.2047,
+      "step": 4072
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.589833169162055e-06,
+      "loss": 2.1723,
+      "step": 4073
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.588178322351113e-06,
+      "loss": 2.2414,
+      "step": 4074
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.586523282025462e-06,
+      "loss": 2.2333,
+      "step": 4075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.58486804838676e-06,
+      "loss": 2.2622,
+      "step": 4076
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.583212621636693e-06,
+      "loss": 2.2839,
+      "step": 4077
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.58155700197697e-06,
+      "loss": 2.2791,
+      "step": 4078
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.579901189609325e-06,
+      "loss": 2.242,
+      "step": 4079
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.578245184735513e-06,
+      "loss": 2.2607,
+      "step": 4080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.576588987557312e-06,
+      "loss": 2.263,
+      "step": 4081
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.574932598276524e-06,
+      "loss": 2.2912,
+      "step": 4082
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.573276017094977e-06,
+      "loss": 2.3073,
+      "step": 4083
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.571619244214521e-06,
+      "loss": 2.3166,
+      "step": 4084
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.569962279837025e-06,
+      "loss": 2.282,
+      "step": 4085
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.5683051241643894e-06,
+      "loss": 2.2591,
+      "step": 4086
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.566647777398529e-06,
+      "loss": 2.2578,
+      "step": 4087
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.5649902397413915e-06,
+      "loss": 2.2545,
+      "step": 4088
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.5633325113949395e-06,
+      "loss": 2.2652,
+      "step": 4089
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.561674592561164e-06,
+      "loss": 2.1773,
+      "step": 4090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.5600164834420754e-06,
+      "loss": 2.2442,
+      "step": 4091
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.558358184239709e-06,
+      "loss": 2.2494,
+      "step": 4092
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.556699695156128e-06,
+      "loss": 2.1789,
+      "step": 4093
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.55504101639341e-06,
+      "loss": 2.2684,
+      "step": 4094
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.55338214815366e-06,
+      "loss": 2.2225,
+      "step": 4095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.551723090639008e-06,
+      "loss": 2.2311,
+      "step": 4096
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.550063844051603e-06,
+      "loss": 2.2966,
+      "step": 4097
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.548404408593622e-06,
+      "loss": 2.2649,
+      "step": 4098
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.546744784467261e-06,
+      "loss": 2.296,
+      "step": 4099
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.545084971874738e-06,
+      "loss": 2.3074,
+      "step": 4100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.543424971018298e-06,
+      "loss": 2.2612,
+      "step": 4101
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.541764782100208e-06,
+      "loss": 2.2505,
+      "step": 4102
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.540104405322757e-06,
+      "loss": 2.2357,
+      "step": 4103
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.538443840888254e-06,
+      "loss": 2.278,
+      "step": 4104
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.536783088999037e-06,
+      "loss": 2.2416,
+      "step": 4105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.53512214985746e-06,
+      "loss": 2.224,
+      "step": 4106
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.533461023665907e-06,
+      "loss": 2.2733,
+      "step": 4107
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.53179971062678e-06,
+      "loss": 2.266,
+      "step": 4108
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.530138210942505e-06,
+      "loss": 2.2267,
+      "step": 4109
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.5284765248155295e-06,
+      "loss": 2.2723,
+      "step": 4110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.526814652448325e-06,
+      "loss": 2.2667,
+      "step": 4111
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.525152594043389e-06,
+      "loss": 2.2689,
+      "step": 4112
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.5234903498032345e-06,
+      "loss": 2.2509,
+      "step": 4113
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.5218279199304014e-06,
+      "loss": 2.1985,
+      "step": 4114
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.520165304627452e-06,
+      "loss": 2.2684,
+      "step": 4115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.518502504096972e-06,
+      "loss": 2.245,
+      "step": 4116
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.516839518541569e-06,
+      "loss": 2.3386,
+      "step": 4117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.5151763481638705e-06,
+      "loss": 2.2478,
+      "step": 4118
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.5135129931665305e-06,
+      "loss": 2.251,
+      "step": 4119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.5118494537522235e-06,
+      "loss": 2.2852,
+      "step": 4120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.510185730123646e-06,
+      "loss": 2.3253,
+      "step": 4121
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.508521822483518e-06,
+      "loss": 2.2607,
+      "step": 4122
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.506857731034582e-06,
+      "loss": 2.2804,
+      "step": 4123
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.505193455979603e-06,
+      "loss": 2.2668,
+      "step": 4124
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.503528997521365e-06,
+      "loss": 2.2305,
+      "step": 4125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.501864355862682e-06,
+      "loss": 2.2568,
+      "step": 4126
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.500199531206381e-06,
+      "loss": 2.2474,
+      "step": 4127
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.49853452375532e-06,
+      "loss": 2.2299,
+      "step": 4128
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.49686933371237e-06,
+      "loss": 2.2207,
+      "step": 4129
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.495203961280434e-06,
+      "loss": 2.1918,
+      "step": 4130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.493538406662429e-06,
+      "loss": 2.2526,
+      "step": 4131
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.491872670061302e-06,
+      "loss": 2.2691,
+      "step": 4132
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.490206751680015e-06,
+      "loss": 2.2744,
+      "step": 4133
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4885406517215535e-06,
+      "loss": 2.3042,
+      "step": 4134
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.48687437038893e-06,
+      "loss": 2.183,
+      "step": 4135
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.485207907885175e-06,
+      "loss": 2.2783,
+      "step": 4136
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.483541264413342e-06,
+      "loss": 2.2593,
+      "step": 4137
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.481874440176506e-06,
+      "loss": 2.2738,
+      "step": 4138
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4802074353777625e-06,
+      "loss": 2.1693,
+      "step": 4139
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4785402502202345e-06,
+      "loss": 2.2446,
+      "step": 4140
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.476872884907061e-06,
+      "loss": 2.1389,
+      "step": 4141
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4752053396414075e-06,
+      "loss": 2.2358,
+      "step": 4142
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.473537614626457e-06,
+      "loss": 2.2488,
+      "step": 4143
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.471869710065418e-06,
+      "loss": 2.2511,
+      "step": 4144
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.47020162616152e-06,
+      "loss": 2.2141,
+      "step": 4145
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4685333631180145e-06,
+      "loss": 2.2791,
+      "step": 4146
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.466864921138171e-06,
+      "loss": 2.2194,
+      "step": 4147
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.465196300425287e-06,
+      "loss": 2.2198,
+      "step": 4148
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.463527501182679e-06,
+      "loss": 2.2547,
+      "step": 4149
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.461858523613684e-06,
+      "loss": 2.2952,
+      "step": 4150
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.460189367921663e-06,
+      "loss": 2.2819,
+      "step": 4151
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.458520034309995e-06,
+      "loss": 2.2508,
+      "step": 4152
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.456850522982086e-06,
+      "loss": 2.2552,
+      "step": 4153
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.455180834141359e-06,
+      "loss": 2.2039,
+      "step": 4154
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.453510967991263e-06,
+      "loss": 2.2224,
+      "step": 4155
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.451840924735264e-06,
+      "loss": 2.2507,
+      "step": 4156
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4501707045768524e-06,
+      "loss": 2.2488,
+      "step": 4157
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.448500307719537e-06,
+      "loss": 2.2597,
+      "step": 4158
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.446829734366855e-06,
+      "loss": 2.2687,
+      "step": 4159
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.445158984722358e-06,
+      "loss": 2.1972,
+      "step": 4160
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.443488058989623e-06,
+      "loss": 2.2307,
+      "step": 4161
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.441816957372247e-06,
+      "loss": 2.2315,
+      "step": 4162
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.440145680073847e-06,
+      "loss": 2.3308,
+      "step": 4163
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.438474227298065e-06,
+      "loss": 2.2762,
+      "step": 4164
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4368025992485615e-06,
+      "loss": 2.2027,
+      "step": 4165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.435130796129019e-06,
+      "loss": 2.2056,
+      "step": 4166
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4334588181431424e-06,
+      "loss": 2.2336,
+      "step": 4167
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.431786665494657e-06,
+      "loss": 2.2036,
+      "step": 4168
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.43011433838731e-06,
+      "loss": 2.2659,
+      "step": 4169
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.428441837024868e-06,
+      "loss": 2.1861,
+      "step": 4170
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.426769161611122e-06,
+      "loss": 2.3113,
+      "step": 4171
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.425096312349881e-06,
+      "loss": 2.2926,
+      "step": 4172
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.423423289444978e-06,
+      "loss": 2.2577,
+      "step": 4173
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.421750093100264e-06,
+      "loss": 2.2296,
+      "step": 4174
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.420076723519615e-06,
+      "loss": 2.2839,
+      "step": 4175
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.418403180906923e-06,
+      "loss": 2.2627,
+      "step": 4176
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.416729465466106e-06,
+      "loss": 2.2664,
+      "step": 4177
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.415055577401101e-06,
+      "loss": 2.185,
+      "step": 4178
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.413381516915868e-06,
+      "loss": 2.1664,
+      "step": 4179
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.411707284214384e-06,
+      "loss": 2.2436,
+      "step": 4180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.410032879500647e-06,
+      "loss": 2.209,
+      "step": 4181
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.408358302978683e-06,
+      "loss": 2.2185,
+      "step": 4182
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.406683554852532e-06,
+      "loss": 2.2636,
+      "step": 4183
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4050086353262565e-06,
+      "loss": 2.3362,
+      "step": 4184
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.40333354460394e-06,
+      "loss": 2.2581,
+      "step": 4185
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.401658282889689e-06,
+      "loss": 2.3022,
+      "step": 4186
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.399982850387625e-06,
+      "loss": 2.2431,
+      "step": 4187
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.3983072473019e-06,
+      "loss": 2.2736,
+      "step": 4188
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.396631473836677e-06,
+      "loss": 2.234,
+      "step": 4189
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.3949555301961474e-06,
+      "loss": 2.2954,
+      "step": 4190
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.3932794165845156e-06,
+      "loss": 2.2291,
+      "step": 4191
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.391603133206015e-06,
+      "loss": 2.214,
+      "step": 4192
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.389926680264893e-06,
+      "loss": 2.2048,
+      "step": 4193
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.388250057965421e-06,
+      "loss": 2.2671,
+      "step": 4194
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.386573266511891e-06,
+      "loss": 2.246,
+      "step": 4195
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.384896306108612e-06,
+      "loss": 2.2373,
+      "step": 4196
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.383219176959921e-06,
+      "loss": 2.3353,
+      "step": 4197
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.3815418792701686e-06,
+      "loss": 2.2525,
+      "step": 4198
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.3798644132437304e-06,
+      "loss": 2.199,
+      "step": 4199
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.378186779084996e-06,
+      "loss": 2.2454,
+      "step": 4200
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.376508976998385e-06,
+      "loss": 2.2097,
+      "step": 4201
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.374831007188331e-06,
+      "loss": 2.3136,
+      "step": 4202
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.373152869859288e-06,
+      "loss": 2.2678,
+      "step": 4203
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.371474565215734e-06,
+      "loss": 2.1701,
+      "step": 4204
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.369796093462164e-06,
+      "loss": 2.1574,
+      "step": 4205
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.368117454803093e-06,
+      "loss": 2.1815,
+      "step": 4206
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.366438649443062e-06,
+      "loss": 2.2857,
+      "step": 4207
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.364759677586627e-06,
+      "loss": 2.2565,
+      "step": 4208
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.363080539438364e-06,
+      "loss": 2.3339,
+      "step": 4209
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.361401235202872e-06,
+      "loss": 2.2884,
+      "step": 4210
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.35972176508477e-06,
+      "loss": 2.2671,
+      "step": 4211
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.358042129288694e-06,
+      "loss": 2.2347,
+      "step": 4212
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.356362328019305e-06,
+      "loss": 2.2261,
+      "step": 4213
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.35468236148128e-06,
+      "loss": 2.2771,
+      "step": 4214
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.353002229879318e-06,
+      "loss": 2.3046,
+      "step": 4215
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.35132193341814e-06,
+      "loss": 2.2903,
+      "step": 4216
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.349641472302484e-06,
+      "loss": 2.2042,
+      "step": 4217
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.3479608467371055e-06,
+      "loss": 2.2585,
+      "step": 4218
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.346280056926788e-06,
+      "loss": 2.2777,
+      "step": 4219
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.344599103076329e-06,
+      "loss": 2.2496,
+      "step": 4220
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.3429179853905485e-06,
+      "loss": 2.2841,
+      "step": 4221
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.341236704074285e-06,
+      "loss": 2.2503,
+      "step": 4222
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.339555259332398e-06,
+      "loss": 2.1784,
+      "step": 4223
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.337873651369764e-06,
+      "loss": 2.2386,
+      "step": 4224
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.336191880391285e-06,
+      "loss": 2.2151,
+      "step": 4225
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.334509946601879e-06,
+      "loss": 2.2304,
+      "step": 4226
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.3328278502064835e-06,
+      "loss": 2.1206,
+      "step": 4227
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.331145591410057e-06,
+      "loss": 2.2782,
+      "step": 4228
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.329463170417578e-06,
+      "loss": 2.3027,
+      "step": 4229
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.327780587434045e-06,
+      "loss": 2.2118,
+      "step": 4230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3260978426644735e-06,
+      "loss": 2.2627,
+      "step": 4231
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.324414936313904e-06,
+      "loss": 2.3159,
+      "step": 4232
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3227318685873915e-06,
+      "loss": 2.236,
+      "step": 4233
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.321048639690013e-06,
+      "loss": 2.2019,
+      "step": 4234
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3193652498268656e-06,
+      "loss": 2.2361,
+      "step": 4235
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.317681699203065e-06,
+      "loss": 2.2464,
+      "step": 4236
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.315997988023744e-06,
+      "loss": 2.2042,
+      "step": 4237
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.314314116494061e-06,
+      "loss": 2.1745,
+      "step": 4238
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.312630084819189e-06,
+      "loss": 2.2865,
+      "step": 4239
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.310945893204324e-06,
+      "loss": 2.2212,
+      "step": 4240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.309261541854679e-06,
+      "loss": 2.2222,
+      "step": 4241
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.307577030975485e-06,
+      "loss": 2.2195,
+      "step": 4242
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.305892360771997e-06,
+      "loss": 2.262,
+      "step": 4243
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.304207531449486e-06,
+      "loss": 2.3188,
+      "step": 4244
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3025225432132434e-06,
+      "loss": 2.2637,
+      "step": 4245
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3008373962685785e-06,
+      "loss": 2.2281,
+      "step": 4246
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2991520908208235e-06,
+      "loss": 2.286,
+      "step": 4247
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.297466627075327e-06,
+      "loss": 2.2312,
+      "step": 4248
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.295781005237458e-06,
+      "loss": 2.2682,
+      "step": 4249
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.294095225512604e-06,
+      "loss": 2.2483,
+      "step": 4250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.292409288106173e-06,
+      "loss": 2.3114,
+      "step": 4251
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.290723193223589e-06,
+      "loss": 2.2246,
+      "step": 4252
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2890369410703e-06,
+      "loss": 2.2586,
+      "step": 4253
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.28735053185177e-06,
+      "loss": 2.2183,
+      "step": 4254
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.285663965773482e-06,
+      "loss": 2.297,
+      "step": 4255
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.28397724304094e-06,
+      "loss": 2.2571,
+      "step": 4256
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2822903638596654e-06,
+      "loss": 2.2562,
+      "step": 4257
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.280603328435199e-06,
+      "loss": 2.2466,
+      "step": 4258
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.278916136973102e-06,
+      "loss": 2.269,
+      "step": 4259
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.277228789678953e-06,
+      "loss": 2.2683,
+      "step": 4260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.275541286758352e-06,
+      "loss": 2.2977,
+      "step": 4261
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.273853628416911e-06,
+      "loss": 2.2568,
+      "step": 4262
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.272165814860272e-06,
+      "loss": 2.1783,
+      "step": 4263
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.270477846294086e-06,
+      "loss": 2.2094,
+      "step": 4264
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.268789722924029e-06,
+      "loss": 2.234,
+      "step": 4265
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.267101444955792e-06,
+      "loss": 2.2253,
+      "step": 4266
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.265413012595088e-06,
+      "loss": 2.2795,
+      "step": 4267
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2637244260476474e-06,
+      "loss": 2.2531,
+      "step": 4268
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.262035685519218e-06,
+      "loss": 2.3285,
+      "step": 4269
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.26034679121557e-06,
+      "loss": 2.3001,
+      "step": 4270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.258657743342486e-06,
+      "loss": 2.2179,
+      "step": 4271
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.256968542105775e-06,
+      "loss": 2.25,
+      "step": 4272
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.25527918771126e-06,
+      "loss": 2.2476,
+      "step": 4273
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2535896803647845e-06,
+      "loss": 2.1908,
+      "step": 4274
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.251900020272208e-06,
+      "loss": 2.2552,
+      "step": 4275
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.250210207639411e-06,
+      "loss": 2.3177,
+      "step": 4276
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2485202426722925e-06,
+      "loss": 2.2484,
+      "step": 4277
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.24683012557677e-06,
+      "loss": 2.2304,
+      "step": 4278
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.24513985655878e-06,
+      "loss": 2.2767,
+      "step": 4279
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.243449435824276e-06,
+      "loss": 2.1899,
+      "step": 4280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.241758863579227e-06,
+      "loss": 2.2335,
+      "step": 4281
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.240068140029628e-06,
+      "loss": 2.1961,
+      "step": 4282
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.238377265381489e-06,
+      "loss": 2.158,
+      "step": 4283
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.236686239840836e-06,
+      "loss": 2.2388,
+      "step": 4284
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.234995063613716e-06,
+      "loss": 2.2894,
+      "step": 4285
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.233303736906193e-06,
+      "loss": 2.1979,
+      "step": 4286
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.231612259924351e-06,
+      "loss": 2.2839,
+      "step": 4287
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.229920632874291e-06,
+      "loss": 2.2794,
+      "step": 4288
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.228228855962133e-06,
+      "loss": 2.2169,
+      "step": 4289
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2265369293940135e-06,
+      "loss": 2.237,
+      "step": 4290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2248448533760895e-06,
+      "loss": 2.2034,
+      "step": 4291
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.223152628114537e-06,
+      "loss": 2.2261,
+      "step": 4292
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.221460253815546e-06,
+      "loss": 2.1965,
+      "step": 4293
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.219767730685329e-06,
+      "loss": 2.237,
+      "step": 4294
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.218075058930113e-06,
+      "loss": 2.2124,
+      "step": 4295
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.216382238756147e-06,
+      "loss": 2.1645,
+      "step": 4296
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.214689270369694e-06,
+      "loss": 2.2076,
+      "step": 4297
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.212996153977038e-06,
+      "loss": 2.1873,
+      "step": 4298
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.21130288978448e-06,
+      "loss": 2.2921,
+      "step": 4299
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.209609477998339e-06,
+      "loss": 2.1754,
+      "step": 4300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.207915918824952e-06,
+      "loss": 2.1975,
+      "step": 4301
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.206222212470675e-06,
+      "loss": 2.2583,
+      "step": 4302
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.20452835914188e-06,
+      "loss": 2.3671,
+      "step": 4303
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.202834359044959e-06,
+      "loss": 2.2503,
+      "step": 4304
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2011402123863194e-06,
+      "loss": 2.237,
+      "step": 4305
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.199445919372388e-06,
+      "loss": 2.2676,
+      "step": 4306
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.1977514802096105e-06,
+      "loss": 2.2586,
+      "step": 4307
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.1960568951044475e-06,
+      "loss": 2.18,
+      "step": 4308
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.19436216426338e-06,
+      "loss": 2.2357,
+      "step": 4309
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.192667287892905e-06,
+      "loss": 2.2768,
+      "step": 4310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.1909722661995394e-06,
+      "loss": 2.2492,
+      "step": 4311
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.189277099389816e-06,
+      "loss": 2.2022,
+      "step": 4312
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.187581787670285e-06,
+      "loss": 2.257,
+      "step": 4313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.185886331247516e-06,
+      "loss": 2.283,
+      "step": 4314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.184190730328095e-06,
+      "loss": 2.1827,
+      "step": 4315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.182494985118625e-06,
+      "loss": 2.1867,
+      "step": 4316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.180799095825727e-06,
+      "loss": 2.2202,
+      "step": 4317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.179103062656042e-06,
+      "loss": 2.296,
+      "step": 4318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.177406885816224e-06,
+      "loss": 2.1878,
+      "step": 4319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.17571056551295e-06,
+      "loss": 2.3075,
+      "step": 4320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.17401410195291e-06,
+      "loss": 2.2564,
+      "step": 4321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.172317495342812e-06,
+      "loss": 2.3012,
+      "step": 4322
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1706207458893855e-06,
+      "loss": 2.2552,
+      "step": 4323
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.168923853799369e-06,
+      "loss": 2.257,
+      "step": 4324
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1672268192795285e-06,
+      "loss": 2.2079,
+      "step": 4325
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.16552964253664e-06,
+      "loss": 2.2307,
+      "step": 4326
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.163832323777499e-06,
+      "loss": 2.2712,
+      "step": 4327
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1621348632089205e-06,
+      "loss": 2.256,
+      "step": 4328
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1604372610377335e-06,
+      "loss": 2.1927,
+      "step": 4329
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.158739517470786e-06,
+      "loss": 2.2114,
+      "step": 4330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.157041632714945e-06,
+      "loss": 2.2521,
+      "step": 4331
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.155343606977091e-06,
+      "loss": 2.2128,
+      "step": 4332
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.15364544046412e-06,
+      "loss": 2.2473,
+      "step": 4333
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.151947133382954e-06,
+      "loss": 2.2098,
+      "step": 4334
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.150248685940523e-06,
+      "loss": 2.3124,
+      "step": 4335
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.148550098343778e-06,
+      "loss": 2.2597,
+      "step": 4336
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.146851370799689e-06,
+      "loss": 2.2051,
+      "step": 4337
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.145152503515239e-06,
+      "loss": 2.2707,
+      "step": 4338
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.143453496697428e-06,
+      "loss": 2.2545,
+      "step": 4339
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.141754350553279e-06,
+      "loss": 2.2837,
+      "step": 4340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.140055065289826e-06,
+      "loss": 2.2064,
+      "step": 4341
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.138355641114121e-06,
+      "loss": 2.267,
+      "step": 4342
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.136656078233233e-06,
+      "loss": 2.1977,
+      "step": 4343
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.134956376854251e-06,
+      "loss": 2.2242,
+      "step": 4344
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.133256537184276e-06,
+      "loss": 2.2663,
+      "step": 4345
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.13155655943043e-06,
+      "loss": 2.2339,
+      "step": 4346
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.12985644379985e-06,
+      "loss": 2.2419,
+      "step": 4347
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.128156190499688e-06,
+      "loss": 2.2736,
+      "step": 4348
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1264557997371185e-06,
+      "loss": 2.2905,
+      "step": 4349
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.124755271719326e-06,
+      "loss": 2.2563,
+      "step": 4350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.123054606653515e-06,
+      "loss": 2.2548,
+      "step": 4351
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.121353804746907e-06,
+      "loss": 2.2366,
+      "step": 4352
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.119652866206739e-06,
+      "loss": 2.2265,
+      "step": 4353
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.117951791240265e-06,
+      "loss": 2.2101,
+      "step": 4354
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.116250580054758e-06,
+      "loss": 2.2915,
+      "step": 4355
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.114549232857503e-06,
+      "loss": 2.2763,
+      "step": 4356
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.112847749855804e-06,
+      "loss": 2.2622,
+      "step": 4357
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.111146131256983e-06,
+      "loss": 2.1819,
+      "step": 4358
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.109444377268376e-06,
+      "loss": 2.214,
+      "step": 4359
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.107742488097338e-06,
+      "loss": 2.2129,
+      "step": 4360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.106040463951237e-06,
+      "loss": 2.2396,
+      "step": 4361
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.10433830503746e-06,
+      "loss": 2.2255,
+      "step": 4362
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.102636011563411e-06,
+      "loss": 2.1904,
+      "step": 4363
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.100933583736508e-06,
+      "loss": 2.2991,
+      "step": 4364
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.099231021764188e-06,
+      "loss": 2.2313,
+      "step": 4365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.097528325853903e-06,
+      "loss": 2.2204,
+      "step": 4366
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.095825496213119e-06,
+      "loss": 2.2074,
+      "step": 4367
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.094122533049324e-06,
+      "loss": 2.2112,
+      "step": 4368
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.092419436570016e-06,
+      "loss": 2.2362,
+      "step": 4369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.090716206982714e-06,
+      "loss": 2.2184,
+      "step": 4370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.08901284449495e-06,
+      "loss": 2.2174,
+      "step": 4371
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.087309349314275e-06,
+      "loss": 2.2492,
+      "step": 4372
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.085605721648253e-06,
+      "loss": 2.2,
+      "step": 4373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.083901961704467e-06,
+      "loss": 2.1872,
+      "step": 4374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.0821980696905145e-06,
+      "loss": 2.2741,
+      "step": 4375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.080494045814011e-06,
+      "loss": 2.2676,
+      "step": 4376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.078789890282585e-06,
+      "loss": 2.2928,
+      "step": 4377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.077085603303883e-06,
+      "loss": 2.2031,
+      "step": 4378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.075381185085568e-06,
+      "loss": 2.2206,
+      "step": 4379
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.073676635835317e-06,
+      "loss": 2.2662,
+      "step": 4380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.071971955760823e-06,
+      "loss": 2.261,
+      "step": 4381
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.0702671450698005e-06,
+      "loss": 2.188,
+      "step": 4382
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.068562203969972e-06,
+      "loss": 2.2693,
+      "step": 4383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.066857132669081e-06,
+      "loss": 2.2117,
+      "step": 4384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.065151931374884e-06,
+      "loss": 2.1456,
+      "step": 4385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.0634466002951545e-06,
+      "loss": 2.3097,
+      "step": 4386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.061741139637682e-06,
+      "loss": 2.2761,
+      "step": 4387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.060035549610275e-06,
+      "loss": 2.273,
+      "step": 4388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.058329830420749e-06,
+      "loss": 2.2153,
+      "step": 4389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.056623982276945e-06,
+      "loss": 2.2279,
+      "step": 4390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.0549180053867114e-06,
+      "loss": 2.2023,
+      "step": 4391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.0532118999579206e-06,
+      "loss": 2.2817,
+      "step": 4392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.051505666198454e-06,
+      "loss": 2.2673,
+      "step": 4393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.049799304316214e-06,
+      "loss": 2.2021,
+      "step": 4394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.048092814519109e-06,
+      "loss": 2.2355,
+      "step": 4395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.046386197015076e-06,
+      "loss": 2.1719,
+      "step": 4396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.044679452012059e-06,
+      "loss": 2.227,
+      "step": 4397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.04297257971802e-06,
+      "loss": 2.197,
+      "step": 4398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.041265580340935e-06,
+      "loss": 2.1828,
+      "step": 4399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.039558454088796e-06,
+      "loss": 2.1646,
+      "step": 4400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.0378512011696155e-06,
+      "loss": 2.2647,
+      "step": 4401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.036143821791413e-06,
+      "loss": 2.2086,
+      "step": 4402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.03443631616223e-06,
+      "loss": 2.2449,
+      "step": 4403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.032728684490118e-06,
+      "loss": 2.3038,
+      "step": 4404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.031020926983149e-06,
+      "loss": 2.1751,
+      "step": 4405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.029313043849407e-06,
+      "loss": 2.2265,
+      "step": 4406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.027605035296994e-06,
+      "loss": 2.2645,
+      "step": 4407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.025896901534023e-06,
+      "loss": 2.2553,
+      "step": 4408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.024188642768628e-06,
+      "loss": 2.2143,
+      "step": 4409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.022480259208951e-06,
+      "loss": 2.2806,
+      "step": 4410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.020771751063159e-06,
+      "loss": 2.2697,
+      "step": 4411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.019063118539425e-06,
+      "loss": 2.217,
+      "step": 4412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.01735436184594e-06,
+      "loss": 2.2302,
+      "step": 4413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.015645481190912e-06,
+      "loss": 2.2674,
+      "step": 4414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.013936476782563e-06,
+      "loss": 2.2655,
+      "step": 4415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.01222734882913e-06,
+      "loss": 2.2031,
+      "step": 4416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.010518097538866e-06,
+      "loss": 2.2048,
+      "step": 4417
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.008808723120035e-06,
+      "loss": 2.2423,
+      "step": 4418
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.007099225780922e-06,
+      "loss": 2.2746,
+      "step": 4419
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.005389605729824e-06,
+      "loss": 2.2515,
+      "step": 4420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.003679863175052e-06,
+      "loss": 2.2354,
+      "step": 4421
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.001969998324932e-06,
+      "loss": 2.2384,
+      "step": 4422
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.000260011387809e-06,
+      "loss": 2.2224,
+      "step": 4423
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9985499025720354e-06,
+      "loss": 2.2255,
+      "step": 4424
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.996839672085986e-06,
+      "loss": 2.2189,
+      "step": 4425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.995129320138047e-06,
+      "loss": 2.2347,
+      "step": 4426
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.993418846936619e-06,
+      "loss": 2.2769,
+      "step": 4427
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.991708252690117e-06,
+      "loss": 2.2296,
+      "step": 4428
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.989997537606973e-06,
+      "loss": 2.2035,
+      "step": 4429
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.988286701895631e-06,
+      "loss": 2.2599,
+      "step": 4430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.986575745764553e-06,
+      "loss": 2.2625,
+      "step": 4431
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.984864669422214e-06,
+      "loss": 2.249,
+      "step": 4432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9831534730771e-06,
+      "loss": 2.2626,
+      "step": 4433
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.98144215693772e-06,
+      "loss": 2.2215,
+      "step": 4434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.979730721212589e-06,
+      "loss": 2.2205,
+      "step": 4435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.978019166110242e-06,
+      "loss": 2.1392,
+      "step": 4436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.976307491839226e-06,
+      "loss": 2.2866,
+      "step": 4437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.974595698608103e-06,
+      "loss": 2.2529,
+      "step": 4438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.972883786625452e-06,
+      "loss": 2.1837,
+      "step": 4439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.97117175609986e-06,
+      "loss": 2.2015,
+      "step": 4440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.969459607239938e-06,
+      "loss": 2.3116,
+      "step": 4441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.967747340254303e-06,
+      "loss": 2.1329,
+      "step": 4442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.966034955351588e-06,
+      "loss": 2.2075,
+      "step": 4443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.964322452740445e-06,
+      "loss": 2.1969,
+      "step": 4444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.962609832629538e-06,
+      "loss": 2.3099,
+      "step": 4445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.960897095227541e-06,
+      "loss": 2.2162,
+      "step": 4446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.959184240743149e-06,
+      "loss": 2.2959,
+      "step": 4447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.957471269385065e-06,
+      "loss": 2.1565,
+      "step": 4448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.955758181362012e-06,
+      "loss": 2.2608,
+      "step": 4449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.954044976882725e-06,
+      "loss": 2.2654,
+      "step": 4450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.952331656155951e-06,
+      "loss": 2.226,
+      "step": 4451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.950618219390451e-06,
+      "loss": 2.2091,
+      "step": 4452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.948904666795007e-06,
+      "loss": 2.2253,
+      "step": 4453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.947190998578407e-06,
+      "loss": 2.2394,
+      "step": 4454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.945477214949457e-06,
+      "loss": 2.3394,
+      "step": 4455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.943763316116977e-06,
+      "loss": 2.2818,
+      "step": 4456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.942049302289798e-06,
+      "loss": 2.1265,
+      "step": 4457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.94033517367677e-06,
+      "loss": 2.1638,
+      "step": 4458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.938620930486754e-06,
+      "loss": 2.2973,
+      "step": 4459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.936906572928625e-06,
+      "loss": 2.2852,
+      "step": 4460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.93519210121127e-06,
+      "loss": 2.1591,
+      "step": 4461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.933477515543595e-06,
+      "loss": 2.2231,
+      "step": 4462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.931762816134517e-06,
+      "loss": 2.2218,
+      "step": 4463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.930048003192965e-06,
+      "loss": 2.2328,
+      "step": 4464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.928333076927888e-06,
+      "loss": 2.1907,
+      "step": 4465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.926618037548237e-06,
+      "loss": 2.2238,
+      "step": 4466
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.924902885262992e-06,
+      "loss": 2.2448,
+      "step": 4467
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.923187620281135e-06,
+      "loss": 2.2528,
+      "step": 4468
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9214722428116675e-06,
+      "loss": 2.1716,
+      "step": 4469
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.919756753063601e-06,
+      "loss": 2.245,
+      "step": 4470
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9180411512459655e-06,
+      "loss": 2.2539,
+      "step": 4471
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9163254375677995e-06,
+      "loss": 2.2473,
+      "step": 4472
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.914609612238159e-06,
+      "loss": 2.2542,
+      "step": 4473
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.912893675466112e-06,
+      "loss": 2.1298,
+      "step": 4474
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.911177627460739e-06,
+      "loss": 2.1393,
+      "step": 4475
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.909461468431135e-06,
+      "loss": 2.2117,
+      "step": 4476
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.907745198586411e-06,
+      "loss": 2.2471,
+      "step": 4477
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.906028818135687e-06,
+      "loss": 2.2156,
+      "step": 4478
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.904312327288101e-06,
+      "loss": 2.3107,
+      "step": 4479
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.902595726252801e-06,
+      "loss": 2.2893,
+      "step": 4480
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.900879015238948e-06,
+      "loss": 2.2409,
+      "step": 4481
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8991621944557224e-06,
+      "loss": 2.2638,
+      "step": 4482
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.897445264112309e-06,
+      "loss": 2.188,
+      "step": 4483
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8957282244179125e-06,
+      "loss": 2.2262,
+      "step": 4484
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8940110755817484e-06,
+      "loss": 2.2611,
+      "step": 4485
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.892293817813048e-06,
+      "loss": 2.2146,
+      "step": 4486
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.89057645132105e-06,
+      "loss": 2.27,
+      "step": 4487
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8888589763150165e-06,
+      "loss": 2.2008,
+      "step": 4488
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.887141393004211e-06,
+      "loss": 2.2143,
+      "step": 4489
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.885423701597918e-06,
+      "loss": 2.2854,
+      "step": 4490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.883705902305432e-06,
+      "loss": 2.2584,
+      "step": 4491
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.881987995336062e-06,
+      "loss": 2.2358,
+      "step": 4492
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.880269980899132e-06,
+      "loss": 2.2548,
+      "step": 4493
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.878551859203974e-06,
+      "loss": 2.1937,
+      "step": 4494
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.876833630459936e-06,
+      "loss": 2.2996,
+      "step": 4495
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8751152948763815e-06,
+      "loss": 2.3265,
+      "step": 4496
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.873396852662682e-06,
+      "loss": 2.2148,
+      "step": 4497
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.871678304028224e-06,
+      "loss": 2.1579,
+      "step": 4498
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.86995964918241e-06,
+      "loss": 2.2061,
+      "step": 4499
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8682408883346535e-06,
+      "loss": 2.2273,
+      "step": 4500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.866522021694376e-06,
+      "loss": 2.2434,
+      "step": 4501
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8648030494710195e-06,
+      "loss": 2.258,
+      "step": 4502
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.863083971874034e-06,
+      "loss": 2.2341,
+      "step": 4503
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8613647891128845e-06,
+      "loss": 2.2506,
+      "step": 4504
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.859645501397048e-06,
+      "loss": 2.2266,
+      "step": 4505
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.857926108936015e-06,
+      "loss": 2.1963,
+      "step": 4506
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.856206611939289e-06,
+      "loss": 2.295,
+      "step": 4507
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.854487010616384e-06,
+      "loss": 2.2413,
+      "step": 4508
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.852767305176829e-06,
+      "loss": 2.1916,
+      "step": 4509
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.851047495830163e-06,
+      "loss": 2.1857,
+      "step": 4510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.849327582785943e-06,
+      "loss": 2.2427,
+      "step": 4511
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.847607566253732e-06,
+      "loss": 2.2367,
+      "step": 4512
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8458874464431115e-06,
+      "loss": 2.2712,
+      "step": 4513
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.844167223563669e-06,
+      "loss": 2.2452,
+      "step": 4514
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.842446897825014e-06,
+      "loss": 2.267,
+      "step": 4515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.840726469436758e-06,
+      "loss": 2.2026,
+      "step": 4516
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.839005938608533e-06,
+      "loss": 2.2053,
+      "step": 4517
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.837285305549978e-06,
+      "loss": 2.1683,
+      "step": 4518
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.83556457047075e-06,
+      "loss": 2.229,
+      "step": 4519
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8338437335805124e-06,
+      "loss": 2.2121,
+      "step": 4520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8321227950889455e-06,
+      "loss": 2.2405,
+      "step": 4521
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.83040175520574e-06,
+      "loss": 2.2235,
+      "step": 4522
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.828680614140599e-06,
+      "loss": 2.1948,
+      "step": 4523
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.826959372103239e-06,
+      "loss": 2.2502,
+      "step": 4524
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.825238029303388e-06,
+      "loss": 2.2579,
+      "step": 4525
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.823516585950787e-06,
+      "loss": 2.2213,
+      "step": 4526
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.821795042255189e-06,
+      "loss": 2.1537,
+      "step": 4527
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8200733984263556e-06,
+      "loss": 2.2072,
+      "step": 4528
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.818351654674067e-06,
+      "loss": 2.187,
+      "step": 4529
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.816629811208112e-06,
+      "loss": 2.2603,
+      "step": 4530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.814907868238291e-06,
+      "loss": 2.2278,
+      "step": 4531
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.813185825974419e-06,
+      "loss": 2.2414,
+      "step": 4532
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.811463684626319e-06,
+      "loss": 2.2136,
+      "step": 4533
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.809741444403831e-06,
+      "loss": 2.2309,
+      "step": 4534
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8080191055168064e-06,
+      "loss": 2.2249,
+      "step": 4535
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8062966681751046e-06,
+      "loss": 2.2453,
+      "step": 4536
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8045741325885965e-06,
+      "loss": 2.295,
+      "step": 4537
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.802851498967173e-06,
+      "loss": 2.1896,
+      "step": 4538
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.80112876752073e-06,
+      "loss": 2.2743,
+      "step": 4539
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.799405938459175e-06,
+      "loss": 2.231,
+      "step": 4540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.797683011992432e-06,
+      "loss": 2.2263,
+      "step": 4541
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.795959988330434e-06,
+      "loss": 2.1902,
+      "step": 4542
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.794236867683125e-06,
+      "loss": 2.2559,
+      "step": 4543
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.792513650260465e-06,
+      "loss": 2.2566,
+      "step": 4544
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7907903362724195e-06,
+      "loss": 2.2472,
+      "step": 4545
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.78906692592897e-06,
+      "loss": 2.2144,
+      "step": 4546
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.787343419440108e-06,
+      "loss": 2.257,
+      "step": 4547
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.78561981701584e-06,
+      "loss": 2.1907,
+      "step": 4548
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.783896118866179e-06,
+      "loss": 2.2309,
+      "step": 4549
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.782172325201155e-06,
+      "loss": 2.164,
+      "step": 4550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.780448436230805e-06,
+      "loss": 2.284,
+      "step": 4551
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.778724452165181e-06,
+      "loss": 2.2133,
+      "step": 4552
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.777000373214345e-06,
+      "loss": 2.2214,
+      "step": 4553
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.77527619958837e-06,
+      "loss": 2.204,
+      "step": 4554
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.773551931497342e-06,
+      "loss": 2.2506,
+      "step": 4555
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.771827569151357e-06,
+      "loss": 2.244,
+      "step": 4556
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.770103112760523e-06,
+      "loss": 2.2294,
+      "step": 4557
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.768378562534962e-06,
+      "loss": 2.2131,
+      "step": 4558
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.766653918684803e-06,
+      "loss": 2.2336,
+      "step": 4559
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.764929181420191e-06,
+      "loss": 2.2524,
+      "step": 4560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.763204350951278e-06,
+      "loss": 2.1877,
+      "step": 4561
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.761479427488229e-06,
+      "loss": 2.1718,
+      "step": 4562
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7597544112412225e-06,
+      "loss": 2.2118,
+      "step": 4563
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7580293024204455e-06,
+      "loss": 2.2254,
+      "step": 4564
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7563041012360975e-06,
+      "loss": 2.2056,
+      "step": 4565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7545788078983875e-06,
+      "loss": 2.2468,
+      "step": 4566
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.752853422617539e-06,
+      "loss": 2.2066,
+      "step": 4567
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.751127945603786e-06,
+      "loss": 2.2312,
+      "step": 4568
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7494023770673705e-06,
+      "loss": 2.3144,
+      "step": 4569
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.747676717218549e-06,
+      "loss": 2.2034,
+      "step": 4570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.745950966267586e-06,
+      "loss": 2.2203,
+      "step": 4571
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.744225124424762e-06,
+      "loss": 2.244,
+      "step": 4572
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.742499191900364e-06,
+      "loss": 2.2422,
+      "step": 4573
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.740773168904691e-06,
+      "loss": 2.1795,
+      "step": 4574
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7390470556480545e-06,
+      "loss": 2.2217,
+      "step": 4575
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.737320852340776e-06,
+      "loss": 2.1938,
+      "step": 4576
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.735594559193187e-06,
+      "loss": 2.2497,
+      "step": 4577
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.733868176415633e-06,
+      "loss": 2.1664,
+      "step": 4578
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.732141704218469e-06,
+      "loss": 2.2075,
+      "step": 4579
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.730415142812059e-06,
+      "loss": 2.1796,
+      "step": 4580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.728688492406778e-06,
+      "loss": 2.2532,
+      "step": 4581
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.726961753213016e-06,
+      "loss": 2.2218,
+      "step": 4582
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.725234925441169e-06,
+      "loss": 2.2065,
+      "step": 4583
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.723508009301646e-06,
+      "loss": 2.2354,
+      "step": 4584
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.721781005004866e-06,
+      "loss": 2.1305,
+      "step": 4585
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.720053912761261e-06,
+      "loss": 2.2101,
+      "step": 4586
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.718326732781271e-06,
+      "loss": 2.2711,
+      "step": 4587
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.716599465275347e-06,
+      "loss": 2.2436,
+      "step": 4588
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.714872110453952e-06,
+      "loss": 2.296,
+      "step": 4589
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7131446685275595e-06,
+      "loss": 2.1969,
+      "step": 4590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.711417139706651e-06,
+      "loss": 2.2523,
+      "step": 4591
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.709689524201723e-06,
+      "loss": 2.2465,
+      "step": 4592
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.707961822223279e-06,
+      "loss": 2.2213,
+      "step": 4593
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.706234033981835e-06,
+      "loss": 2.2342,
+      "step": 4594
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.704506159687914e-06,
+      "loss": 2.2542,
+      "step": 4595
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.702778199552055e-06,
+      "loss": 2.2078,
+      "step": 4596
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.701050153784806e-06,
+      "loss": 2.2171,
+      "step": 4597
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6993220225967214e-06,
+      "loss": 2.2242,
+      "step": 4598
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.697593806198369e-06,
+      "loss": 2.2346,
+      "step": 4599
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.695865504800328e-06,
+      "loss": 2.2802,
+      "step": 4600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.694137118613185e-06,
+      "loss": 2.2063,
+      "step": 4601
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.692408647847542e-06,
+      "loss": 2.2184,
+      "step": 4602
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.690680092714004e-06,
+      "loss": 2.2441,
+      "step": 4603
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.68895145342319e-06,
+      "loss": 2.2407,
+      "step": 4604
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.687222730185733e-06,
+      "loss": 2.1928,
+      "step": 4605
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.685493923212273e-06,
+      "loss": 2.1854,
+      "step": 4606
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.683765032713455e-06,
+      "loss": 2.2884,
+      "step": 4607
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.682036058899942e-06,
+      "loss": 2.2171,
+      "step": 4608
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.680307001982405e-06,
+      "loss": 2.2051,
+      "step": 4609
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.678577862171523e-06,
+      "loss": 2.2352,
+      "step": 4610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.676848639677987e-06,
+      "loss": 2.3394,
+      "step": 4611
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.675119334712496e-06,
+      "loss": 2.2171,
+      "step": 4612
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.673389947485763e-06,
+      "loss": 2.313,
+      "step": 4613
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.671660478208508e-06,
+      "loss": 2.2704,
+      "step": 4614
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6699309270914615e-06,
+      "loss": 2.2418,
+      "step": 4615
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.668201294345363e-06,
+      "loss": 2.2114,
+      "step": 4616
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.666471580180963e-06,
+      "loss": 2.2586,
+      "step": 4617
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6647417848090225e-06,
+      "loss": 2.2538,
+      "step": 4618
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6630119084403125e-06,
+      "loss": 2.1992,
+      "step": 4619
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.661281951285613e-06,
+      "loss": 2.2497,
+      "step": 4620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.659551913555713e-06,
+      "loss": 2.1854,
+      "step": 4621
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.657821795461413e-06,
+      "loss": 2.2481,
+      "step": 4622
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.656091597213523e-06,
+      "loss": 2.3648,
+      "step": 4623
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.654361319022862e-06,
+      "loss": 2.2178,
+      "step": 4624
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.65263096110026e-06,
+      "loss": 2.2544,
+      "step": 4625
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.650900523656553e-06,
+      "loss": 2.2413,
+      "step": 4626
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.649170006902592e-06,
+      "loss": 2.1647,
+      "step": 4627
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.647439411049235e-06,
+      "loss": 2.2322,
+      "step": 4628
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6457087363073505e-06,
+      "loss": 2.2599,
+      "step": 4629
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.643977982887815e-06,
+      "loss": 2.2376,
+      "step": 4630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.642247151001515e-06,
+      "loss": 2.1146,
+      "step": 4631
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.640516240859348e-06,
+      "loss": 2.1809,
+      "step": 4632
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.63878525267222e-06,
+      "loss": 2.2342,
+      "step": 4633
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6370541866510476e-06,
+      "loss": 2.2438,
+      "step": 4634
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.635323043006753e-06,
+      "loss": 2.136,
+      "step": 4635
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.633591821950274e-06,
+      "loss": 2.1651,
+      "step": 4636
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.631860523692553e-06,
+      "loss": 2.2394,
+      "step": 4637
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.630129148444543e-06,
+      "loss": 2.1948,
+      "step": 4638
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.62839769641721e-06,
+      "loss": 2.2911,
+      "step": 4639
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.626666167821522e-06,
+      "loss": 2.2604,
+      "step": 4640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.624934562868463e-06,
+      "loss": 2.2521,
+      "step": 4641
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.623202881769023e-06,
+      "loss": 2.1674,
+      "step": 4642
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6214711247342015e-06,
+      "loss": 2.2325,
+      "step": 4643
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6197392919750095e-06,
+      "loss": 2.1499,
+      "step": 4644
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.618007383702464e-06,
+      "loss": 2.1897,
+      "step": 4645
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.616275400127594e-06,
+      "loss": 2.2641,
+      "step": 4646
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6145433414614345e-06,
+      "loss": 2.1921,
+      "step": 4647
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.612811207915034e-06,
+      "loss": 2.2439,
+      "step": 4648
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.611078999699448e-06,
+      "loss": 2.2002,
+      "step": 4649
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.609346717025738e-06,
+      "loss": 2.255,
+      "step": 4650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6076143601049795e-06,
+      "loss": 2.2732,
+      "step": 4651
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.605881929148254e-06,
+      "loss": 2.1802,
+      "step": 4652
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.604149424366653e-06,
+      "loss": 2.2485,
+      "step": 4653
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6024168459712765e-06,
+      "loss": 2.2553,
+      "step": 4654
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.600684194173236e-06,
+      "loss": 2.3059,
+      "step": 4655
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.598951469183649e-06,
+      "loss": 2.3031,
+      "step": 4656
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.59721867121364e-06,
+      "loss": 2.2795,
+      "step": 4657
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.59548580047435e-06,
+      "loss": 2.1731,
+      "step": 4658
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.593752857176921e-06,
+      "loss": 2.2391,
+      "step": 4659
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.592019841532507e-06,
+      "loss": 2.2464,
+      "step": 4660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.590286753752269e-06,
+      "loss": 2.2352,
+      "step": 4661
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.588553594047382e-06,
+      "loss": 2.2467,
+      "step": 4662
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.5868203626290266e-06,
+      "loss": 2.2638,
+      "step": 4663
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.585087059708389e-06,
+      "loss": 2.251,
+      "step": 4664
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.5833536854966665e-06,
+      "loss": 2.2043,
+      "step": 4665
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.581620240205068e-06,
+      "loss": 2.2069,
+      "step": 4666
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.5798867240448075e-06,
+      "loss": 2.3025,
+      "step": 4667
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.578153137227109e-06,
+      "loss": 2.2974,
+      "step": 4668
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.576419479963204e-06,
+      "loss": 2.2127,
+      "step": 4669
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.5746857524643335e-06,
+      "loss": 2.2724,
+      "step": 4670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.572951954941748e-06,
+      "loss": 2.2307,
+      "step": 4671
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.5712180876067045e-06,
+      "loss": 2.1947,
+      "step": 4672
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.56948415067047e-06,
+      "loss": 2.199,
+      "step": 4673
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.567750144344318e-06,
+      "loss": 2.2012,
+      "step": 4674
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.566016068839535e-06,
+      "loss": 2.2097,
+      "step": 4675
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.5642819243674085e-06,
+      "loss": 2.2174,
+      "step": 4676
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.562547711139243e-06,
+      "loss": 2.2428,
+      "step": 4677
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.560813429366345e-06,
+      "loss": 2.1691,
+      "step": 4678
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.559079079260032e-06,
+      "loss": 2.1984,
+      "step": 4679
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.557344661031628e-06,
+      "loss": 2.2698,
+      "step": 4680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.555610174892468e-06,
+      "loss": 2.2556,
+      "step": 4681
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.553875621053893e-06,
+      "loss": 2.2109,
+      "step": 4682
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.552140999727256e-06,
+      "loss": 2.2438,
+      "step": 4683
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.5504063111239116e-06,
+      "loss": 2.1626,
+      "step": 4684
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.548671555455226e-06,
+      "loss": 2.2384,
+      "step": 4685
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.546936732932578e-06,
+      "loss": 2.2481,
+      "step": 4686
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.545201843767348e-06,
+      "loss": 2.2401,
+      "step": 4687
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.543466888170927e-06,
+      "loss": 2.2027,
+      "step": 4688
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.541731866354713e-06,
+      "loss": 2.2526,
+      "step": 4689
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.539996778530114e-06,
+      "loss": 2.2302,
+      "step": 4690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.5382616249085476e-06,
+      "loss": 2.2486,
+      "step": 4691
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.536526405701433e-06,
+      "loss": 2.2883,
+      "step": 4692
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.534791121120205e-06,
+      "loss": 2.2522,
+      "step": 4693
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.5330557713763e-06,
+      "loss": 2.2615,
+      "step": 4694
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.5313203566811666e-06,
+      "loss": 2.2146,
+      "step": 4695
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.52958487724626e-06,
+      "loss": 2.2695,
+      "step": 4696
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.527849333283042e-06,
+      "loss": 2.2255,
+      "step": 4697
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.526113725002984e-06,
+      "loss": 2.2231,
+      "step": 4698
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.524378052617563e-06,
+      "loss": 2.2513,
+      "step": 4699
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.522642316338268e-06,
+      "loss": 2.2291,
+      "step": 4700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.520906516376592e-06,
+      "loss": 2.2197,
+      "step": 4701
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.519170652944037e-06,
+      "loss": 2.2444,
+      "step": 4702
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.517434726252113e-06,
+      "loss": 2.1846,
+      "step": 4703
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.515698736512337e-06,
+      "loss": 2.2091,
+      "step": 4704
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.513962683936235e-06,
+      "loss": 2.2302,
+      "step": 4705
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.512226568735338e-06,
+      "loss": 2.1891,
+      "step": 4706
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.510490391121188e-06,
+      "loss": 2.2641,
+      "step": 4707
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.508754151305332e-06,
+      "loss": 2.2619,
+      "step": 4708
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.507017849499326e-06,
+      "loss": 2.2768,
+      "step": 4709
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.505281485914732e-06,
+      "loss": 2.2641,
+      "step": 4710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.503545060763123e-06,
+      "loss": 2.2883,
+      "step": 4711
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5018085742560745e-06,
+      "loss": 2.1662,
+      "step": 4712
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.500072026605175e-06,
+      "loss": 2.2552,
+      "step": 4713
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.498335418022015e-06,
+      "loss": 2.2286,
+      "step": 4714
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.496598748718196e-06,
+      "loss": 2.1828,
+      "step": 4715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4948620189053255e-06,
+      "loss": 2.2472,
+      "step": 4716
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.49312522879502e-06,
+      "loss": 2.1887,
+      "step": 4717
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.491388378598899e-06,
+      "loss": 2.21,
+      "step": 4718
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.489651468528596e-06,
+      "loss": 2.1633,
+      "step": 4719
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.487914498795748e-06,
+      "loss": 2.271,
+      "step": 4720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.486177469611999e-06,
+      "loss": 2.2662,
+      "step": 4721
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.484440381188997e-06,
+      "loss": 2.2802,
+      "step": 4722
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.482703233738405e-06,
+      "loss": 2.1781,
+      "step": 4723
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.480966027471889e-06,
+      "loss": 2.2084,
+      "step": 4724
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4792287626011206e-06,
+      "loss": 2.1922,
+      "step": 4725
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.477491439337782e-06,
+      "loss": 2.2365,
+      "step": 4726
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4757540578935595e-06,
+      "loss": 2.2887,
+      "step": 4727
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.474016618480147e-06,
+      "loss": 2.1904,
+      "step": 4728
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.472279121309248e-06,
+      "loss": 2.2854,
+      "step": 4729
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.470541566592573e-06,
+      "loss": 2.2134,
+      "step": 4730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.468803954541834e-06,
+      "loss": 2.2363,
+      "step": 4731
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.467066285368754e-06,
+      "loss": 2.1832,
+      "step": 4732
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.465328559285064e-06,
+      "loss": 2.3133,
+      "step": 4733
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.463590776502501e-06,
+      "loss": 2.2691,
+      "step": 4734
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.461852937232809e-06,
+      "loss": 2.3251,
+      "step": 4735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.460115041687737e-06,
+      "loss": 2.186,
+      "step": 4736
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4583770900790415e-06,
+      "loss": 2.1732,
+      "step": 4737
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.456639082618489e-06,
+      "loss": 2.2337,
+      "step": 4738
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.454901019517851e-06,
+      "loss": 2.2293,
+      "step": 4739
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.453162900988902e-06,
+      "loss": 2.1769,
+      "step": 4740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.451424727243428e-06,
+      "loss": 2.1636,
+      "step": 4741
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.449686498493219e-06,
+      "loss": 2.2329,
+      "step": 4742
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.447948214950078e-06,
+      "loss": 2.2635,
+      "step": 4743
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.446209876825803e-06,
+      "loss": 2.2273,
+      "step": 4744
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4444714843322085e-06,
+      "loss": 2.2888,
+      "step": 4745
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.442733037681112e-06,
+      "loss": 2.2458,
+      "step": 4746
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.440994537084337e-06,
+      "loss": 2.2561,
+      "step": 4747
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.439255982753717e-06,
+      "loss": 2.1484,
+      "step": 4748
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.437517374901087e-06,
+      "loss": 2.2283,
+      "step": 4749
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.435778713738292e-06,
+      "loss": 2.2385,
+      "step": 4750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.434039999477182e-06,
+      "loss": 2.1981,
+      "step": 4751
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.432301232329615e-06,
+      "loss": 2.2361,
+      "step": 4752
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.430562412507454e-06,
+      "loss": 2.2478,
+      "step": 4753
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.42882354022257e-06,
+      "loss": 2.2246,
+      "step": 4754
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4270846156868386e-06,
+      "loss": 2.2064,
+      "step": 4755
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.425345639112141e-06,
+      "loss": 2.2881,
+      "step": 4756
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.423606610710368e-06,
+      "loss": 2.1914,
+      "step": 4757
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4218675306934145e-06,
+      "loss": 2.2067,
+      "step": 4758
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.420128399273183e-06,
+      "loss": 2.1953,
+      "step": 4759
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.41838921666158e-06,
+      "loss": 2.262,
+      "step": 4760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.416649983070518e-06,
+      "loss": 2.2366,
+      "step": 4761
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.41491069871192e-06,
+      "loss": 2.2465,
+      "step": 4762
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.413171363797713e-06,
+      "loss": 2.2149,
+      "step": 4763
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.411431978539829e-06,
+      "loss": 2.1713,
+      "step": 4764
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.409692543150206e-06,
+      "loss": 2.2917,
+      "step": 4765
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4079530578407895e-06,
+      "loss": 2.1868,
+      "step": 4766
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.406213522823529e-06,
+      "loss": 2.221,
+      "step": 4767
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.404473938310384e-06,
+      "loss": 2.2384,
+      "step": 4768
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.402734304513316e-06,
+      "loss": 2.2632,
+      "step": 4769
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.400994621644294e-06,
+      "loss": 2.2602,
+      "step": 4770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.399254889915296e-06,
+      "loss": 2.2039,
+      "step": 4771
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3975151095383e-06,
+      "loss": 2.2954,
+      "step": 4772
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3957752807252925e-06,
+      "loss": 2.2213,
+      "step": 4773
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.394035403688268e-06,
+      "loss": 2.2539,
+      "step": 4774
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.392295478639226e-06,
+      "loss": 2.2395,
+      "step": 4775
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.390555505790168e-06,
+      "loss": 2.1943,
+      "step": 4776
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.388815485353109e-06,
+      "loss": 2.2596,
+      "step": 4777
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3870754175400595e-06,
+      "loss": 2.2312,
+      "step": 4778
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.385335302563046e-06,
+      "loss": 2.1635,
+      "step": 4779
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.383595140634093e-06,
+      "loss": 2.2619,
+      "step": 4780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.381854931965238e-06,
+      "loss": 2.2274,
+      "step": 4781
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.380114676768516e-06,
+      "loss": 2.2325,
+      "step": 4782
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.378374375255977e-06,
+      "loss": 2.2791,
+      "step": 4783
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.376634027639664e-06,
+      "loss": 2.238,
+      "step": 4784
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3748936341316395e-06,
+      "loss": 2.2161,
+      "step": 4785
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.373153194943962e-06,
+      "loss": 2.2007,
+      "step": 4786
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.371412710288701e-06,
+      "loss": 2.2338,
+      "step": 4787
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3696721803779265e-06,
+      "loss": 2.2326,
+      "step": 4788
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3679316054237165e-06,
+      "loss": 2.1872,
+      "step": 4789
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.366190985638159e-06,
+      "loss": 2.2276,
+      "step": 4790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3644503212333395e-06,
+      "loss": 2.2824,
+      "step": 4791
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.362709612421355e-06,
+      "loss": 2.1614,
+      "step": 4792
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.360968859414305e-06,
+      "loss": 2.178,
+      "step": 4793
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.359228062424292e-06,
+      "loss": 2.1041,
+      "step": 4794
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.35748722166343e-06,
+      "loss": 2.2509,
+      "step": 4795
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.355746337343835e-06,
+      "loss": 2.2122,
+      "step": 4796
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.354005409677628e-06,
+      "loss": 2.2109,
+      "step": 4797
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.352264438876935e-06,
+      "loss": 2.1734,
+      "step": 4798
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3505234251538885e-06,
+      "loss": 2.2204,
+      "step": 4799
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.348782368720627e-06,
+      "loss": 2.2627,
+      "step": 4800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.347041269789293e-06,
+      "loss": 2.2008,
+      "step": 4801
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.345300128572031e-06,
+      "loss": 2.2465,
+      "step": 4802
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.343558945280994e-06,
+      "loss": 2.1748,
+      "step": 4803
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.341817720128344e-06,
+      "loss": 2.2608,
+      "step": 4804
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3400764533262415e-06,
+      "loss": 2.1608,
+      "step": 4805
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.338335145086855e-06,
+      "loss": 2.157,
+      "step": 4806
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.336593795622357e-06,
+      "loss": 2.2003,
+      "step": 4807
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.334852405144926e-06,
+      "loss": 2.255,
+      "step": 4808
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.333110973866743e-06,
+      "loss": 2.2259,
+      "step": 4809
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3313695020000026e-06,
+      "loss": 2.2586,
+      "step": 4810
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.32962798975689e-06,
+      "loss": 2.229,
+      "step": 4811
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.327886437349609e-06,
+      "loss": 2.1683,
+      "step": 4812
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.326144844990357e-06,
+      "loss": 2.1898,
+      "step": 4813
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3244032128913485e-06,
+      "loss": 2.2634,
+      "step": 4814
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.322661541264791e-06,
+      "loss": 2.2321,
+      "step": 4815
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.320919830322903e-06,
+      "loss": 2.2107,
+      "step": 4816
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.319178080277908e-06,
+      "loss": 2.3481,
+      "step": 4817
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.317436291342031e-06,
+      "loss": 2.1919,
+      "step": 4818
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.315694463727506e-06,
+      "loss": 2.1901,
+      "step": 4819
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3139525976465675e-06,
+      "loss": 2.215,
+      "step": 4820
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.312210693311458e-06,
+      "loss": 2.1949,
+      "step": 4821
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.310468750934421e-06,
+      "loss": 2.2524,
+      "step": 4822
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.30872677072771e-06,
+      "loss": 2.1965,
+      "step": 4823
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.306984752903578e-06,
+      "loss": 2.2046,
+      "step": 4824
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3052426976742855e-06,
+      "loss": 2.2676,
+      "step": 4825
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.303500605252095e-06,
+      "loss": 2.223,
+      "step": 4826
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.301758475849278e-06,
+      "loss": 2.2436,
+      "step": 4827
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.300016309678104e-06,
+      "loss": 2.2272,
+      "step": 4828
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.298274106950855e-06,
+      "loss": 2.1827,
+      "step": 4829
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.296531867879809e-06,
+      "loss": 2.221,
+      "step": 4830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.294789592677255e-06,
+      "loss": 2.2529,
+      "step": 4831
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.293047281555482e-06,
+      "loss": 2.2458,
+      "step": 4832
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.291304934726789e-06,
+      "loss": 2.2308,
+      "step": 4833
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.289562552403472e-06,
+      "loss": 2.2502,
+      "step": 4834
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.287820134797837e-06,
+      "loss": 2.2473,
+      "step": 4835
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2860776821221915e-06,
+      "loss": 2.2779,
+      "step": 4836
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.284335194588848e-06,
+      "loss": 2.2376,
+      "step": 4837
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.282592672410124e-06,
+      "loss": 2.2085,
+      "step": 4838
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.28085011579834e-06,
+      "loss": 2.2982,
+      "step": 4839
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.27910752496582e-06,
+      "loss": 2.2177,
+      "step": 4840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2773649001248965e-06,
+      "loss": 2.2144,
+      "step": 4841
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.275622241487899e-06,
+      "loss": 2.2423,
+      "step": 4842
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.273879549267168e-06,
+      "loss": 2.2265,
+      "step": 4843
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.272136823675046e-06,
+      "loss": 2.2389,
+      "step": 4844
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.270394064923878e-06,
+      "loss": 2.2051,
+      "step": 4845
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.268651273226011e-06,
+      "loss": 2.1765,
+      "step": 4846
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2669084487938025e-06,
+      "loss": 2.2005,
+      "step": 4847
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.26516559183961e-06,
+      "loss": 2.2113,
+      "step": 4848
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.263422702575793e-06,
+      "loss": 2.2366,
+      "step": 4849
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2616797812147205e-06,
+      "loss": 2.1913,
+      "step": 4850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.259936827968758e-06,
+      "loss": 2.181,
+      "step": 4851
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.258193843050283e-06,
+      "loss": 2.1999,
+      "step": 4852
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.256450826671671e-06,
+      "loss": 2.2633,
+      "step": 4853
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.254707779045305e-06,
+      "loss": 2.2266,
+      "step": 4854
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.252964700383567e-06,
+      "loss": 2.2054,
+      "step": 4855
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.251221590898848e-06,
+      "loss": 2.194,
+      "step": 4856
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.249478450803541e-06,
+      "loss": 2.2441,
+      "step": 4857
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.247735280310041e-06,
+      "loss": 2.1454,
+      "step": 4858
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.245992079630748e-06,
+      "loss": 2.2284,
+      "step": 4859
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.244248848978067e-06,
+      "loss": 2.182,
+      "step": 4860
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.242505588564404e-06,
+      "loss": 2.1945,
+      "step": 4861
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.240762298602171e-06,
+      "loss": 2.2055,
+      "step": 4862
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.239018979303784e-06,
+      "loss": 2.2017,
+      "step": 4863
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.237275630881658e-06,
+      "loss": 2.2455,
+      "step": 4864
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.235532253548213e-06,
+      "loss": 2.2357,
+      "step": 4865
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.233788847515882e-06,
+      "loss": 2.2718,
+      "step": 4866
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2320454129970866e-06,
+      "loss": 2.2349,
+      "step": 4867
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.230301950204261e-06,
+      "loss": 2.2513,
+      "step": 4868
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.228558459349844e-06,
+      "loss": 2.2771,
+      "step": 4869
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.226814940646268e-06,
+      "loss": 2.2256,
+      "step": 4870
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2250713943059826e-06,
+      "loss": 2.2119,
+      "step": 4871
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.223327820541432e-06,
+      "loss": 2.2542,
+      "step": 4872
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.221584219565061e-06,
+      "loss": 2.2488,
+      "step": 4873
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.219840591589325e-06,
+      "loss": 2.2156,
+      "step": 4874
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.218096936826681e-06,
+      "loss": 2.2914,
+      "step": 4875
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.216353255489586e-06,
+      "loss": 2.274,
+      "step": 4876
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.214609547790504e-06,
+      "loss": 2.2528,
+      "step": 4877
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.212865813941899e-06,
+      "loss": 2.2403,
+      "step": 4878
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.21112205415624e-06,
+      "loss": 2.1842,
+      "step": 4879
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.209378268645998e-06,
+      "loss": 2.2137,
+      "step": 4880
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.207634457623652e-06,
+      "loss": 2.2252,
+      "step": 4881
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.205890621301676e-06,
+      "loss": 2.1993,
+      "step": 4882
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.204146759892551e-06,
+      "loss": 2.2208,
+      "step": 4883
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.202402873608763e-06,
+      "loss": 2.3027,
+      "step": 4884
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.200658962662799e-06,
+      "loss": 2.2124,
+      "step": 4885
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.19891502726715e-06,
+      "loss": 2.2265,
+      "step": 4886
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.197171067634307e-06,
+      "loss": 2.2153,
+      "step": 4887
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.195427083976768e-06,
+      "loss": 2.2085,
+      "step": 4888
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.193683076507031e-06,
+      "loss": 2.1937,
+      "step": 4889
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1919390454376e-06,
+      "loss": 2.2227,
+      "step": 4890
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.190194990980979e-06,
+      "loss": 2.2193,
+      "step": 4891
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.188450913349674e-06,
+      "loss": 2.1655,
+      "step": 4892
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.186706812756197e-06,
+      "loss": 2.2366,
+      "step": 4893
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.18496268941306e-06,
+      "loss": 2.1995,
+      "step": 4894
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.183218543532782e-06,
+      "loss": 2.1251,
+      "step": 4895
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.18147437532788e-06,
+      "loss": 2.2648,
+      "step": 4896
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.179730185010875e-06,
+      "loss": 2.3035,
+      "step": 4897
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.177985972794293e-06,
+      "loss": 2.262,
+      "step": 4898
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.17624173889066e-06,
+      "loss": 2.2145,
+      "step": 4899
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.174497483512506e-06,
+      "loss": 2.2546,
+      "step": 4900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.172753206872363e-06,
+      "loss": 2.3246,
+      "step": 4901
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.171008909182765e-06,
+      "loss": 2.2622,
+      "step": 4902
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.169264590656249e-06,
+      "loss": 2.2277,
+      "step": 4903
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.167520251505358e-06,
+      "loss": 2.2174,
+      "step": 4904
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.165775891942631e-06,
+      "loss": 2.1746,
+      "step": 4905
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.164031512180616e-06,
+      "loss": 2.2091,
+      "step": 4906
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.162287112431858e-06,
+      "loss": 2.1859,
+      "step": 4907
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.160542692908909e-06,
+      "loss": 2.2618,
+      "step": 4908
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.158798253824319e-06,
+      "loss": 2.1803,
+      "step": 4909
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.157053795390642e-06,
+      "loss": 2.1613,
+      "step": 4910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.155309317820438e-06,
+      "loss": 2.2584,
+      "step": 4911
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.153564821326265e-06,
+      "loss": 2.2242,
+      "step": 4912
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.151820306120682e-06,
+      "loss": 2.1486,
+      "step": 4913
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.150075772416256e-06,
+      "loss": 2.1766,
+      "step": 4914
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.148331220425554e-06,
+      "loss": 2.2272,
+      "step": 4915
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.146586650361143e-06,
+      "loss": 2.238,
+      "step": 4916
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.144842062435591e-06,
+      "loss": 2.1808,
+      "step": 4917
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.143097456861474e-06,
+      "loss": 2.1627,
+      "step": 4918
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.141352833851367e-06,
+      "loss": 2.2567,
+      "step": 4919
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.139608193617846e-06,
+      "loss": 2.2334,
+      "step": 4920
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1378635363734884e-06,
+      "loss": 2.2201,
+      "step": 4921
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.136118862330876e-06,
+      "loss": 2.2265,
+      "step": 4922
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.134374171702596e-06,
+      "loss": 2.2002,
+      "step": 4923
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.13262946470123e-06,
+      "loss": 2.1417,
+      "step": 4924
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.130884741539367e-06,
+      "loss": 2.2066,
+      "step": 4925
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1291400024295946e-06,
+      "loss": 2.3189,
+      "step": 4926
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.127395247584503e-06,
+      "loss": 2.2441,
+      "step": 4927
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1256504772166885e-06,
+      "loss": 2.2078,
+      "step": 4928
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.123905691538744e-06,
+      "loss": 2.168,
+      "step": 4929
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1221608907632665e-06,
+      "loss": 2.2465,
+      "step": 4930
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.120416075102855e-06,
+      "loss": 2.2199,
+      "step": 4931
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.118671244770111e-06,
+      "loss": 2.2188,
+      "step": 4932
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.116926399977634e-06,
+      "loss": 2.2529,
+      "step": 4933
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.115181540938032e-06,
+      "loss": 2.2442,
+      "step": 4934
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.113436667863908e-06,
+      "loss": 2.2531,
+      "step": 4935
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.111691780967869e-06,
+      "loss": 2.1627,
+      "step": 4936
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.109946880462526e-06,
+      "loss": 2.2338,
+      "step": 4937
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1082019665604895e-06,
+      "loss": 2.2889,
+      "step": 4938
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1064570394743705e-06,
+      "loss": 2.1988,
+      "step": 4939
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1047120994167855e-06,
+      "loss": 2.1919,
+      "step": 4940
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.102967146600348e-06,
+      "loss": 2.244,
+      "step": 4941
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.101222181237676e-06,
+      "loss": 2.1716,
+      "step": 4942
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.09947720354139e-06,
+      "loss": 2.2906,
+      "step": 4943
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.097732213724107e-06,
+      "loss": 2.1912,
+      "step": 4944
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.09598721199845e-06,
+      "loss": 2.2999,
+      "step": 4945
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.0942421985770415e-06,
+      "loss": 2.2275,
+      "step": 4946
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.092497173672508e-06,
+      "loss": 2.2229,
+      "step": 4947
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.090752137497474e-06,
+      "loss": 2.2621,
+      "step": 4948
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.089007090264568e-06,
+      "loss": 2.1619,
+      "step": 4949
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.087262032186418e-06,
+      "loss": 2.2913,
+      "step": 4950
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.085516963475653e-06,
+      "loss": 2.1847,
+      "step": 4951
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.083771884344908e-06,
+      "loss": 2.2083,
+      "step": 4952
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.0820267950068115e-06,
+      "loss": 2.2936,
+      "step": 4953
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.080281695673999e-06,
+      "loss": 2.2761,
+      "step": 4954
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.078536586559104e-06,
+      "loss": 2.1958,
+      "step": 4955
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.0767914678747655e-06,
+      "loss": 2.23,
+      "step": 4956
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.0750463398336195e-06,
+      "loss": 2.2476,
+      "step": 4957
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.073301202648304e-06,
+      "loss": 2.2281,
+      "step": 4958
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.0715560565314595e-06,
+      "loss": 2.2485,
+      "step": 4959
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.069810901695727e-06,
+      "loss": 2.2033,
+      "step": 4960
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.068065738353748e-06,
+      "loss": 2.2128,
+      "step": 4961
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.066320566718165e-06,
+      "loss": 2.2043,
+      "step": 4962
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.0645753870016224e-06,
+      "loss": 2.2305,
+      "step": 4963
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.062830199416764e-06,
+      "loss": 2.2547,
+      "step": 4964
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.061085004176236e-06,
+      "loss": 2.2308,
+      "step": 4965
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.059339801492687e-06,
+      "loss": 2.1452,
+      "step": 4966
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.0575945915787616e-06,
+      "loss": 2.2283,
+      "step": 4967
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.055849374647112e-06,
+      "loss": 2.2579,
+      "step": 4968
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.054104150910384e-06,
+      "loss": 2.1888,
+      "step": 4969
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.05235892058123e-06,
+      "loss": 2.289,
+      "step": 4970
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.050613683872299e-06,
+      "loss": 2.1908,
+      "step": 4971
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.048868440996246e-06,
+      "loss": 2.2422,
+      "step": 4972
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.047123192165722e-06,
+      "loss": 2.2266,
+      "step": 4973
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.045377937593376e-06,
+      "loss": 2.1637,
+      "step": 4974
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.04363267749187e-06,
+      "loss": 2.1639,
+      "step": 4975
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.041887412073853e-06,
+      "loss": 2.163,
+      "step": 4976
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.040142141551982e-06,
+      "loss": 2.1949,
+      "step": 4977
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.038396866138915e-06,
+      "loss": 2.1909,
+      "step": 4978
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.036651586047303e-06,
+      "loss": 2.2076,
+      "step": 4979
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.034906301489808e-06,
+      "loss": 2.2707,
+      "step": 4980
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.033161012679087e-06,
+      "loss": 2.2175,
+      "step": 4981
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.031415719827796e-06,
+      "loss": 2.2174,
+      "step": 4982
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.029670423148595e-06,
+      "loss": 2.1714,
+      "step": 4983
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.027925122854141e-06,
+      "loss": 2.1997,
+      "step": 4984
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.0261798191570975e-06,
+      "loss": 2.1501,
+      "step": 4985
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.024434512270123e-06,
+      "loss": 2.2631,
+      "step": 4986
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.0226892024058785e-06,
+      "loss": 2.2114,
+      "step": 4987
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0209438897770205e-06,
+      "loss": 2.2332,
+      "step": 4988
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.019198574596213e-06,
+      "loss": 2.1891,
+      "step": 4989
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0174532570761194e-06,
+      "loss": 2.2468,
+      "step": 4990
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.015707937429398e-06,
+      "loss": 2.2167,
+      "step": 4991
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.013962615868714e-06,
+      "loss": 2.2841,
+      "step": 4992
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.012217292606726e-06,
+      "loss": 2.1776,
+      "step": 4993
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.010471967856096e-06,
+      "loss": 2.2258,
+      "step": 4994
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.008726641829492e-06,
+      "loss": 2.1393,
+      "step": 4995
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.006981314739573e-06,
+      "loss": 2.1918,
+      "step": 4996
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.005235986799001e-06,
+      "loss": 2.2068,
+      "step": 4997
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.003490658220438e-06,
+      "loss": 2.2191,
+      "step": 4998
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.001745329216551e-06,
+      "loss": 2.2368,
+      "step": 4999
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5e-06,
+      "loss": 2.1916,
+      "step": 5000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9982546707834514e-06,
+      "loss": 2.2063,
+      "step": 5001
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.996509341779563e-06,
+      "loss": 2.2058,
+      "step": 5002
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.994764013201002e-06,
+      "loss": 2.2254,
+      "step": 5003
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.993018685260428e-06,
+      "loss": 2.1927,
+      "step": 5004
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9912733581705095e-06,
+      "loss": 2.2388,
+      "step": 5005
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9895280321439036e-06,
+      "loss": 2.2,
+      "step": 5006
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.987782707393276e-06,
+      "loss": 2.1986,
+      "step": 5007
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.986037384131288e-06,
+      "loss": 2.228,
+      "step": 5008
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.984292062570603e-06,
+      "loss": 2.1705,
+      "step": 5009
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.982546742923883e-06,
+      "loss": 2.2742,
+      "step": 5010
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.980801425403788e-06,
+      "loss": 2.1968,
+      "step": 5011
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.979056110222982e-06,
+      "loss": 2.1886,
+      "step": 5012
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.977310797594124e-06,
+      "loss": 2.1766,
+      "step": 5013
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.975565487729879e-06,
+      "loss": 2.2606,
+      "step": 5014
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9738201808429025e-06,
+      "loss": 2.2637,
+      "step": 5015
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9720748771458595e-06,
+      "loss": 2.2278,
+      "step": 5016
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.970329576851406e-06,
+      "loss": 2.1934,
+      "step": 5017
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.968584280172206e-06,
+      "loss": 2.2083,
+      "step": 5018
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.966838987320916e-06,
+      "loss": 2.188,
+      "step": 5019
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.965093698510192e-06,
+      "loss": 2.2166,
+      "step": 5020
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9633484139526975e-06,
+      "loss": 2.2222,
+      "step": 5021
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.961603133861086e-06,
+      "loss": 2.2244,
+      "step": 5022
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9598578584480186e-06,
+      "loss": 2.1359,
+      "step": 5023
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9581125879261476e-06,
+      "loss": 2.2529,
+      "step": 5024
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.956367322508131e-06,
+      "loss": 2.2301,
+      "step": 5025
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.954622062406623e-06,
+      "loss": 2.2593,
+      "step": 5026
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.952876807834281e-06,
+      "loss": 2.2319,
+      "step": 5027
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.951131559003756e-06,
+      "loss": 2.1842,
+      "step": 5028
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9493863161277016e-06,
+      "loss": 2.2165,
+      "step": 5029
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9476410794187726e-06,
+      "loss": 2.1223,
+      "step": 5030
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.945895849089618e-06,
+      "loss": 2.1777,
+      "step": 5031
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.94415062535289e-06,
+      "loss": 2.1628,
+      "step": 5032
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.942405408421238e-06,
+      "loss": 2.2915,
+      "step": 5033
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.940660198507315e-06,
+      "loss": 2.1927,
+      "step": 5034
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.938914995823764e-06,
+      "loss": 2.2387,
+      "step": 5035
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.937169800583237e-06,
+      "loss": 2.2248,
+      "step": 5036
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.93542461299838e-06,
+      "loss": 2.1578,
+      "step": 5037
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.933679433281837e-06,
+      "loss": 2.2916,
+      "step": 5038
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.931934261646255e-06,
+      "loss": 2.2775,
+      "step": 5039
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9301890983042744e-06,
+      "loss": 2.2087,
+      "step": 5040
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.928443943468541e-06,
+      "loss": 2.2523,
+      "step": 5041
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.926698797351697e-06,
+      "loss": 2.1839,
+      "step": 5042
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.924953660166383e-06,
+      "loss": 2.2659,
+      "step": 5043
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.923208532125235e-06,
+      "loss": 2.202,
+      "step": 5044
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.921463413440898e-06,
+      "loss": 2.2232,
+      "step": 5045
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9197183043260035e-06,
+      "loss": 2.2004,
+      "step": 5046
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.91797320499319e-06,
+      "loss": 2.2698,
+      "step": 5047
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9162281156550945e-06,
+      "loss": 2.2502,
+      "step": 5048
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9144830365243464e-06,
+      "loss": 2.2401,
+      "step": 5049
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9127379678135825e-06,
+      "loss": 2.3162,
+      "step": 5050
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.910992909735432e-06,
+      "loss": 2.2503,
+      "step": 5051
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9092478625025266e-06,
+      "loss": 2.2046,
+      "step": 5052
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9075028263274925e-06,
+      "loss": 2.2562,
+      "step": 5053
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.90575780142296e-06,
+      "loss": 2.2712,
+      "step": 5054
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.904012788001553e-06,
+      "loss": 2.2786,
+      "step": 5055
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.902267786275895e-06,
+      "loss": 2.2046,
+      "step": 5056
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.900522796458613e-06,
+      "loss": 2.1748,
+      "step": 5057
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.898777818762325e-06,
+      "loss": 2.1797,
+      "step": 5058
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.897032853399653e-06,
+      "loss": 2.1734,
+      "step": 5059
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.895287900583216e-06,
+      "loss": 2.2688,
+      "step": 5060
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.89354296052563e-06,
+      "loss": 2.1826,
+      "step": 5061
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.891798033439511e-06,
+      "loss": 2.1917,
+      "step": 5062
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.890053119537475e-06,
+      "loss": 2.1431,
+      "step": 5063
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.888308219032133e-06,
+      "loss": 2.2257,
+      "step": 5064
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.886563332136093e-06,
+      "loss": 2.2724,
+      "step": 5065
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.88481845906197e-06,
+      "loss": 2.2359,
+      "step": 5066
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.883073600022366e-06,
+      "loss": 2.2148,
+      "step": 5067
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.881328755229892e-06,
+      "loss": 2.2626,
+      "step": 5068
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.879583924897146e-06,
+      "loss": 2.2464,
+      "step": 5069
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.877839109236735e-06,
+      "loss": 2.247,
+      "step": 5070
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8760943084612585e-06,
+      "loss": 2.2084,
+      "step": 5071
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.874349522783313e-06,
+      "loss": 2.2878,
+      "step": 5072
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8726047524154985e-06,
+      "loss": 2.2611,
+      "step": 5073
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.870859997570407e-06,
+      "loss": 2.1407,
+      "step": 5074
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.869115258460636e-06,
+      "loss": 2.2124,
+      "step": 5075
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.86737053529877e-06,
+      "loss": 2.2283,
+      "step": 5076
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.865625828297405e-06,
+      "loss": 2.1363,
+      "step": 5077
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.863881137669123e-06,
+      "loss": 2.2383,
+      "step": 5078
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.862136463626512e-06,
+      "loss": 2.2074,
+      "step": 5079
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.860391806382157e-06,
+      "loss": 2.1944,
+      "step": 5080
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8586471661486345e-06,
+      "loss": 2.2588,
+      "step": 5081
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.856902543138528e-06,
+      "loss": 2.1619,
+      "step": 5082
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.85515793756441e-06,
+      "loss": 2.2235,
+      "step": 5083
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.853413349638859e-06,
+      "loss": 2.2221,
+      "step": 5084
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.851668779574446e-06,
+      "loss": 2.2132,
+      "step": 5085
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8499242275837444e-06,
+      "loss": 2.1699,
+      "step": 5086
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.848179693879319e-06,
+      "loss": 2.1855,
+      "step": 5087
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.846435178673737e-06,
+      "loss": 2.2154,
+      "step": 5088
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8446906821795645e-06,
+      "loss": 2.2764,
+      "step": 5089
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.842946204609359e-06,
+      "loss": 2.1898,
+      "step": 5090
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8412017461756845e-06,
+      "loss": 2.2373,
+      "step": 5091
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.839457307091093e-06,
+      "loss": 2.187,
+      "step": 5092
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.837712887568143e-06,
+      "loss": 2.2529,
+      "step": 5093
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.835968487819384e-06,
+      "loss": 2.1902,
+      "step": 5094
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8342241080573696e-06,
+      "loss": 2.2336,
+      "step": 5095
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.832479748494643e-06,
+      "loss": 2.2079,
+      "step": 5096
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.830735409343752e-06,
+      "loss": 2.1713,
+      "step": 5097
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.828991090817238e-06,
+      "loss": 2.2622,
+      "step": 5098
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.827246793127639e-06,
+      "loss": 2.2163,
+      "step": 5099
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.825502516487497e-06,
+      "loss": 2.2728,
+      "step": 5100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.823758261109341e-06,
+      "loss": 2.2218,
+      "step": 5101
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.822014027205708e-06,
+      "loss": 2.2303,
+      "step": 5102
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.820269814989125e-06,
+      "loss": 2.1828,
+      "step": 5103
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.818525624672122e-06,
+      "loss": 2.1541,
+      "step": 5104
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.816781456467218e-06,
+      "loss": 2.2335,
+      "step": 5105
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.815037310586941e-06,
+      "loss": 2.1412,
+      "step": 5106
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.813293187243806e-06,
+      "loss": 2.2182,
+      "step": 5107
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.811549086650327e-06,
+      "loss": 2.2262,
+      "step": 5108
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.809805009019024e-06,
+      "loss": 2.2032,
+      "step": 5109
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8080609545624004e-06,
+      "loss": 2.2108,
+      "step": 5110
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.80631692349297e-06,
+      "loss": 2.2304,
+      "step": 5111
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8045729160232326e-06,
+      "loss": 2.2621,
+      "step": 5112
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.802828932365694e-06,
+      "loss": 2.2377,
+      "step": 5113
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.801084972732851e-06,
+      "loss": 2.2083,
+      "step": 5114
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.799341037337203e-06,
+      "loss": 2.215,
+      "step": 5115
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.797597126391238e-06,
+      "loss": 2.2488,
+      "step": 5116
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.79585324010745e-06,
+      "loss": 2.2624,
+      "step": 5117
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.794109378698327e-06,
+      "loss": 2.1653,
+      "step": 5118
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.79236554237635e-06,
+      "loss": 2.1801,
+      "step": 5119
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7906217313540035e-06,
+      "loss": 2.2472,
+      "step": 5120
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.788877945843762e-06,
+      "loss": 2.2057,
+      "step": 5121
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.787134186058103e-06,
+      "loss": 2.23,
+      "step": 5122
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.785390452209497e-06,
+      "loss": 2.2718,
+      "step": 5123
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.783646744510416e-06,
+      "loss": 2.2848,
+      "step": 5124
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.781903063173321e-06,
+      "loss": 2.2566,
+      "step": 5125
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.780159408410677e-06,
+      "loss": 2.2328,
+      "step": 5126
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.778415780434941e-06,
+      "loss": 2.2403,
+      "step": 5127
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7766721794585704e-06,
+      "loss": 2.2864,
+      "step": 5128
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.774928605694018e-06,
+      "loss": 2.2858,
+      "step": 5129
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7731850593537316e-06,
+      "loss": 2.2875,
+      "step": 5130
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.771441540650158e-06,
+      "loss": 2.2079,
+      "step": 5131
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.769698049795739e-06,
+      "loss": 2.2329,
+      "step": 5132
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.767954587002915e-06,
+      "loss": 2.2124,
+      "step": 5133
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.766211152484122e-06,
+      "loss": 2.2248,
+      "step": 5134
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7644677464517874e-06,
+      "loss": 2.2347,
+      "step": 5135
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.762724369118346e-06,
+      "loss": 2.2244,
+      "step": 5136
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.760981020696218e-06,
+      "loss": 2.2378,
+      "step": 5137
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.759237701397831e-06,
+      "loss": 2.2355,
+      "step": 5138
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.757494411435597e-06,
+      "loss": 2.2568,
+      "step": 5139
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.755751151021934e-06,
+      "loss": 2.206,
+      "step": 5140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.754007920369252e-06,
+      "loss": 2.1919,
+      "step": 5141
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.752264719689961e-06,
+      "loss": 2.2449,
+      "step": 5142
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7505215491964615e-06,
+      "loss": 2.2241,
+      "step": 5143
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.748778409101153e-06,
+      "loss": 2.2006,
+      "step": 5144
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.747035299616434e-06,
+      "loss": 2.2897,
+      "step": 5145
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.745292220954696e-06,
+      "loss": 2.1723,
+      "step": 5146
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.74354917332833e-06,
+      "loss": 2.2694,
+      "step": 5147
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.741806156949718e-06,
+      "loss": 2.233,
+      "step": 5148
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.740063172031243e-06,
+      "loss": 2.2315,
+      "step": 5149
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.738320218785281e-06,
+      "loss": 2.2301,
+      "step": 5150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7365772974242075e-06,
+      "loss": 2.2414,
+      "step": 5151
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.734834408160393e-06,
+      "loss": 2.1581,
+      "step": 5152
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.733091551206198e-06,
+      "loss": 2.2315,
+      "step": 5153
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.73134872677399e-06,
+      "loss": 2.1717,
+      "step": 5154
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.729605935076123e-06,
+      "loss": 2.1848,
+      "step": 5155
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.727863176324955e-06,
+      "loss": 2.1725,
+      "step": 5156
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7261204507328315e-06,
+      "loss": 2.2707,
+      "step": 5157
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7243777585121034e-06,
+      "loss": 2.157,
+      "step": 5158
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.722635099875106e-06,
+      "loss": 2.1741,
+      "step": 5159
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.720892475034181e-06,
+      "loss": 2.2588,
+      "step": 5160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.719149884201664e-06,
+      "loss": 2.1338,
+      "step": 5161
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.717407327589878e-06,
+      "loss": 2.2687,
+      "step": 5162
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.715664805411155e-06,
+      "loss": 2.2216,
+      "step": 5163
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.71392231787781e-06,
+      "loss": 2.2573,
+      "step": 5164
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.712179865202164e-06,
+      "loss": 2.1677,
+      "step": 5165
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.710437447596528e-06,
+      "loss": 2.215,
+      "step": 5166
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.708695065273213e-06,
+      "loss": 2.2986,
+      "step": 5167
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.706952718444518e-06,
+      "loss": 2.2412,
+      "step": 5168
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.705210407322746e-06,
+      "loss": 2.24,
+      "step": 5169
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.703468132120193e-06,
+      "loss": 2.1733,
+      "step": 5170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7017258930491474e-06,
+      "loss": 2.213,
+      "step": 5171
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.699983690321898e-06,
+      "loss": 2.2308,
+      "step": 5172
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.698241524150724e-06,
+      "loss": 2.2293,
+      "step": 5173
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.696499394747906e-06,
+      "loss": 2.243,
+      "step": 5174
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.694757302325715e-06,
+      "loss": 2.1947,
+      "step": 5175
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.693015247096423e-06,
+      "loss": 2.204,
+      "step": 5176
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.691273229272291e-06,
+      "loss": 2.1884,
+      "step": 5177
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.689531249065581e-06,
+      "loss": 2.178,
+      "step": 5178
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.687789306688544e-06,
+      "loss": 2.2302,
+      "step": 5179
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.686047402353433e-06,
+      "loss": 2.2063,
+      "step": 5180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.684305536272496e-06,
+      "loss": 2.252,
+      "step": 5181
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.68256370865797e-06,
+      "loss": 2.2061,
+      "step": 5182
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.680821919722094e-06,
+      "loss": 2.2581,
+      "step": 5183
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.679080169677097e-06,
+      "loss": 2.2491,
+      "step": 5184
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.677338458735211e-06,
+      "loss": 2.3249,
+      "step": 5185
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.675596787108652e-06,
+      "loss": 2.3016,
+      "step": 5186
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.673855155009644e-06,
+      "loss": 2.1948,
+      "step": 5187
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.672113562650394e-06,
+      "loss": 2.1946,
+      "step": 5188
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.670372010243111e-06,
+      "loss": 2.2335,
+      "step": 5189
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.668630498000001e-06,
+      "loss": 2.2357,
+      "step": 5190
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.666889026133257e-06,
+      "loss": 2.1965,
+      "step": 5191
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6651475948550765e-06,
+      "loss": 2.2653,
+      "step": 5192
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.663406204377645e-06,
+      "loss": 2.1774,
+      "step": 5193
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.661664854913147e-06,
+      "loss": 2.1682,
+      "step": 5194
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.659923546673761e-06,
+      "loss": 2.1731,
+      "step": 5195
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.658182279871657e-06,
+      "loss": 2.2449,
+      "step": 5196
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.656441054719007e-06,
+      "loss": 2.2206,
+      "step": 5197
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.654699871427972e-06,
+      "loss": 2.2267,
+      "step": 5198
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.652958730210711e-06,
+      "loss": 2.1848,
+      "step": 5199
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.651217631279374e-06,
+      "loss": 2.2847,
+      "step": 5200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.649476574846113e-06,
+      "loss": 2.2239,
+      "step": 5201
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6477355611230655e-06,
+      "loss": 2.1881,
+      "step": 5202
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.645994590322373e-06,
+      "loss": 2.194,
+      "step": 5203
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.644253662656167e-06,
+      "loss": 2.2802,
+      "step": 5204
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.642512778336571e-06,
+      "loss": 2.2363,
+      "step": 5205
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6407719375757095e-06,
+      "loss": 2.2564,
+      "step": 5206
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.639031140585697e-06,
+      "loss": 2.213,
+      "step": 5207
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.637290387578647e-06,
+      "loss": 2.2256,
+      "step": 5208
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.635549678766661e-06,
+      "loss": 2.1771,
+      "step": 5209
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6338090143618435e-06,
+      "loss": 2.2331,
+      "step": 5210
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6320683945762835e-06,
+      "loss": 2.2108,
+      "step": 5211
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.630327819622076e-06,
+      "loss": 2.1983,
+      "step": 5212
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.628587289711303e-06,
+      "loss": 2.2605,
+      "step": 5213
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6268468050560394e-06,
+      "loss": 2.183,
+      "step": 5214
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.625106365868363e-06,
+      "loss": 2.277,
+      "step": 5215
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6233659723603374e-06,
+      "loss": 2.2338,
+      "step": 5216
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.621625624744026e-06,
+      "loss": 2.2941,
+      "step": 5217
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.619885323231484e-06,
+      "loss": 2.2284,
+      "step": 5218
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.618145068034764e-06,
+      "loss": 2.1724,
+      "step": 5219
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6164048593659076e-06,
+      "loss": 2.2256,
+      "step": 5220
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.614664697436956e-06,
+      "loss": 2.1933,
+      "step": 5221
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.612924582459943e-06,
+      "loss": 2.185,
+      "step": 5222
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.611184514646894e-06,
+      "loss": 2.194,
+      "step": 5223
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.609444494209834e-06,
+      "loss": 2.1959,
+      "step": 5224
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6077045213607765e-06,
+      "loss": 2.1952,
+      "step": 5225
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.605964596311733e-06,
+      "loss": 2.2362,
+      "step": 5226
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.604224719274708e-06,
+      "loss": 2.2923,
+      "step": 5227
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.602484890461702e-06,
+      "loss": 2.2569,
+      "step": 5228
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.600745110084704e-06,
+      "loss": 2.179,
+      "step": 5229
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5990053783557066e-06,
+      "loss": 2.1954,
+      "step": 5230
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.597265695486685e-06,
+      "loss": 2.2423,
+      "step": 5231
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.595526061689617e-06,
+      "loss": 2.2005,
+      "step": 5232
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.593786477176473e-06,
+      "loss": 2.1584,
+      "step": 5233
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.592046942159213e-06,
+      "loss": 2.2796,
+      "step": 5234
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.590307456849797e-06,
+      "loss": 2.208,
+      "step": 5235
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.588568021460172e-06,
+      "loss": 2.2168,
+      "step": 5236
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.586828636202288e-06,
+      "loss": 2.1934,
+      "step": 5237
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5850893012880806e-06,
+      "loss": 2.2517,
+      "step": 5238
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.583350016929484e-06,
+      "loss": 2.2206,
+      "step": 5239
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.581610783338424e-06,
+      "loss": 2.2544,
+      "step": 5240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.579871600726819e-06,
+      "loss": 2.2784,
+      "step": 5241
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.578132469306588e-06,
+      "loss": 2.2082,
+      "step": 5242
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.576393389289633e-06,
+      "loss": 2.1918,
+      "step": 5243
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.57465436088786e-06,
+      "loss": 2.1172,
+      "step": 5244
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.572915384313163e-06,
+      "loss": 2.2265,
+      "step": 5245
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.571176459777431e-06,
+      "loss": 2.2357,
+      "step": 5246
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.569437587492545e-06,
+      "loss": 2.1611,
+      "step": 5247
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5676987676703865e-06,
+      "loss": 2.2405,
+      "step": 5248
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.56596000052282e-06,
+      "loss": 2.1761,
+      "step": 5249
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.564221286261709e-06,
+      "loss": 2.1847,
+      "step": 5250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5624826250989156e-06,
+      "loss": 2.1965,
+      "step": 5251
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.560744017246284e-06,
+      "loss": 2.2794,
+      "step": 5252
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.559005462915665e-06,
+      "loss": 2.2294,
+      "step": 5253
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.557266962318889e-06,
+      "loss": 2.1879,
+      "step": 5254
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.555528515667793e-06,
+      "loss": 2.2363,
+      "step": 5255
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.553790123174198e-06,
+      "loss": 2.2114,
+      "step": 5256
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.552051785049925e-06,
+      "loss": 2.2084,
+      "step": 5257
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5503135015067815e-06,
+      "loss": 2.2081,
+      "step": 5258
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.548575272756573e-06,
+      "loss": 2.1928,
+      "step": 5259
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.546837099011101e-06,
+      "loss": 2.2143,
+      "step": 5260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.545098980482151e-06,
+      "loss": 2.2197,
+      "step": 5261
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.543360917381512e-06,
+      "loss": 2.2219,
+      "step": 5262
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.541622909920959e-06,
+      "loss": 2.1902,
+      "step": 5263
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.539884958312265e-06,
+      "loss": 2.2348,
+      "step": 5264
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.538147062767191e-06,
+      "loss": 2.2015,
+      "step": 5265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5364092234975e-06,
+      "loss": 2.229,
+      "step": 5266
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.534671440714939e-06,
+      "loss": 2.2544,
+      "step": 5267
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.532933714631248e-06,
+      "loss": 2.2372,
+      "step": 5268
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5311960454581685e-06,
+      "loss": 2.2055,
+      "step": 5269
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.529458433407429e-06,
+      "loss": 2.1912,
+      "step": 5270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.527720878690752e-06,
+      "loss": 2.1896,
+      "step": 5271
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.525983381519853e-06,
+      "loss": 2.2155,
+      "step": 5272
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.524245942106442e-06,
+      "loss": 2.2331,
+      "step": 5273
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.522508560662219e-06,
+      "loss": 2.2078,
+      "step": 5274
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.52077123739888e-06,
+      "loss": 2.2012,
+      "step": 5275
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.519033972528114e-06,
+      "loss": 2.2059,
+      "step": 5276
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.517296766261596e-06,
+      "loss": 2.1545,
+      "step": 5277
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5155596188110055e-06,
+      "loss": 2.2129,
+      "step": 5278
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.513822530388004e-06,
+      "loss": 2.1455,
+      "step": 5279
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.512085501204254e-06,
+      "loss": 2.169,
+      "step": 5280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.510348531471403e-06,
+      "loss": 2.2275,
+      "step": 5281
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.508611621401102e-06,
+      "loss": 2.2438,
+      "step": 5282
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.506874771204981e-06,
+      "loss": 2.216,
+      "step": 5283
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.505137981094675e-06,
+      "loss": 2.2483,
+      "step": 5284
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5034012512818065e-06,
+      "loss": 2.1895,
+      "step": 5285
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5016645819779865e-06,
+      "loss": 2.2568,
+      "step": 5286
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.499927973394826e-06,
+      "loss": 2.21,
+      "step": 5287
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4981914257439254e-06,
+      "loss": 2.1966,
+      "step": 5288
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.496454939236879e-06,
+      "loss": 2.2167,
+      "step": 5289
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.494718514085269e-06,
+      "loss": 2.1927,
+      "step": 5290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.492982150500677e-06,
+      "loss": 2.2922,
+      "step": 5291
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.491245848694669e-06,
+      "loss": 2.2184,
+      "step": 5292
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.489509608878813e-06,
+      "loss": 2.2426,
+      "step": 5293
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.487773431264664e-06,
+      "loss": 2.2498,
+      "step": 5294
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4860373160637665e-06,
+      "loss": 2.2526,
+      "step": 5295
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.484301263487664e-06,
+      "loss": 2.229,
+      "step": 5296
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.482565273747888e-06,
+      "loss": 2.2275,
+      "step": 5297
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4808293470559645e-06,
+      "loss": 2.2066,
+      "step": 5298
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.479093483623409e-06,
+      "loss": 2.1649,
+      "step": 5299
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.477357683661734e-06,
+      "loss": 2.2689,
+      "step": 5300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.475621947382438e-06,
+      "loss": 2.1843,
+      "step": 5301
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.473886274997018e-06,
+      "loss": 2.2051,
+      "step": 5302
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.472150666716961e-06,
+      "loss": 2.1924,
+      "step": 5303
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.470415122753742e-06,
+      "loss": 2.2393,
+      "step": 5304
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.468679643318836e-06,
+      "loss": 2.2515,
+      "step": 5305
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.466944228623701e-06,
+      "loss": 2.1606,
+      "step": 5306
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4652088788797965e-06,
+      "loss": 2.2145,
+      "step": 5307
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.463473594298567e-06,
+      "loss": 2.2176,
+      "step": 5308
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.461738375091454e-06,
+      "loss": 2.1584,
+      "step": 5309
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.460003221469886e-06,
+      "loss": 2.2757,
+      "step": 5310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.458268133645289e-06,
+      "loss": 2.264,
+      "step": 5311
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.456533111829076e-06,
+      "loss": 2.1922,
+      "step": 5312
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4547981562326535e-06,
+      "loss": 2.2519,
+      "step": 5313
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.453063267067424e-06,
+      "loss": 2.2024,
+      "step": 5314
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.451328444544774e-06,
+      "loss": 2.1679,
+      "step": 5315
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.44959368887609e-06,
+      "loss": 2.2331,
+      "step": 5316
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.447859000272744e-06,
+      "loss": 2.2106,
+      "step": 5317
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.446124378946108e-06,
+      "loss": 2.2952,
+      "step": 5318
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.444389825107534e-06,
+      "loss": 2.2178,
+      "step": 5319
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.442655338968373e-06,
+      "loss": 2.1786,
+      "step": 5320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.44092092073997e-06,
+      "loss": 2.2373,
+      "step": 5321
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.439186570633656e-06,
+      "loss": 2.2084,
+      "step": 5322
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.437452288860759e-06,
+      "loss": 2.2273,
+      "step": 5323
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4357180756325915e-06,
+      "loss": 2.194,
+      "step": 5324
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4339839311604675e-06,
+      "loss": 2.2291,
+      "step": 5325
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.432249855655681e-06,
+      "loss": 2.2184,
+      "step": 5326
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.430515849329532e-06,
+      "loss": 2.2023,
+      "step": 5327
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.428781912393299e-06,
+      "loss": 2.2814,
+      "step": 5328
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.427048045058254e-06,
+      "loss": 2.2267,
+      "step": 5329
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.425314247535668e-06,
+      "loss": 2.233,
+      "step": 5330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.423580520036797e-06,
+      "loss": 2.2989,
+      "step": 5331
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4218468627728935e-06,
+      "loss": 2.2123,
+      "step": 5332
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.420113275955193e-06,
+      "loss": 2.2239,
+      "step": 5333
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.418379759794934e-06,
+      "loss": 2.1823,
+      "step": 5334
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.416646314503334e-06,
+      "loss": 2.1915,
+      "step": 5335
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.414912940291614e-06,
+      "loss": 2.1553,
+      "step": 5336
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.413179637370977e-06,
+      "loss": 2.2083,
+      "step": 5337
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4114464059526185e-06,
+      "loss": 2.201,
+      "step": 5338
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.409713246247732e-06,
+      "loss": 2.1988,
+      "step": 5339
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4079801584674955e-06,
+      "loss": 2.1805,
+      "step": 5340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.406247142823082e-06,
+      "loss": 2.2438,
+      "step": 5341
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.404514199525651e-06,
+      "loss": 2.2543,
+      "step": 5342
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.402781328786361e-06,
+      "loss": 2.2706,
+      "step": 5343
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.401048530816353e-06,
+      "loss": 2.2354,
+      "step": 5344
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.399315805826765e-06,
+      "loss": 2.2206,
+      "step": 5345
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.397583154028725e-06,
+      "loss": 2.1804,
+      "step": 5346
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.395850575633348e-06,
+      "loss": 2.1755,
+      "step": 5347
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.394118070851749e-06,
+      "loss": 2.2626,
+      "step": 5348
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.392385639895022e-06,
+      "loss": 2.2166,
+      "step": 5349
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.390653282974264e-06,
+      "loss": 2.1156,
+      "step": 5350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.388921000300553e-06,
+      "loss": 2.1604,
+      "step": 5351
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.387188792084967e-06,
+      "loss": 2.2931,
+      "step": 5352
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.385456658538565e-06,
+      "loss": 2.2504,
+      "step": 5353
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.383724599872407e-06,
+      "loss": 2.1905,
+      "step": 5354
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.381992616297538e-06,
+      "loss": 2.1714,
+      "step": 5355
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.380260708024991e-06,
+      "loss": 2.2314,
+      "step": 5356
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.378528875265801e-06,
+      "loss": 2.2504,
+      "step": 5357
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.376797118230978e-06,
+      "loss": 2.2261,
+      "step": 5358
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.375065437131539e-06,
+      "loss": 2.285,
+      "step": 5359
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.373333832178478e-06,
+      "loss": 2.0859,
+      "step": 5360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.371602303582792e-06,
+      "loss": 2.1239,
+      "step": 5361
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.369870851555457e-06,
+      "loss": 2.2238,
+      "step": 5362
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3681394763074495e-06,
+      "loss": 2.2457,
+      "step": 5363
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.366408178049728e-06,
+      "loss": 2.2007,
+      "step": 5364
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3646769569932475e-06,
+      "loss": 2.2224,
+      "step": 5365
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.362945813348956e-06,
+      "loss": 2.1949,
+      "step": 5366
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.361214747327781e-06,
+      "loss": 2.2161,
+      "step": 5367
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.359483759140654e-06,
+      "loss": 2.157,
+      "step": 5368
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.357752848998486e-06,
+      "loss": 2.2664,
+      "step": 5369
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.356022017112187e-06,
+      "loss": 2.1952,
+      "step": 5370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.35429126369265e-06,
+      "loss": 2.2874,
+      "step": 5371
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.352560588950766e-06,
+      "loss": 2.1835,
+      "step": 5372
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.350829993097409e-06,
+      "loss": 2.1432,
+      "step": 5373
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.349099476343448e-06,
+      "loss": 2.2468,
+      "step": 5374
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.347369038899744e-06,
+      "loss": 2.1488,
+      "step": 5375
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.34563868097714e-06,
+      "loss": 2.2601,
+      "step": 5376
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.343908402786478e-06,
+      "loss": 2.2183,
+      "step": 5377
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.342178204538588e-06,
+      "loss": 2.1676,
+      "step": 5378
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.340448086444288e-06,
+      "loss": 2.2066,
+      "step": 5379
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3387180487143875e-06,
+      "loss": 2.0948,
+      "step": 5380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.336988091559688e-06,
+      "loss": 2.2448,
+      "step": 5381
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.335258215190979e-06,
+      "loss": 2.2008,
+      "step": 5382
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3335284198190385e-06,
+      "loss": 2.229,
+      "step": 5383
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.331798705654639e-06,
+      "loss": 2.2286,
+      "step": 5384
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.33006907290854e-06,
+      "loss": 2.1949,
+      "step": 5385
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.328339521791493e-06,
+      "loss": 2.2169,
+      "step": 5386
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.326610052514238e-06,
+      "loss": 2.147,
+      "step": 5387
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3248806652875045e-06,
+      "loss": 2.2611,
+      "step": 5388
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.323151360322014e-06,
+      "loss": 2.1345,
+      "step": 5389
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.321422137828479e-06,
+      "loss": 2.2447,
+      "step": 5390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.319692998017597e-06,
+      "loss": 2.2193,
+      "step": 5391
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.317963941100059e-06,
+      "loss": 2.2399,
+      "step": 5392
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.316234967286548e-06,
+      "loss": 2.262,
+      "step": 5393
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.314506076787729e-06,
+      "loss": 2.1856,
+      "step": 5394
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.312777269814268e-06,
+      "loss": 2.1418,
+      "step": 5395
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.31104854657681e-06,
+      "loss": 2.3005,
+      "step": 5396
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.309319907285998e-06,
+      "loss": 2.2036,
+      "step": 5397
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.307591352152459e-06,
+      "loss": 2.2757,
+      "step": 5398
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3058628813868154e-06,
+      "loss": 2.262,
+      "step": 5399
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.304134495199675e-06,
+      "loss": 2.2154,
+      "step": 5400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.302406193801632e-06,
+      "loss": 2.1531,
+      "step": 5401
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.300677977403281e-06,
+      "loss": 2.222,
+      "step": 5402
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.298949846215195e-06,
+      "loss": 2.2932,
+      "step": 5403
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.297221800447946e-06,
+      "loss": 2.168,
+      "step": 5404
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.295493840312087e-06,
+      "loss": 2.202,
+      "step": 5405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.293765966018167e-06,
+      "loss": 2.2246,
+      "step": 5406
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.292038177776722e-06,
+      "loss": 2.2918,
+      "step": 5407
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.290310475798278e-06,
+      "loss": 2.1068,
+      "step": 5408
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.288582860293351e-06,
+      "loss": 2.2586,
+      "step": 5409
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.286855331472442e-06,
+      "loss": 2.2076,
+      "step": 5410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.285127889546049e-06,
+      "loss": 2.2231,
+      "step": 5411
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.283400534724654e-06,
+      "loss": 2.2861,
+      "step": 5412
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.281673267218731e-06,
+      "loss": 2.2301,
+      "step": 5413
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.279946087238739e-06,
+      "loss": 2.2209,
+      "step": 5414
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.278218994995135e-06,
+      "loss": 2.2553,
+      "step": 5415
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2764919906983545e-06,
+      "loss": 2.2651,
+      "step": 5416
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.274765074558832e-06,
+      "loss": 2.3319,
+      "step": 5417
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.273038246786986e-06,
+      "loss": 2.2381,
+      "step": 5418
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2713115075932225e-06,
+      "loss": 2.2485,
+      "step": 5419
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.269584857187942e-06,
+      "loss": 2.2713,
+      "step": 5420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.267858295781531e-06,
+      "loss": 2.2389,
+      "step": 5421
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.266131823584368e-06,
+      "loss": 2.2297,
+      "step": 5422
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.264405440806813e-06,
+      "loss": 2.2497,
+      "step": 5423
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.262679147659227e-06,
+      "loss": 2.2772,
+      "step": 5424
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.260952944351947e-06,
+      "loss": 2.1689,
+      "step": 5425
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.259226831095311e-06,
+      "loss": 2.2325,
+      "step": 5426
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.25750080809964e-06,
+      "loss": 2.3088,
+      "step": 5427
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.255774875575239e-06,
+      "loss": 2.1936,
+      "step": 5428
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2540490337324156e-06,
+      "loss": 2.2576,
+      "step": 5429
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2523232827814534e-06,
+      "loss": 2.2828,
+      "step": 5430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.250597622932631e-06,
+      "loss": 2.1867,
+      "step": 5431
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.248872054396215e-06,
+      "loss": 2.1886,
+      "step": 5432
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.247146577382462e-06,
+      "loss": 2.3121,
+      "step": 5433
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.245421192101613e-06,
+      "loss": 2.3182,
+      "step": 5434
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.243695898763904e-06,
+      "loss": 2.1987,
+      "step": 5435
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.241970697579557e-06,
+      "loss": 2.1882,
+      "step": 5436
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.240245588758778e-06,
+      "loss": 2.2651,
+      "step": 5437
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.238520572511773e-06,
+      "loss": 2.2495,
+      "step": 5438
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2367956490487235e-06,
+      "loss": 2.3066,
+      "step": 5439
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.23507081857981e-06,
+      "loss": 2.2572,
+      "step": 5440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.233346081315197e-06,
+      "loss": 2.2554,
+      "step": 5441
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.23162143746504e-06,
+      "loss": 2.2222,
+      "step": 5442
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2298968872394784e-06,
+      "loss": 2.2384,
+      "step": 5443
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.228172430848645e-06,
+      "loss": 2.2189,
+      "step": 5444
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.226448068502661e-06,
+      "loss": 2.2252,
+      "step": 5445
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.224723800411631e-06,
+      "loss": 2.2088,
+      "step": 5446
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.222999626785658e-06,
+      "loss": 2.2657,
+      "step": 5447
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.22127554783482e-06,
+      "loss": 2.2577,
+      "step": 5448
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.219551563769196e-06,
+      "loss": 2.3225,
+      "step": 5449
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.217827674798845e-06,
+      "loss": 2.2188,
+      "step": 5450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.216103881133822e-06,
+      "loss": 2.2395,
+      "step": 5451
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2143801829841635e-06,
+      "loss": 2.194,
+      "step": 5452
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.212656580559894e-06,
+      "loss": 2.2096,
+      "step": 5453
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.210933074071033e-06,
+      "loss": 2.2015,
+      "step": 5454
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.209209663727583e-06,
+      "loss": 2.2746,
+      "step": 5455
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.207486349739538e-06,
+      "loss": 2.1953,
+      "step": 5456
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.205763132316875e-06,
+      "loss": 2.2763,
+      "step": 5457
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.204040011669567e-06,
+      "loss": 2.1677,
+      "step": 5458
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.202316988007568e-06,
+      "loss": 2.228,
+      "step": 5459
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.200594061540827e-06,
+      "loss": 2.242,
+      "step": 5460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.198871232479274e-06,
+      "loss": 2.2004,
+      "step": 5461
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.197148501032829e-06,
+      "loss": 2.2131,
+      "step": 5462
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.195425867411404e-06,
+      "loss": 2.2015,
+      "step": 5463
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.193703331824898e-06,
+      "loss": 2.2031,
+      "step": 5464
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.191980894483195e-06,
+      "loss": 2.2776,
+      "step": 5465
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.190258555596168e-06,
+      "loss": 2.1845,
+      "step": 5466
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1885363153736825e-06,
+      "loss": 2.2298,
+      "step": 5467
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.186814174025582e-06,
+      "loss": 2.2744,
+      "step": 5468
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.18509213176171e-06,
+      "loss": 2.2803,
+      "step": 5469
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.183370188791891e-06,
+      "loss": 2.2794,
+      "step": 5470
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.181648345325934e-06,
+      "loss": 2.1559,
+      "step": 5471
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.179926601573645e-06,
+      "loss": 2.2965,
+      "step": 5472
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.178204957744812e-06,
+      "loss": 2.2317,
+      "step": 5473
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.176483414049214e-06,
+      "loss": 2.2295,
+      "step": 5474
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.174761970696612e-06,
+      "loss": 2.2513,
+      "step": 5475
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.173040627896762e-06,
+      "loss": 2.275,
+      "step": 5476
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.171319385859402e-06,
+      "loss": 2.2103,
+      "step": 5477
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.169598244794261e-06,
+      "loss": 2.2505,
+      "step": 5478
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.167877204911057e-06,
+      "loss": 2.1922,
+      "step": 5479
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.166156266419489e-06,
+      "loss": 2.298,
+      "step": 5480
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.164435429529253e-06,
+      "loss": 2.2114,
+      "step": 5481
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.162714694450023e-06,
+      "loss": 2.275,
+      "step": 5482
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.160994061391469e-06,
+      "loss": 2.1963,
+      "step": 5483
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.159273530563243e-06,
+      "loss": 2.2101,
+      "step": 5484
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.157553102174988e-06,
+      "loss": 2.233,
+      "step": 5485
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.155832776436331e-06,
+      "loss": 2.26,
+      "step": 5486
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.15411255355689e-06,
+      "loss": 2.2103,
+      "step": 5487
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.15239243374627e-06,
+      "loss": 2.1982,
+      "step": 5488
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.150672417214058e-06,
+      "loss": 2.2106,
+      "step": 5489
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.148952504169839e-06,
+      "loss": 2.2947,
+      "step": 5490
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.147232694823173e-06,
+      "loss": 2.2806,
+      "step": 5491
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.145512989383618e-06,
+      "loss": 2.25,
+      "step": 5492
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.143793388060712e-06,
+      "loss": 2.2632,
+      "step": 5493
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.142073891063986e-06,
+      "loss": 2.3015,
+      "step": 5494
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.140354498602952e-06,
+      "loss": 2.2507,
+      "step": 5495
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.138635210887117e-06,
+      "loss": 2.2903,
+      "step": 5496
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1369160281259685e-06,
+      "loss": 2.2204,
+      "step": 5497
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.135196950528982e-06,
+      "loss": 2.2477,
+      "step": 5498
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.133477978305626e-06,
+      "loss": 2.2365,
+      "step": 5499
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.131759111665349e-06,
+      "loss": 2.3031,
+      "step": 5500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.13004035081759e-06,
+      "loss": 2.1711,
+      "step": 5501
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.128321695971775e-06,
+      "loss": 2.1077,
+      "step": 5502
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.12660314733732e-06,
+      "loss": 2.2849,
+      "step": 5503
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.124884705123619e-06,
+      "loss": 2.2293,
+      "step": 5504
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.123166369540066e-06,
+      "loss": 2.2979,
+      "step": 5505
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.121448140796029e-06,
+      "loss": 2.2759,
+      "step": 5506
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.119730019100869e-06,
+      "loss": 2.275,
+      "step": 5507
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.118012004663939e-06,
+      "loss": 2.1387,
+      "step": 5508
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1162940976945695e-06,
+      "loss": 2.3389,
+      "step": 5509
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.114576298402085e-06,
+      "loss": 2.3273,
+      "step": 5510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.11285860699579e-06,
+      "loss": 2.2976,
+      "step": 5511
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.111141023684986e-06,
+      "loss": 2.2867,
+      "step": 5512
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.109423548678949e-06,
+      "loss": 2.286,
+      "step": 5513
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.107706182186954e-06,
+      "loss": 2.2502,
+      "step": 5514
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.105988924418252e-06,
+      "loss": 2.2929,
+      "step": 5515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.104271775582089e-06,
+      "loss": 2.3225,
+      "step": 5516
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.102554735887694e-06,
+      "loss": 2.1848,
+      "step": 5517
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.100837805544279e-06,
+      "loss": 2.2469,
+      "step": 5518
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.099120984761053e-06,
+      "loss": 2.3279,
+      "step": 5519
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0974042737472005e-06,
+      "loss": 2.2602,
+      "step": 5520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0956876727119e-06,
+      "loss": 2.3575,
+      "step": 5521
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.093971181864313e-06,
+      "loss": 2.2452,
+      "step": 5522
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.092254801413591e-06,
+      "loss": 2.2163,
+      "step": 5523
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.090538531568867e-06,
+      "loss": 2.2733,
+      "step": 5524
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.088822372539263e-06,
+      "loss": 2.2063,
+      "step": 5525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.087106324533891e-06,
+      "loss": 2.2402,
+      "step": 5526
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0853903877618425e-06,
+      "loss": 2.2029,
+      "step": 5527
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.083674562432203e-06,
+      "loss": 2.2826,
+      "step": 5528
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.081958848754035e-06,
+      "loss": 2.2437,
+      "step": 5529
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0802432469364e-06,
+      "loss": 2.2816,
+      "step": 5530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.078527757188333e-06,
+      "loss": 2.2453,
+      "step": 5531
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0768123797188665e-06,
+      "loss": 2.2704,
+      "step": 5532
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.075097114737011e-06,
+      "loss": 2.2196,
+      "step": 5533
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.073381962451764e-06,
+      "loss": 2.2256,
+      "step": 5534
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0716669230721154e-06,
+      "loss": 2.2437,
+      "step": 5535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.069951996807034e-06,
+      "loss": 2.2869,
+      "step": 5536
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.068237183865485e-06,
+      "loss": 2.1933,
+      "step": 5537
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.066522484456406e-06,
+      "loss": 2.2268,
+      "step": 5538
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.064807898788731e-06,
+      "loss": 2.253,
+      "step": 5539
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.063093427071376e-06,
+      "loss": 2.2962,
+      "step": 5540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.061379069513248e-06,
+      "loss": 2.2347,
+      "step": 5541
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0596648263232315e-06,
+      "loss": 2.1719,
+      "step": 5542
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.057950697710203e-06,
+      "loss": 2.1901,
+      "step": 5543
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0562366838830255e-06,
+      "loss": 2.1865,
+      "step": 5544
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.054522785050543e-06,
+      "loss": 2.1969,
+      "step": 5545
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.052809001421595e-06,
+      "loss": 2.2881,
+      "step": 5546
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.051095333204994e-06,
+      "loss": 2.1804,
+      "step": 5547
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0493817806095504e-06,
+      "loss": 2.2533,
+      "step": 5548
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.047668343844051e-06,
+      "loss": 2.2671,
+      "step": 5549
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.045955023117276e-06,
+      "loss": 2.255,
+      "step": 5550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0442418186379895e-06,
+      "loss": 2.2204,
+      "step": 5551
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.042528730614935e-06,
+      "loss": 2.2148,
+      "step": 5552
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.040815759256852e-06,
+      "loss": 2.2287,
+      "step": 5553
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.039102904772459e-06,
+      "loss": 2.2332,
+      "step": 5554
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.037390167370464e-06,
+      "loss": 2.1568,
+      "step": 5555
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.035677547259555e-06,
+      "loss": 2.2531,
+      "step": 5556
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0339650446484135e-06,
+      "loss": 2.2325,
+      "step": 5557
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.032252659745699e-06,
+      "loss": 2.1892,
+      "step": 5558
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.030540392760064e-06,
+      "loss": 2.2819,
+      "step": 5559
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.028828243900141e-06,
+      "loss": 2.2956,
+      "step": 5560
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.02711621337455e-06,
+      "loss": 2.2195,
+      "step": 5561
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.025404301391898e-06,
+      "loss": 2.2404,
+      "step": 5562
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.023692508160776e-06,
+      "loss": 2.2017,
+      "step": 5563
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.02198083388976e-06,
+      "loss": 2.2523,
+      "step": 5564
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.020269278787411e-06,
+      "loss": 2.232,
+      "step": 5565
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.018557843062282e-06,
+      "loss": 2.2167,
+      "step": 5566
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.016846526922901e-06,
+      "loss": 2.2368,
+      "step": 5567
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.015135330577787e-06,
+      "loss": 2.2404,
+      "step": 5568
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0134242542354486e-06,
+      "loss": 2.2372,
+      "step": 5569
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0117132981043695e-06,
+      "loss": 2.2111,
+      "step": 5570
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.01000246239303e-06,
+      "loss": 2.2459,
+      "step": 5571
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0082917473098845e-06,
+      "loss": 2.1445,
+      "step": 5572
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.006581153063383e-06,
+      "loss": 2.2416,
+      "step": 5573
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.004870679861953e-06,
+      "loss": 2.2314,
+      "step": 5574
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.003160327914015e-06,
+      "loss": 2.258,
+      "step": 5575
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.001450097427965e-06,
+      "loss": 2.1605,
+      "step": 5576
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.999739988612192e-06,
+      "loss": 2.2895,
+      "step": 5577
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9980300016750696e-06,
+      "loss": 2.198,
+      "step": 5578
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9963201368249495e-06,
+      "loss": 2.1925,
+      "step": 5579
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.994610394270178e-06,
+      "loss": 2.1319,
+      "step": 5580
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.992900774219078e-06,
+      "loss": 2.1931,
+      "step": 5581
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.991191276879966e-06,
+      "loss": 2.1857,
+      "step": 5582
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.989481902461135e-06,
+      "loss": 2.1882,
+      "step": 5583
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.987772651170871e-06,
+      "loss": 2.1769,
+      "step": 5584
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.986063523217439e-06,
+      "loss": 2.207,
+      "step": 5585
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.98435451880909e-06,
+      "loss": 2.1173,
+      "step": 5586
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.982645638154062e-06,
+      "loss": 2.2597,
+      "step": 5587
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.980936881460576e-06,
+      "loss": 2.2057,
+      "step": 5588
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.979228248936843e-06,
+      "loss": 2.2403,
+      "step": 5589
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.977519740791049e-06,
+      "loss": 2.2516,
+      "step": 5590
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9758113572313735e-06,
+      "loss": 2.1355,
+      "step": 5591
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.974103098465976e-06,
+      "loss": 2.2354,
+      "step": 5592
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.972394964703008e-06,
+      "loss": 2.2754,
+      "step": 5593
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.970686956150595e-06,
+      "loss": 2.2177,
+      "step": 5594
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.968979073016853e-06,
+      "loss": 2.1645,
+      "step": 5595
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.967271315509884e-06,
+      "loss": 2.2784,
+      "step": 5596
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.965563683837772e-06,
+      "loss": 2.245,
+      "step": 5597
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.963856178208588e-06,
+      "loss": 2.2148,
+      "step": 5598
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.962148798830385e-06,
+      "loss": 2.2351,
+      "step": 5599
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.960441545911205e-06,
+      "loss": 2.2364,
+      "step": 5600
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9587344196590665e-06,
+      "loss": 2.2013,
+      "step": 5601
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.957027420281981e-06,
+      "loss": 2.1999,
+      "step": 5602
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.955320547987943e-06,
+      "loss": 2.2573,
+      "step": 5603
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9536138029849244e-06,
+      "loss": 2.1795,
+      "step": 5604
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.951907185480892e-06,
+      "loss": 2.1822,
+      "step": 5605
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.950200695683788e-06,
+      "loss": 2.2481,
+      "step": 5606
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9484943338015465e-06,
+      "loss": 2.1415,
+      "step": 5607
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.94678810004208e-06,
+      "loss": 2.2204,
+      "step": 5608
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.94508199461329e-06,
+      "loss": 2.244,
+      "step": 5609
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.943376017723058e-06,
+      "loss": 2.2444,
+      "step": 5610
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.941670169579252e-06,
+      "loss": 2.2466,
+      "step": 5611
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.939964450389728e-06,
+      "loss": 2.2362,
+      "step": 5612
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.938258860362319e-06,
+      "loss": 2.2479,
+      "step": 5613
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.936553399704848e-06,
+      "loss": 2.2522,
+      "step": 5614
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.934848068625117e-06,
+      "loss": 2.2511,
+      "step": 5615
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.933142867330921e-06,
+      "loss": 2.1976,
+      "step": 5616
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.931437796030028e-06,
+      "loss": 2.2654,
+      "step": 5617
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9297328549302e-06,
+      "loss": 2.2842,
+      "step": 5618
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.928028044239176e-06,
+      "loss": 2.2339,
+      "step": 5619
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.926323364164684e-06,
+      "loss": 2.2274,
+      "step": 5620
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.924618814914435e-06,
+      "loss": 2.2334,
+      "step": 5621
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.922914396696118e-06,
+      "loss": 2.1987,
+      "step": 5622
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.921210109717417e-06,
+      "loss": 2.2406,
+      "step": 5623
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.91950595418599e-06,
+      "loss": 2.2132,
+      "step": 5624
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.917801930309486e-06,
+      "loss": 2.2148,
+      "step": 5625
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9160980382955336e-06,
+      "loss": 2.1797,
+      "step": 5626
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.914394278351749e-06,
+      "loss": 2.1957,
+      "step": 5627
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.912690650685726e-06,
+      "loss": 2.1505,
+      "step": 5628
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9109871555050514e-06,
+      "loss": 2.1461,
+      "step": 5629
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.909283793017289e-06,
+      "loss": 2.2393,
+      "step": 5630
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.907580563429985e-06,
+      "loss": 2.2557,
+      "step": 5631
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.905877466950679e-06,
+      "loss": 2.1643,
+      "step": 5632
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.904174503786882e-06,
+      "loss": 2.1621,
+      "step": 5633
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.902471674146099e-06,
+      "loss": 2.212,
+      "step": 5634
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.900768978235812e-06,
+      "loss": 2.2143,
+      "step": 5635
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.899066416263493e-06,
+      "loss": 2.2403,
+      "step": 5636
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.89736398843659e-06,
+      "loss": 2.1385,
+      "step": 5637
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.895661694962542e-06,
+      "loss": 2.1662,
+      "step": 5638
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8939595360487655e-06,
+      "loss": 2.1667,
+      "step": 5639
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.892257511902664e-06,
+      "loss": 2.2365,
+      "step": 5640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.890555622731626e-06,
+      "loss": 2.2561,
+      "step": 5641
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.888853868743018e-06,
+      "loss": 2.2083,
+      "step": 5642
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.887152250144197e-06,
+      "loss": 2.1852,
+      "step": 5643
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.885450767142498e-06,
+      "loss": 2.2556,
+      "step": 5644
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.883749419945244e-06,
+      "loss": 2.2444,
+      "step": 5645
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.882048208759735e-06,
+      "loss": 2.2103,
+      "step": 5646
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.880347133793263e-06,
+      "loss": 2.2468,
+      "step": 5647
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8786461952530955e-06,
+      "loss": 2.2282,
+      "step": 5648
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.876945393346486e-06,
+      "loss": 2.2714,
+      "step": 5649
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.875244728280676e-06,
+      "loss": 2.1602,
+      "step": 5650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.873544200262882e-06,
+      "loss": 2.1904,
+      "step": 5651
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.871843809500313e-06,
+      "loss": 2.2331,
+      "step": 5652
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.870143556200152e-06,
+      "loss": 2.2191,
+      "step": 5653
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.868443440569571e-06,
+      "loss": 2.1895,
+      "step": 5654
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.866743462815724e-06,
+      "loss": 2.1656,
+      "step": 5655
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.865043623145751e-06,
+      "loss": 2.2226,
+      "step": 5656
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.863343921766769e-06,
+      "loss": 2.1885,
+      "step": 5657
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.86164435888588e-06,
+      "loss": 2.1113,
+      "step": 5658
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.859944934710177e-06,
+      "loss": 2.2251,
+      "step": 5659
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8582456494467214e-06,
+      "loss": 2.1894,
+      "step": 5660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.856546503302573e-06,
+      "loss": 2.269,
+      "step": 5661
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.854847496484762e-06,
+      "loss": 2.1735,
+      "step": 5662
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.853148629200312e-06,
+      "loss": 2.1899,
+      "step": 5663
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8514499016562216e-06,
+      "loss": 2.2075,
+      "step": 5664
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.849751314059479e-06,
+      "loss": 2.1727,
+      "step": 5665
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8480528666170495e-06,
+      "loss": 2.2538,
+      "step": 5666
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.846354559535881e-06,
+      "loss": 2.2876,
+      "step": 5667
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.844656393022912e-06,
+      "loss": 2.2505,
+      "step": 5668
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.842958367285056e-06,
+      "loss": 2.1837,
+      "step": 5669
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.841260482529215e-06,
+      "loss": 2.1884,
+      "step": 5670
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.839562738962267e-06,
+      "loss": 2.2334,
+      "step": 5671
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.83786513679108e-06,
+      "loss": 2.2523,
+      "step": 5672
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.836167676222501e-06,
+      "loss": 2.271,
+      "step": 5673
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.834470357463362e-06,
+      "loss": 2.1998,
+      "step": 5674
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.832773180720475e-06,
+      "loss": 2.2121,
+      "step": 5675
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.831076146200633e-06,
+      "loss": 2.1518,
+      "step": 5676
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.829379254110617e-06,
+      "loss": 2.225,
+      "step": 5677
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.827682504657187e-06,
+      "loss": 2.217,
+      "step": 5678
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.825985898047091e-06,
+      "loss": 2.2159,
+      "step": 5679
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.82428943448705e-06,
+      "loss": 2.143,
+      "step": 5680
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.822593114183777e-06,
+      "loss": 2.1854,
+      "step": 5681
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.820896937343959e-06,
+      "loss": 2.2073,
+      "step": 5682
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.819200904174274e-06,
+      "loss": 2.1846,
+      "step": 5683
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.817505014881378e-06,
+      "loss": 2.2553,
+      "step": 5684
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.815809269671908e-06,
+      "loss": 2.1956,
+      "step": 5685
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.814113668752486e-06,
+      "loss": 2.1528,
+      "step": 5686
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8124182123297153e-06,
+      "loss": 2.1646,
+      "step": 5687
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.810722900610186e-06,
+      "loss": 2.1984,
+      "step": 5688
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.809027733800461e-06,
+      "loss": 2.2543,
+      "step": 5689
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8073327121070968e-06,
+      "loss": 2.1576,
+      "step": 5690
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8056378357366224e-06,
+      "loss": 2.1465,
+      "step": 5691
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8039431048955537e-06,
+      "loss": 2.2725,
+      "step": 5692
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8022485197903924e-06,
+      "loss": 2.1687,
+      "step": 5693
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8005540806276132e-06,
+      "loss": 2.1987,
+      "step": 5694
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.798859787613682e-06,
+      "loss": 2.2135,
+      "step": 5695
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.797165640955041e-06,
+      "loss": 2.175,
+      "step": 5696
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7954716408581206e-06,
+      "loss": 2.154,
+      "step": 5697
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.793777787529325e-06,
+      "loss": 2.2523,
+      "step": 5698
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7920840811750485e-06,
+      "loss": 2.1717,
+      "step": 5699
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.790390522001662e-06,
+      "loss": 2.2,
+      "step": 5700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7886971102155205e-06,
+      "loss": 2.1712,
+      "step": 5701
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.787003846022964e-06,
+      "loss": 2.1783,
+      "step": 5702
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.785310729630307e-06,
+      "loss": 2.2471,
+      "step": 5703
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7836177612438557e-06,
+      "loss": 2.1806,
+      "step": 5704
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7819249410698877e-06,
+      "loss": 2.1513,
+      "step": 5705
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7802322693146726e-06,
+      "loss": 2.1228,
+      "step": 5706
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.778539746184454e-06,
+      "loss": 2.2165,
+      "step": 5707
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.776847371885464e-06,
+      "loss": 2.1837,
+      "step": 5708
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7751551466239113e-06,
+      "loss": 2.2816,
+      "step": 5709
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7734630706059873e-06,
+      "loss": 2.1699,
+      "step": 5710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7717711440378695e-06,
+      "loss": 2.2087,
+      "step": 5711
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.77007936712571e-06,
+      "loss": 2.2204,
+      "step": 5712
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7683877400756513e-06,
+      "loss": 2.2577,
+      "step": 5713
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7666962630938084e-06,
+      "loss": 2.1393,
+      "step": 5714
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.765004936386286e-06,
+      "loss": 2.2234,
+      "step": 5715
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7633137601591647e-06,
+      "loss": 2.2218,
+      "step": 5716
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.761622734618513e-06,
+      "loss": 2.1925,
+      "step": 5717
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.759931859970374e-06,
+      "loss": 2.2784,
+      "step": 5718
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.758241136420775e-06,
+      "loss": 2.1497,
+      "step": 5719
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.756550564175727e-06,
+      "loss": 2.1228,
+      "step": 5720
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.75486014344122e-06,
+      "loss": 2.179,
+      "step": 5721
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7531698744232307e-06,
+      "loss": 2.2193,
+      "step": 5722
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7514797573277075e-06,
+      "loss": 2.141,
+      "step": 5723
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.74978979236059e-06,
+      "loss": 2.1908,
+      "step": 5724
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.748099979727792e-06,
+      "loss": 2.1704,
+      "step": 5725
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7464103196352176e-06,
+      "loss": 2.2462,
+      "step": 5726
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7447208122887425e-06,
+      "loss": 2.1369,
+      "step": 5727
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7430314578942263e-06,
+      "loss": 2.1537,
+      "step": 5728
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7413422566575153e-06,
+      "loss": 2.1774,
+      "step": 5729
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7396532087844318e-06,
+      "loss": 2.1233,
+      "step": 5730
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7379643144807835e-06,
+      "loss": 2.2372,
+      "step": 5731
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.736275573952354e-06,
+      "loss": 2.2217,
+      "step": 5732
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7345869874049136e-06,
+      "loss": 2.1287,
+      "step": 5733
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7328985550442086e-06,
+      "loss": 2.2337,
+      "step": 5734
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7312102770759724e-06,
+      "loss": 2.2309,
+      "step": 5735
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7295221537059162e-06,
+      "loss": 2.1629,
+      "step": 5736
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.72783418513973e-06,
+      "loss": 2.268,
+      "step": 5737
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7261463715830902e-06,
+      "loss": 2.2642,
+      "step": 5738
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7244587132416497e-06,
+      "loss": 2.2322,
+      "step": 5739
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7227712103210485e-06,
+      "loss": 2.2065,
+      "step": 5740
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7210838630268986e-06,
+      "loss": 2.1643,
+      "step": 5741
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7193966715648026e-06,
+      "loss": 2.253,
+      "step": 5742
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7177096361403362e-06,
+      "loss": 2.2237,
+      "step": 5743
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.716022756959061e-06,
+      "loss": 2.1186,
+      "step": 5744
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7143360342265206e-06,
+      "loss": 2.2144,
+      "step": 5745
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7126494681482317e-06,
+      "loss": 2.2894,
+      "step": 5746
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7109630589297014e-06,
+      "loss": 2.1396,
+      "step": 5747
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.709276806776412e-06,
+      "loss": 2.271,
+      "step": 5748
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.707590711893829e-06,
+      "loss": 2.2772,
+      "step": 5749
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.705904774487396e-06,
+      "loss": 2.263,
+      "step": 5750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.704218994762543e-06,
+      "loss": 2.1918,
+      "step": 5751
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7025333729246733e-06,
+      "loss": 2.159,
+      "step": 5752
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.700847909179177e-06,
+      "loss": 2.2271,
+      "step": 5753
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.699162603731423e-06,
+      "loss": 2.2404,
+      "step": 5754
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6974774567867586e-06,
+      "loss": 2.2441,
+      "step": 5755
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.695792468550517e-06,
+      "loss": 2.2148,
+      "step": 5756
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.694107639228005e-06,
+      "loss": 2.191,
+      "step": 5757
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6924229690245163e-06,
+      "loss": 2.1902,
+      "step": 5758
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.690738458145322e-06,
+      "loss": 2.2081,
+      "step": 5759
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6890541067956775e-06,
+      "loss": 2.1999,
+      "step": 5760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6873699151808105e-06,
+      "loss": 2.2245,
+      "step": 5761
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.68568588350594e-06,
+      "loss": 2.2065,
+      "step": 5762
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.684002011976259e-06,
+      "loss": 2.1706,
+      "step": 5763
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6823183007969375e-06,
+      "loss": 2.2372,
+      "step": 5764
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.680634750173137e-06,
+      "loss": 2.2151,
+      "step": 5765
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.678951360309988e-06,
+      "loss": 2.155,
+      "step": 5766
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6772681314126097e-06,
+      "loss": 2.2116,
+      "step": 5767
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6755850636860956e-06,
+      "loss": 2.1964,
+      "step": 5768
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6739021573355273e-06,
+      "loss": 2.2074,
+      "step": 5769
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.672219412565956e-06,
+      "loss": 2.2064,
+      "step": 5770
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.670536829582424e-06,
+      "loss": 2.2131,
+      "step": 5771
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.668854408589945e-06,
+      "loss": 2.1552,
+      "step": 5772
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6671721497935177e-06,
+      "loss": 2.1512,
+      "step": 5773
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6654900533981234e-06,
+      "loss": 2.1983,
+      "step": 5774
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.663808119608716e-06,
+      "loss": 2.167,
+      "step": 5775
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6621263486302373e-06,
+      "loss": 2.1744,
+      "step": 5776
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6604447406676036e-06,
+      "loss": 2.2078,
+      "step": 5777
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6587632959257168e-06,
+      "loss": 2.2391,
+      "step": 5778
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.657082014609452e-06,
+      "loss": 2.2217,
+      "step": 5779
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.655400896923672e-06,
+      "loss": 2.2516,
+      "step": 5780
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.653719943073214e-06,
+      "loss": 2.285,
+      "step": 5781
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6520391532628953e-06,
+      "loss": 2.2852,
+      "step": 5782
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6503585276975196e-06,
+      "loss": 2.1865,
+      "step": 5783
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.648678066581861e-06,
+      "loss": 2.1735,
+      "step": 5784
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6469977701206833e-06,
+      "loss": 2.2127,
+      "step": 5785
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.645317638518721e-06,
+      "loss": 2.153,
+      "step": 5786
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6436376719806965e-06,
+      "loss": 2.2174,
+      "step": 5787
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6419578707113055e-06,
+      "loss": 2.1887,
+      "step": 5788
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.640278234915232e-06,
+      "loss": 2.1171,
+      "step": 5789
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6385987647971287e-06,
+      "loss": 2.1808,
+      "step": 5790
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6369194605616364e-06,
+      "loss": 2.1533,
+      "step": 5791
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.635240322413375e-06,
+      "loss": 2.1132,
+      "step": 5792
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6335613505569386e-06,
+      "loss": 2.2383,
+      "step": 5793
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6318825451969085e-06,
+      "loss": 2.2034,
+      "step": 5794
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.630203906537838e-06,
+      "loss": 2.206,
+      "step": 5795
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.628525434784268e-06,
+      "loss": 2.2614,
+      "step": 5796
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6268471301407127e-06,
+      "loss": 2.2054,
+      "step": 5797
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.625168992811671e-06,
+      "loss": 2.1388,
+      "step": 5798
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6234910230016173e-06,
+      "loss": 2.2244,
+      "step": 5799
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6218132209150047e-06,
+      "loss": 2.2075,
+      "step": 5800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6201355867562725e-06,
+      "loss": 2.1916,
+      "step": 5801
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.618458120729832e-06,
+      "loss": 2.2368,
+      "step": 5802
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.616780823040081e-06,
+      "loss": 2.1697,
+      "step": 5803
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6151036938913887e-06,
+      "loss": 2.221,
+      "step": 5804
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.613426733488111e-06,
+      "loss": 2.2605,
+      "step": 5805
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.61174994203458e-06,
+      "loss": 2.2493,
+      "step": 5806
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.610073319735109e-06,
+      "loss": 2.2072,
+      "step": 5807
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.608396866793988e-06,
+      "loss": 2.1826,
+      "step": 5808
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.606720583415485e-06,
+      "loss": 2.221,
+      "step": 5809
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6050444698038547e-06,
+      "loss": 2.2108,
+      "step": 5810
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.603368526163323e-06,
+      "loss": 2.2097,
+      "step": 5811
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6016927526981014e-06,
+      "loss": 2.2325,
+      "step": 5812
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.600017149612375e-06,
+      "loss": 2.1795,
+      "step": 5813
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.598341717110313e-06,
+      "loss": 2.2402,
+      "step": 5814
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5966664553960622e-06,
+      "loss": 2.163,
+      "step": 5815
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5949913646737456e-06,
+      "loss": 2.2505,
+      "step": 5816
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5933164451474708e-06,
+      "loss": 2.2799,
+      "step": 5817
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5916416970213173e-06,
+      "loss": 2.2066,
+      "step": 5818
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5899671204993535e-06,
+      "loss": 2.1999,
+      "step": 5819
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5882927157856175e-06,
+      "loss": 2.1676,
+      "step": 5820
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.586618483084134e-06,
+      "loss": 2.1847,
+      "step": 5821
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.584944422598899e-06,
+      "loss": 2.1658,
+      "step": 5822
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.583270534533896e-06,
+      "loss": 2.1322,
+      "step": 5823
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5815968190930793e-06,
+      "loss": 2.1641,
+      "step": 5824
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.579923276480387e-06,
+      "loss": 2.1788,
+      "step": 5825
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5782499068997386e-06,
+      "loss": 2.2501,
+      "step": 5826
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5765767105550236e-06,
+      "loss": 2.1567,
+      "step": 5827
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5749036876501196e-06,
+      "loss": 2.1597,
+      "step": 5828
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.573230838388878e-06,
+      "loss": 2.2013,
+      "step": 5829
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.571558162975133e-06,
+      "loss": 2.1381,
+      "step": 5830
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.569885661612691e-06,
+      "loss": 2.1156,
+      "step": 5831
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.568213334505345e-06,
+      "loss": 2.2011,
+      "step": 5832
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5665411818568596e-06,
+      "loss": 2.1371,
+      "step": 5833
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.564869203870982e-06,
+      "loss": 2.1791,
+      "step": 5834
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5631974007514414e-06,
+      "loss": 2.1746,
+      "step": 5835
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.561525772701937e-06,
+      "loss": 2.2717,
+      "step": 5836
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.559854319926156e-06,
+      "loss": 2.2049,
+      "step": 5837
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5581830426277554e-06,
+      "loss": 2.154,
+      "step": 5838
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.556511941010378e-06,
+      "loss": 2.1787,
+      "step": 5839
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5548410152776414e-06,
+      "loss": 2.2125,
+      "step": 5840
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.553170265633146e-06,
+      "loss": 2.2867,
+      "step": 5841
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5514996922804636e-06,
+      "loss": 2.168,
+      "step": 5842
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5498292954231497e-06,
+      "loss": 2.1972,
+      "step": 5843
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.548159075264738e-06,
+      "loss": 2.1738,
+      "step": 5844
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5464890320087374e-06,
+      "loss": 2.2012,
+      "step": 5845
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5448191658586423e-06,
+      "loss": 2.2366,
+      "step": 5846
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5431494770179154e-06,
+      "loss": 2.2624,
+      "step": 5847
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5414799656900057e-06,
+      "loss": 2.185,
+      "step": 5848
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.539810632078338e-06,
+      "loss": 2.1956,
+      "step": 5849
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.538141476386317e-06,
+      "loss": 2.1866,
+      "step": 5850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.536472498817323e-06,
+      "loss": 2.1606,
+      "step": 5851
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5348036995747135e-06,
+      "loss": 2.2222,
+      "step": 5852
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5331350788618303e-06,
+      "loss": 2.2199,
+      "step": 5853
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.531466636881987e-06,
+      "loss": 2.1936,
+      "step": 5854
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5297983738384813e-06,
+      "loss": 2.1737,
+      "step": 5855
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5281302899345825e-06,
+      "loss": 2.1605,
+      "step": 5856
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5264623853735435e-06,
+      "loss": 2.2761,
+      "step": 5857
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.524794660358593e-06,
+      "loss": 2.2053,
+      "step": 5858
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5231271150929403e-06,
+      "loss": 2.1681,
+      "step": 5859
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.521459749779769e-06,
+      "loss": 2.2236,
+      "step": 5860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5197925646222387e-06,
+      "loss": 2.2035,
+      "step": 5861
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5181255598234963e-06,
+      "loss": 2.2081,
+      "step": 5862
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5164587355866593e-06,
+      "loss": 2.2307,
+      "step": 5863
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5147920921148267e-06,
+      "loss": 2.1547,
+      "step": 5864
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5131256296110703e-06,
+      "loss": 2.2229,
+      "step": 5865
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.511459348278448e-06,
+      "loss": 2.2272,
+      "step": 5866
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.509793248319987e-06,
+      "loss": 2.1789,
+      "step": 5867
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.508127329938699e-06,
+      "loss": 2.1838,
+      "step": 5868
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5064615933375724e-06,
+      "loss": 2.2305,
+      "step": 5869
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5047960387195673e-06,
+      "loss": 2.1468,
+      "step": 5870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.503130666287631e-06,
+      "loss": 2.1543,
+      "step": 5871
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.501465476244681e-06,
+      "loss": 2.1955,
+      "step": 5872
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.49980046879362e-06,
+      "loss": 2.1671,
+      "step": 5873
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.498135644137318e-06,
+      "loss": 2.2263,
+      "step": 5874
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4964710024786354e-06,
+      "loss": 2.2464,
+      "step": 5875
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4948065440203982e-06,
+      "loss": 2.2325,
+      "step": 5876
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4931422689654186e-06,
+      "loss": 2.1738,
+      "step": 5877
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.491478177516484e-06,
+      "loss": 2.2468,
+      "step": 5878
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4898142698763555e-06,
+      "loss": 2.1791,
+      "step": 5879
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.488150546247778e-06,
+      "loss": 2.2552,
+      "step": 5880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.486487006833471e-06,
+      "loss": 2.1338,
+      "step": 5881
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.484823651836131e-06,
+      "loss": 2.1797,
+      "step": 5882
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.483160481458432e-06,
+      "loss": 2.2511,
+      "step": 5883
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4814974959030294e-06,
+      "loss": 2.2094,
+      "step": 5884
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4798346953725487e-06,
+      "loss": 2.1153,
+      "step": 5885
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4781720800696006e-06,
+      "loss": 2.1788,
+      "step": 5886
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.476509650196769e-06,
+      "loss": 2.2349,
+      "step": 5887
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.474847405956613e-06,
+      "loss": 2.1998,
+      "step": 5888
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4731853475516763e-06,
+      "loss": 2.273,
+      "step": 5889
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.471523475184472e-06,
+      "loss": 2.2123,
+      "step": 5890
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4698617890574972e-06,
+      "loss": 2.1829,
+      "step": 5891
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4682002893732203e-06,
+      "loss": 2.184,
+      "step": 5892
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4665389763340945e-06,
+      "loss": 2.151,
+      "step": 5893
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.464877850142541e-06,
+      "loss": 2.2142,
+      "step": 5894
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.463216911000965e-06,
+      "loss": 2.1897,
+      "step": 5895
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4615561591117486e-06,
+      "loss": 2.1728,
+      "step": 5896
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.459895594677245e-06,
+      "loss": 2.1737,
+      "step": 5897
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4582352178997937e-06,
+      "loss": 2.2189,
+      "step": 5898
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4565750289817024e-06,
+      "loss": 2.2332,
+      "step": 5899
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4549150281252635e-06,
+      "loss": 2.1853,
+      "step": 5900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4532552155327404e-06,
+      "loss": 2.2017,
+      "step": 5901
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4515955914063796e-06,
+      "loss": 2.2296,
+      "step": 5902
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4499361559483976e-06,
+      "loss": 2.1894,
+      "step": 5903
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4482769093609945e-06,
+      "loss": 2.22,
+      "step": 5904
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4466178518463424e-06,
+      "loss": 2.1884,
+      "step": 5905
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.444958983606592e-06,
+      "loss": 2.1842,
+      "step": 5906
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4433003048438748e-06,
+      "loss": 2.1448,
+      "step": 5907
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.441641815760291e-06,
+      "loss": 2.188,
+      "step": 5908
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4399835165579266e-06,
+      "loss": 2.2261,
+      "step": 5909
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4383254074388373e-06,
+      "loss": 2.1723,
+      "step": 5910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4366674886050618e-06,
+      "loss": 2.2442,
+      "step": 5911
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4350097602586085e-06,
+      "loss": 2.198,
+      "step": 5912
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4333522226014715e-06,
+      "loss": 2.2805,
+      "step": 5913
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4316948758356127e-06,
+      "loss": 2.2544,
+      "step": 5914
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4300377201629753e-06,
+      "loss": 2.1466,
+      "step": 5915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4283807557854814e-06,
+      "loss": 2.1718,
+      "step": 5916
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.426723982905023e-06,
+      "loss": 2.2887,
+      "step": 5917
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4250674017234774e-06,
+      "loss": 2.221,
+      "step": 5918
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4234110124426893e-06,
+      "loss": 2.2088,
+      "step": 5919
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4217548152644887e-06,
+      "loss": 2.2185,
+      "step": 5920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4200988103906747e-06,
+      "loss": 2.1363,
+      "step": 5921
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4184429980230305e-06,
+      "loss": 2.1758,
+      "step": 5922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4167873783633087e-06,
+      "loss": 2.2413,
+      "step": 5923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4151319516132414e-06,
+      "loss": 2.2533,
+      "step": 5924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4134767179745404e-06,
+      "loss": 2.1366,
+      "step": 5925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.411821677648887e-06,
+      "loss": 2.2228,
+      "step": 5926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4101668308379465e-06,
+      "loss": 2.0968,
+      "step": 5927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4085121777433532e-06,
+      "loss": 2.1966,
+      "step": 5928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4068577185667253e-06,
+      "loss": 2.216,
+      "step": 5929
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.40520345350965e-06,
+      "loss": 2.1115,
+      "step": 5930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.403549382773699e-06,
+      "loss": 2.2405,
+      "step": 5931
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.401895506560411e-06,
+      "loss": 2.1851,
+      "step": 5932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.400241825071309e-06,
+      "loss": 2.2237,
+      "step": 5933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3985883385078875e-06,
+      "loss": 2.1699,
+      "step": 5934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.396935047071619e-06,
+      "loss": 2.2575,
+      "step": 5935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3952819509639534e-06,
+      "loss": 2.2181,
+      "step": 5936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3936290503863132e-06,
+      "loss": 2.1662,
+      "step": 5937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3919763455401016e-06,
+      "loss": 2.2727,
+      "step": 5938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3903238366266956e-06,
+      "loss": 2.0583,
+      "step": 5939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3886715238474454e-06,
+      "loss": 2.2156,
+      "step": 5940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3870194074036846e-06,
+      "loss": 2.1605,
+      "step": 5941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3853674874967134e-06,
+      "loss": 2.2082,
+      "step": 5942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3837157643278173e-06,
+      "loss": 2.2527,
+      "step": 5943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3820642380982527e-06,
+      "loss": 2.2344,
+      "step": 5944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3804129090092542e-06,
+      "loss": 2.2012,
+      "step": 5945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.378761777262028e-06,
+      "loss": 2.177,
+      "step": 5946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3771108430577624e-06,
+      "loss": 2.2344,
+      "step": 5947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.375460106597619e-06,
+      "loss": 2.2454,
+      "step": 5948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.373809568082732e-06,
+      "loss": 2.264,
+      "step": 5949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.372159227714218e-06,
+      "loss": 2.2123,
+      "step": 5950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3705090856931626e-06,
+      "loss": 2.2101,
+      "step": 5951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3688591422206333e-06,
+      "loss": 2.1895,
+      "step": 5952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.367209397497669e-06,
+      "loss": 2.2307,
+      "step": 5953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3655598517252886e-06,
+      "loss": 2.2343,
+      "step": 5954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3639105051044807e-06,
+      "loss": 2.1956,
+      "step": 5955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3622613578362162e-06,
+      "loss": 2.198,
+      "step": 5956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.360612410121438e-06,
+      "loss": 2.2292,
+      "step": 5957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.358963662161062e-06,
+      "loss": 2.2347,
+      "step": 5958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.357315114155989e-06,
+      "loss": 2.1713,
+      "step": 5959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.355666766307084e-06,
+      "loss": 2.2132,
+      "step": 5960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3540186188151976e-06,
+      "loss": 2.1924,
+      "step": 5961
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.352370671881148e-06,
+      "loss": 2.1676,
+      "step": 5962
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.350722925705736e-06,
+      "loss": 2.1525,
+      "step": 5963
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3490753804897315e-06,
+      "loss": 2.2077,
+      "step": 5964
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.347428036433886e-06,
+      "loss": 2.2105,
+      "step": 5965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.34578089373892e-06,
+      "loss": 2.197,
+      "step": 5966
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.344133952605534e-06,
+      "loss": 2.1897,
+      "step": 5967
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3424872132344044e-06,
+      "loss": 2.2463,
+      "step": 5968
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.340840675826178e-06,
+      "loss": 2.2442,
+      "step": 5969
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.339194340581485e-06,
+      "loss": 2.195,
+      "step": 5970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3375482077009213e-06,
+      "loss": 2.1896,
+      "step": 5971
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3359022773850673e-06,
+      "loss": 2.176,
+      "step": 5972
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.334256549834472e-06,
+      "loss": 2.2083,
+      "step": 5973
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3326110252496652e-06,
+      "loss": 2.1974,
+      "step": 5974
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.330965703831146e-06,
+      "loss": 2.2575,
+      "step": 5975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3293205857793924e-06,
+      "loss": 2.2297,
+      "step": 5976
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.32767567129486e-06,
+      "loss": 2.185,
+      "step": 5977
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3260309605779717e-06,
+      "loss": 2.1575,
+      "step": 5978
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3243864538291358e-06,
+      "loss": 2.2291,
+      "step": 5979
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.322742151248726e-06,
+      "loss": 2.221,
+      "step": 5980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3210980530370974e-06,
+      "loss": 2.1568,
+      "step": 5981
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.319454159394578e-06,
+      "loss": 2.1214,
+      "step": 5982
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.317810470521473e-06,
+      "loss": 2.2272,
+      "step": 5983
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.31616698661806e-06,
+      "loss": 2.1961,
+      "step": 5984
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3145237078845903e-06,
+      "loss": 2.1819,
+      "step": 5985
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.312880634521295e-06,
+      "loss": 2.2178,
+      "step": 5986
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3112377667283756e-06,
+      "loss": 2.1338,
+      "step": 5987
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3095951047060147e-06,
+      "loss": 2.2218,
+      "step": 5988
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.30795264865436e-06,
+      "loss": 2.1417,
+      "step": 5989
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3063103987735433e-06,
+      "loss": 2.1429,
+      "step": 5990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3046683552636665e-06,
+      "loss": 2.1441,
+      "step": 5991
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.30302651832481e-06,
+      "loss": 2.2077,
+      "step": 5992
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3013848881570243e-06,
+      "loss": 2.1343,
+      "step": 5993
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2997434649603368e-06,
+      "loss": 2.1888,
+      "step": 5994
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2981022489347503e-06,
+      "loss": 2.1937,
+      "step": 5995
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2964612402802422e-06,
+      "loss": 2.1072,
+      "step": 5996
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2948204391967657e-06,
+      "loss": 2.1996,
+      "step": 5997
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.293179845884245e-06,
+      "loss": 2.2439,
+      "step": 5998
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2915394605425836e-06,
+      "loss": 2.222,
+      "step": 5999
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.289899283371657e-06,
+      "loss": 2.1732,
+      "step": 6000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2882593145713148e-06,
+      "loss": 2.1912,
+      "step": 6001
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2866195543413843e-06,
+      "loss": 2.1774,
+      "step": 6002
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2849800028816613e-06,
+      "loss": 2.2352,
+      "step": 6003
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2833406603919243e-06,
+      "loss": 2.1968,
+      "step": 6004
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.28170152707192e-06,
+      "loss": 2.1545,
+      "step": 6005
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.280062603121373e-06,
+      "loss": 2.185,
+      "step": 6006
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.27842388873998e-06,
+      "loss": 2.1439,
+      "step": 6007
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2767853841274154e-06,
+      "loss": 2.1906,
+      "step": 6008
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2751470894833236e-06,
+      "loss": 2.2533,
+      "step": 6009
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.273509005007327e-06,
+      "loss": 2.2252,
+      "step": 6010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2718711308990226e-06,
+      "loss": 2.2469,
+      "step": 6011
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2702334673579765e-06,
+      "loss": 2.2143,
+      "step": 6012
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.268596014583737e-06,
+      "loss": 2.219,
+      "step": 6013
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.26695877277582e-06,
+      "loss": 2.1973,
+      "step": 6014
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2653217421337213e-06,
+      "loss": 2.2311,
+      "step": 6015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.263684922856905e-06,
+      "loss": 2.2452,
+      "step": 6016
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.262048315144816e-06,
+      "loss": 2.1369,
+      "step": 6017
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.260411919196866e-06,
+      "loss": 2.185,
+      "step": 6018
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.258775735212447e-06,
+      "loss": 2.1247,
+      "step": 6019
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2571397633909252e-06,
+      "loss": 2.1581,
+      "step": 6020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2555040039316344e-06,
+      "loss": 2.2228,
+      "step": 6021
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2538684570338908e-06,
+      "loss": 2.2186,
+      "step": 6022
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2522331228969774e-06,
+      "loss": 2.2359,
+      "step": 6023
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2505980017201564e-06,
+      "loss": 2.2036,
+      "step": 6024
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.248963093702663e-06,
+      "loss": 2.1725,
+      "step": 6025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.247328399043706e-06,
+      "loss": 2.2511,
+      "step": 6026
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.245693917942465e-06,
+      "loss": 2.1838,
+      "step": 6027
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2440596505981005e-06,
+      "loss": 2.1933,
+      "step": 6028
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.242425597209742e-06,
+      "loss": 2.1562,
+      "step": 6029
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2407917579764914e-06,
+      "loss": 2.2084,
+      "step": 6030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2391581330974307e-06,
+      "loss": 2.1492,
+      "step": 6031
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2375247227716077e-06,
+      "loss": 2.1712,
+      "step": 6032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.235891527198053e-06,
+      "loss": 2.1966,
+      "step": 6033
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2342585465757625e-06,
+      "loss": 2.2601,
+      "step": 6034
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2326257811037154e-06,
+      "loss": 2.1493,
+      "step": 6035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.230993230980853e-06,
+      "loss": 2.144,
+      "step": 6036
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.229360896406102e-06,
+      "loss": 2.1961,
+      "step": 6037
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.227728777578353e-06,
+      "loss": 2.2199,
+      "step": 6038
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.226096874696476e-06,
+      "loss": 2.1351,
+      "step": 6039
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.224465187959316e-06,
+      "loss": 2.1528,
+      "step": 6040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2228337175656856e-06,
+      "loss": 2.1085,
+      "step": 6041
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2212024637143756e-06,
+      "loss": 2.257,
+      "step": 6042
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.21957142660415e-06,
+      "loss": 2.2198,
+      "step": 6043
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.217940606433747e-06,
+      "loss": 2.2171,
+      "step": 6044
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2163100034018735e-06,
+      "loss": 2.213,
+      "step": 6045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2146796177072183e-06,
+      "loss": 2.2369,
+      "step": 6046
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2130494495484345e-06,
+      "loss": 2.2104,
+      "step": 6047
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.211419499124154e-06,
+      "loss": 2.2146,
+      "step": 6048
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.209789766632984e-06,
+      "loss": 2.2407,
+      "step": 6049
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2081602522734987e-06,
+      "loss": 2.15,
+      "step": 6050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2065309562442536e-06,
+      "loss": 2.2445,
+      "step": 6051
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2049018787437693e-06,
+      "loss": 2.2231,
+      "step": 6052
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2032730199705477e-06,
+      "loss": 2.2111,
+      "step": 6053
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.201644380123056e-06,
+      "loss": 2.2233,
+      "step": 6054
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2000159593997447e-06,
+      "loss": 2.2822,
+      "step": 6055
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1983877579990276e-06,
+      "loss": 2.2624,
+      "step": 6056
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.196759776119296e-06,
+      "loss": 2.1768,
+      "step": 6057
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.195132013958918e-06,
+      "loss": 2.1637,
+      "step": 6058
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.193504471716228e-06,
+      "loss": 2.2094,
+      "step": 6059
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1918771495895395e-06,
+      "loss": 2.1867,
+      "step": 6060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.190250047777134e-06,
+      "loss": 2.2381,
+      "step": 6061
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.188623166477272e-06,
+      "loss": 2.2451,
+      "step": 6062
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1869965058881836e-06,
+      "loss": 2.2466,
+      "step": 6063
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.185370066208069e-06,
+      "loss": 2.1903,
+      "step": 6064
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.183743847635109e-06,
+      "loss": 2.1602,
+      "step": 6065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1821178503674515e-06,
+      "loss": 2.2489,
+      "step": 6066
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1804920746032197e-06,
+      "loss": 2.1942,
+      "step": 6067
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.178866520540509e-06,
+      "loss": 2.1842,
+      "step": 6068
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.17724118837739e-06,
+      "loss": 2.1702,
+      "step": 6069
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1756160783119015e-06,
+      "loss": 2.168,
+      "step": 6070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1739911905420617e-06,
+      "loss": 2.1849,
+      "step": 6071
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1723665252658564e-06,
+      "loss": 2.1021,
+      "step": 6072
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.170742082681244e-06,
+      "loss": 2.1721,
+      "step": 6073
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.169117862986163e-06,
+      "loss": 2.1681,
+      "step": 6074
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.167493866378514e-06,
+      "loss": 2.1814,
+      "step": 6075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.16587009305618e-06,
+      "loss": 2.2367,
+      "step": 6076
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.164246543217011e-06,
+      "loss": 2.1699,
+      "step": 6077
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1626232170588343e-06,
+      "loss": 2.2158,
+      "step": 6078
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.161000114779443e-06,
+      "loss": 2.1467,
+      "step": 6079
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1593772365766107e-06,
+      "loss": 2.1998,
+      "step": 6080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.157754582648079e-06,
+      "loss": 2.2073,
+      "step": 6081
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1561321531915622e-06,
+      "loss": 2.1159,
+      "step": 6082
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1545099484047514e-06,
+      "loss": 2.2319,
+      "step": 6083
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.152887968485303e-06,
+      "loss": 2.2339,
+      "step": 6084
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.151266213630854e-06,
+      "loss": 2.2527,
+      "step": 6085
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.149644684039008e-06,
+      "loss": 2.1695,
+      "step": 6086
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1480233799073467e-06,
+      "loss": 2.1994,
+      "step": 6087
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1464023014334164e-06,
+      "loss": 2.1938,
+      "step": 6088
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.144781448814746e-06,
+      "loss": 2.1141,
+      "step": 6089
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1431608222488276e-06,
+      "loss": 2.1112,
+      "step": 6090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1415404219331287e-06,
+      "loss": 2.2526,
+      "step": 6091
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.139920248065095e-06,
+      "loss": 2.2011,
+      "step": 6092
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1383003008421336e-06,
+      "loss": 2.19,
+      "step": 6093
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1366805804616353e-06,
+      "loss": 2.1705,
+      "step": 6094
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1350610871209553e-06,
+      "loss": 2.2127,
+      "step": 6095
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1334418210174268e-06,
+      "loss": 2.1764,
+      "step": 6096
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1318227823483483e-06,
+      "loss": 2.2241,
+      "step": 6097
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.130203971310999e-06,
+      "loss": 2.2232,
+      "step": 6098
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.128585388102623e-06,
+      "loss": 2.2129,
+      "step": 6099
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.12696703292044e-06,
+      "loss": 2.1074,
+      "step": 6100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1253489059616448e-06,
+      "loss": 2.1455,
+      "step": 6101
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1237310074233964e-06,
+      "loss": 2.1281,
+      "step": 6102
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1221133375028357e-06,
+      "loss": 2.1783,
+      "step": 6103
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1204958963970666e-06,
+      "loss": 2.177,
+      "step": 6104
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1188786843031727e-06,
+      "loss": 2.2021,
+      "step": 6105
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.117261701418204e-06,
+      "loss": 2.2287,
+      "step": 6106
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1156449479391876e-06,
+      "loss": 2.176,
+      "step": 6107
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.114028424063118e-06,
+      "loss": 2.1505,
+      "step": 6108
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1124121299869627e-06,
+      "loss": 2.1883,
+      "step": 6109
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.110796065907665e-06,
+      "loss": 2.1443,
+      "step": 6110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.109180232022135e-06,
+      "loss": 2.2097,
+      "step": 6111
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1075646285272608e-06,
+      "loss": 2.1343,
+      "step": 6112
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1059492556198933e-06,
+      "loss": 2.166,
+      "step": 6113
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1043341134968653e-06,
+      "loss": 2.1853,
+      "step": 6114
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.102719202354974e-06,
+      "loss": 2.1478,
+      "step": 6115
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1011045223909954e-06,
+      "loss": 2.2793,
+      "step": 6116
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0994900738016693e-06,
+      "loss": 2.1358,
+      "step": 6117
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.097875856783713e-06,
+      "loss": 2.1925,
+      "step": 6118
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0962618715338135e-06,
+      "loss": 2.1877,
+      "step": 6119
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.09464811824863e-06,
+      "loss": 2.2165,
+      "step": 6120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.093034597124795e-06,
+      "loss": 2.18,
+      "step": 6121
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0914213083589086e-06,
+      "loss": 2.1487,
+      "step": 6122
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0898082521475463e-06,
+      "loss": 2.2086,
+      "step": 6123
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.088195428687254e-06,
+      "loss": 2.2934,
+      "step": 6124
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0865828381745515e-06,
+      "loss": 2.16,
+      "step": 6125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0849704808059266e-06,
+      "loss": 2.2045,
+      "step": 6126
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.083358356777837e-06,
+      "loss": 2.2037,
+      "step": 6127
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0817464662867192e-06,
+      "loss": 2.1946,
+      "step": 6128
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.080134809528975e-06,
+      "loss": 2.1427,
+      "step": 6129
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.078523386700982e-06,
+      "loss": 2.1992,
+      "step": 6130
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0769121979990845e-06,
+      "loss": 2.155,
+      "step": 6131
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0753012436196033e-06,
+      "loss": 2.1703,
+      "step": 6132
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0736905237588275e-06,
+      "loss": 2.1928,
+      "step": 6133
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0720800386130176e-06,
+      "loss": 2.2142,
+      "step": 6134
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0704697883784083e-06,
+      "loss": 2.2015,
+      "step": 6135
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0688597732512004e-06,
+      "loss": 2.1775,
+      "step": 6136
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0672499934275725e-06,
+      "loss": 2.1312,
+      "step": 6137
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0656404491036696e-06,
+      "loss": 2.1922,
+      "step": 6138
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0640311404756116e-06,
+      "loss": 2.2959,
+      "step": 6139
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0624220677394854e-06,
+      "loss": 2.2685,
+      "step": 6140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.060813231091354e-06,
+      "loss": 2.2288,
+      "step": 6141
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.059204630727247e-06,
+      "loss": 2.2087,
+      "step": 6142
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0575962668431704e-06,
+      "loss": 2.144,
+      "step": 6143
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0559881396350967e-06,
+      "loss": 2.1393,
+      "step": 6144
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0543802492989693e-06,
+      "loss": 2.1606,
+      "step": 6145
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0527725960307083e-06,
+      "loss": 2.2017,
+      "step": 6146
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0511651800261987e-06,
+      "loss": 2.2115,
+      "step": 6147
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.049558001481302e-06,
+      "loss": 2.1681,
+      "step": 6148
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.047951060591845e-06,
+      "loss": 2.1818,
+      "step": 6149
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0463443575536324e-06,
+      "loss": 2.215,
+      "step": 6150
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0447378925624316e-06,
+      "loss": 2.2447,
+      "step": 6151
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.043131665813988e-06,
+      "loss": 2.1974,
+      "step": 6152
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.041525677504017e-06,
+      "loss": 2.2355,
+      "step": 6153
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0399199278281986e-06,
+      "loss": 2.146,
+      "step": 6154
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0383144169821944e-06,
+      "loss": 2.1156,
+      "step": 6155
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0367091451616254e-06,
+      "loss": 2.225,
+      "step": 6156
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.035104112562093e-06,
+      "loss": 2.1661,
+      "step": 6157
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.033499319379163e-06,
+      "loss": 2.1623,
+      "step": 6158
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0318947658083787e-06,
+      "loss": 2.1416,
+      "step": 6159
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.030290452045245e-06,
+      "loss": 2.1506,
+      "step": 6160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.028686378285245e-06,
+      "loss": 2.2227,
+      "step": 6161
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0270825447238316e-06,
+      "loss": 2.1699,
+      "step": 6162
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.025478951556424e-06,
+      "loss": 2.1915,
+      "step": 6163
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.023875598978419e-06,
+      "loss": 2.2302,
+      "step": 6164
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.022272487185176e-06,
+      "loss": 2.1662,
+      "step": 6165
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0206696163720317e-06,
+      "loss": 2.1589,
+      "step": 6166
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0190669867342902e-06,
+      "loss": 2.1204,
+      "step": 6167
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0174645984672298e-06,
+      "loss": 2.1737,
+      "step": 6168
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.015862451766092e-06,
+      "loss": 2.2607,
+      "step": 6169
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0142605468260976e-06,
+      "loss": 2.1886,
+      "step": 6170
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.012658883842432e-06,
+      "loss": 2.1569,
+      "step": 6171
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.011057463010252e-06,
+      "loss": 2.1885,
+      "step": 6172
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.009456284524688e-06,
+      "loss": 2.1903,
+      "step": 6173
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.007855348580837e-06,
+      "loss": 2.1972,
+      "step": 6174
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0062546553737692e-06,
+      "loss": 2.2486,
+      "step": 6175
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.004654205098524e-06,
+      "loss": 2.1391,
+      "step": 6176
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.003053997950112e-06,
+      "loss": 2.2097,
+      "step": 6177
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.001454034123512e-06,
+      "loss": 2.0991,
+      "step": 6178
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9998543138136773e-06,
+      "loss": 2.1712,
+      "step": 6179
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9982548372155264e-06,
+      "loss": 2.2028,
+      "step": 6180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9966556045239504e-06,
+      "loss": 2.2171,
+      "step": 6181
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9950566159338146e-06,
+      "loss": 2.2269,
+      "step": 6182
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9934578716399465e-06,
+      "loss": 2.147,
+      "step": 6183
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.991859371837151e-06,
+      "loss": 2.1365,
+      "step": 6184
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9902611167202e-06,
+      "loss": 2.2113,
+      "step": 6185
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9886631064838355e-06,
+      "loss": 2.2423,
+      "step": 6186
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9870653413227692e-06,
+      "loss": 2.1507,
+      "step": 6187
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9854678214316875e-06,
+      "loss": 2.1763,
+      "step": 6188
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9838705470052397e-06,
+      "loss": 2.1631,
+      "step": 6189
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.98227351823805e-06,
+      "loss": 2.2423,
+      "step": 6190
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9806767353247127e-06,
+      "loss": 2.1874,
+      "step": 6191
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9790801984597885e-06,
+      "loss": 2.1523,
+      "step": 6192
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.977483907837814e-06,
+      "loss": 2.1738,
+      "step": 6193
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9758878636532884e-06,
+      "loss": 2.1981,
+      "step": 6194
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.974292066100688e-06,
+      "loss": 2.1974,
+      "step": 6195
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.972696515374455e-06,
+      "loss": 2.248,
+      "step": 6196
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9711012116690007e-06,
+      "loss": 2.1521,
+      "step": 6197
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.969506155178711e-06,
+      "loss": 2.1854,
+      "step": 6198
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9679113460979347e-06,
+      "loss": 2.1531,
+      "step": 6199
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.966316784621e-06,
+      "loss": 2.2509,
+      "step": 6200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.964722470942194e-06,
+      "loss": 2.2442,
+      "step": 6201
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.963128405255783e-06,
+      "loss": 2.174,
+      "step": 6202
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9615345877559953e-06,
+      "loss": 2.1868,
+      "step": 6203
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9599410186370363e-06,
+      "loss": 2.1805,
+      "step": 6204
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9583476980930768e-06,
+      "loss": 2.1572,
+      "step": 6205
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9567546263182554e-06,
+      "loss": 2.2302,
+      "step": 6206
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9551618035066863e-06,
+      "loss": 2.2477,
+      "step": 6207
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9535692298524477e-06,
+      "loss": 2.1975,
+      "step": 6208
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9519769055495917e-06,
+      "loss": 2.2232,
+      "step": 6209
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9503848307921363e-06,
+      "loss": 2.1882,
+      "step": 6210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.948793005774074e-06,
+      "loss": 2.1869,
+      "step": 6211
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9472014306893605e-06,
+      "loss": 2.192,
+      "step": 6212
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9456101057319266e-06,
+      "loss": 2.1569,
+      "step": 6213
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.94401903109567e-06,
+      "loss": 2.2008,
+      "step": 6214
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9424282069744564e-06,
+      "loss": 2.2319,
+      "step": 6215
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.940837633562127e-06,
+      "loss": 2.18,
+      "step": 6216
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9392473110524834e-06,
+      "loss": 2.2321,
+      "step": 6217
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9376572396393047e-06,
+      "loss": 2.2377,
+      "step": 6218
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9360674195163354e-06,
+      "loss": 2.196,
+      "step": 6219
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.934477850877292e-06,
+      "loss": 2.1647,
+      "step": 6220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9328885339158554e-06,
+      "loss": 2.2192,
+      "step": 6221
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.931299468825682e-06,
+      "loss": 2.1701,
+      "step": 6222
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.929710655800393e-06,
+      "loss": 2.2008,
+      "step": 6223
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.92812209503358e-06,
+      "loss": 2.1817,
+      "step": 6224
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.926533786718806e-06,
+      "loss": 2.2367,
+      "step": 6225
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9249457310495994e-06,
+      "loss": 2.1107,
+      "step": 6226
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9233579282194617e-06,
+      "loss": 2.1609,
+      "step": 6227
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.921770378421861e-06,
+      "loss": 2.1569,
+      "step": 6228
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.920183081850237e-06,
+      "loss": 2.1729,
+      "step": 6229
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.918596038697995e-06,
+      "loss": 2.2335,
+      "step": 6230
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9170092491585122e-06,
+      "loss": 2.2384,
+      "step": 6231
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.915422713425134e-06,
+      "loss": 2.1315,
+      "step": 6232
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9138364316911747e-06,
+      "loss": 2.1597,
+      "step": 6233
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.912250404149918e-06,
+      "loss": 2.1775,
+      "step": 6234
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.910664630994615e-06,
+      "loss": 2.2035,
+      "step": 6235
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9090791124184934e-06,
+      "loss": 2.1383,
+      "step": 6236
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9074938486147357e-06,
+      "loss": 2.1848,
+      "step": 6237
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.905908839776509e-06,
+      "loss": 2.2066,
+      "step": 6238
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9043240860969342e-06,
+      "loss": 2.1583,
+      "step": 6239
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9027395877691143e-06,
+      "loss": 2.172,
+      "step": 6240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9011553449861163e-06,
+      "loss": 2.2255,
+      "step": 6241
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.899571357940969e-06,
+      "loss": 2.203,
+      "step": 6242
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.897987626826682e-06,
+      "loss": 2.14,
+      "step": 6243
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.896404151836227e-06,
+      "loss": 2.1764,
+      "step": 6244
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8948209331625454e-06,
+      "loss": 2.1636,
+      "step": 6245
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.893237970998547e-06,
+      "loss": 2.2005,
+      "step": 6246
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8916552655371117e-06,
+      "loss": 2.1952,
+      "step": 6247
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8900728169710866e-06,
+      "loss": 2.2352,
+      "step": 6248
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.888490625493289e-06,
+      "loss": 2.1257,
+      "step": 6249
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.886908691296504e-06,
+      "loss": 2.2444,
+      "step": 6250
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8853270145734846e-06,
+      "loss": 2.1347,
+      "step": 6251
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8837455955169547e-06,
+      "loss": 2.1392,
+      "step": 6252
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8821644343196042e-06,
+      "loss": 2.2514,
+      "step": 6253
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8805835311740933e-06,
+      "loss": 2.1832,
+      "step": 6254
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8790028862730487e-06,
+      "loss": 2.2407,
+      "step": 6255
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.877422499809072e-06,
+      "loss": 2.2406,
+      "step": 6256
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.875842371974722e-06,
+      "loss": 2.1862,
+      "step": 6257
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.874262502962537e-06,
+      "loss": 2.1646,
+      "step": 6258
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8726828929650206e-06,
+      "loss": 2.1654,
+      "step": 6259
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.871103542174637e-06,
+      "loss": 2.2081,
+      "step": 6260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8695244507838326e-06,
+      "loss": 2.1978,
+      "step": 6261
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8679456189850076e-06,
+      "loss": 2.1884,
+      "step": 6262
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8663670469705434e-06,
+      "loss": 2.1505,
+      "step": 6263
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.864788734932783e-06,
+      "loss": 2.2231,
+      "step": 6264
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.863210683064038e-06,
+      "loss": 2.2518,
+      "step": 6265
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8616328915565907e-06,
+      "loss": 2.1735,
+      "step": 6266
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8600553606026883e-06,
+      "loss": 2.2333,
+      "step": 6267
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.858478090394549e-06,
+      "loss": 2.2689,
+      "step": 6268
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.856901081124359e-06,
+      "loss": 2.0632,
+      "step": 6269
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8553243329842715e-06,
+      "loss": 2.2364,
+      "step": 6270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.853747846166408e-06,
+      "loss": 2.2484,
+      "step": 6271
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8521716208628597e-06,
+      "loss": 2.2025,
+      "step": 6272
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.850595657265682e-06,
+      "loss": 2.2091,
+      "step": 6273
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.849019955566908e-06,
+      "loss": 2.2169,
+      "step": 6274
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8474445159585235e-06,
+      "loss": 2.1802,
+      "step": 6275
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8458693386325e-06,
+      "loss": 2.1114,
+      "step": 6276
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.84429442378076e-06,
+      "loss": 2.1453,
+      "step": 6277
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8427197715952047e-06,
+      "loss": 2.1647,
+      "step": 6278
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.841145382267705e-06,
+      "loss": 2.1945,
+      "step": 6279
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.839571255990088e-06,
+      "loss": 2.2349,
+      "step": 6280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.837997392954165e-06,
+      "loss": 2.1644,
+      "step": 6281
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8364237933516964e-06,
+      "loss": 2.1653,
+      "step": 6282
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8348504573744283e-06,
+      "loss": 2.1501,
+      "step": 6283
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8332773852140644e-06,
+      "loss": 2.2317,
+      "step": 6284
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8317045770622784e-06,
+      "loss": 2.1615,
+      "step": 6285
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.830132033110713e-06,
+      "loss": 2.1695,
+      "step": 6286
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8285597535509775e-06,
+      "loss": 2.1869,
+      "step": 6287
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.826987738574649e-06,
+      "loss": 2.194,
+      "step": 6288
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8254159883732735e-06,
+      "loss": 2.1601,
+      "step": 6289
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8238445031383634e-06,
+      "loss": 2.276,
+      "step": 6290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8222732830613995e-06,
+      "loss": 2.115,
+      "step": 6291
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8207023283338304e-06,
+      "loss": 2.1392,
+      "step": 6292
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8191316391470703e-06,
+      "loss": 2.1893,
+      "step": 6293
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8175612156925082e-06,
+      "loss": 2.1662,
+      "step": 6294
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8159910581614904e-06,
+      "loss": 2.1793,
+      "step": 6295
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.814421166745337e-06,
+      "loss": 2.2471,
+      "step": 6296
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8128515416353345e-06,
+      "loss": 2.17,
+      "step": 6297
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.811282183022736e-06,
+      "loss": 2.165,
+      "step": 6298
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.809713091098768e-06,
+      "loss": 2.1522,
+      "step": 6299
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8081442660546126e-06,
+      "loss": 2.2396,
+      "step": 6300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8065757080814315e-06,
+      "loss": 2.1448,
+      "step": 6301
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.805007417370347e-06,
+      "loss": 2.18,
+      "step": 6302
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8034393941124505e-06,
+      "loss": 2.2217,
+      "step": 6303
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8018716384988034e-06,
+      "loss": 2.1307,
+      "step": 6304
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.800304150720424e-06,
+      "loss": 2.1912,
+      "step": 6305
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.798736930968315e-06,
+      "loss": 2.164,
+      "step": 6306
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7971699794334332e-06,
+      "loss": 2.1128,
+      "step": 6307
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.795603296306708e-06,
+      "loss": 2.1366,
+      "step": 6308
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.794036881779034e-06,
+      "loss": 2.2025,
+      "step": 6309
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7924707360412743e-06,
+      "loss": 2.2239,
+      "step": 6310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7909048592842602e-06,
+      "loss": 2.1619,
+      "step": 6311
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7893392516987873e-06,
+      "loss": 2.1832,
+      "step": 6312
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.787773913475621e-06,
+      "loss": 2.2033,
+      "step": 6313
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7862088448054936e-06,
+      "loss": 2.2039,
+      "step": 6314
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7846440458791024e-06,
+      "loss": 2.2221,
+      "step": 6315
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7830795168871127e-06,
+      "loss": 2.1804,
+      "step": 6316
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7815152580201637e-06,
+      "loss": 2.1676,
+      "step": 6317
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.779951269468847e-06,
+      "loss": 2.1861,
+      "step": 6318
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7783875514237373e-06,
+      "loss": 2.182,
+      "step": 6319
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.776824104075364e-06,
+      "loss": 2.2436,
+      "step": 6320
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7752609276142282e-06,
+      "loss": 2.1926,
+      "step": 6321
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7736980222308042e-06,
+      "loss": 2.2016,
+      "step": 6322
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.772135388115519e-06,
+      "loss": 2.1907,
+      "step": 6323
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7705730254587802e-06,
+      "loss": 2.1426,
+      "step": 6324
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7690109344509563e-06,
+      "loss": 2.1803,
+      "step": 6325
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7674491152823825e-06,
+      "loss": 2.1876,
+      "step": 6326
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.765887568143362e-06,
+      "loss": 2.1942,
+      "step": 6327
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7643262932241642e-06,
+      "loss": 2.2724,
+      "step": 6328
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7627652907150272e-06,
+      "loss": 2.1745,
+      "step": 6329
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.761204560806152e-06,
+      "loss": 2.2338,
+      "step": 6330
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.759644103687711e-06,
+      "loss": 2.1744,
+      "step": 6331
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7580839195498397e-06,
+      "loss": 2.2098,
+      "step": 6332
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7565240085826423e-06,
+      "loss": 2.1198,
+      "step": 6333
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.75496437097619e-06,
+      "loss": 2.1955,
+      "step": 6334
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.753405006920518e-06,
+      "loss": 2.1532,
+      "step": 6335
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.75184591660563e-06,
+      "loss": 2.1586,
+      "step": 6336
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7502871002215016e-06,
+      "loss": 2.2435,
+      "step": 6337
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7487285579580635e-06,
+      "loss": 2.2128,
+      "step": 6338
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7471702900052204e-06,
+      "loss": 2.1796,
+      "step": 6339
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7456122965528475e-06,
+      "loss": 2.1854,
+      "step": 6340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7440545777907747e-06,
+      "loss": 2.2084,
+      "step": 6341
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.742497133908812e-06,
+      "loss": 2.1507,
+      "step": 6342
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7409399650967217e-06,
+      "loss": 2.1692,
+      "step": 6343
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.739383071544246e-06,
+      "loss": 2.1796,
+      "step": 6344
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7378264534410865e-06,
+      "loss": 2.257,
+      "step": 6345
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.736270110976912e-06,
+      "loss": 2.1841,
+      "step": 6346
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.734714044341359e-06,
+      "loss": 2.1273,
+      "step": 6347
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7331582537240243e-06,
+      "loss": 2.1914,
+      "step": 6348
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7316027393144827e-06,
+      "loss": 2.2114,
+      "step": 6349
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7300475013022666e-06,
+      "loss": 2.1796,
+      "step": 6350
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.728492539876877e-06,
+      "loss": 2.153,
+      "step": 6351
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.726937855227781e-06,
+      "loss": 2.1628,
+      "step": 6352
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7253834475444126e-06,
+      "loss": 2.1685,
+      "step": 6353
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.723829317016169e-06,
+      "loss": 2.1639,
+      "step": 6354
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7222754638324235e-06,
+      "loss": 2.1974,
+      "step": 6355
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7207218881825016e-06,
+      "loss": 2.2083,
+      "step": 6356
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.719168590255703e-06,
+      "loss": 2.1451,
+      "step": 6357
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.717615570241294e-06,
+      "loss": 2.1809,
+      "step": 6358
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.716062828328502e-06,
+      "loss": 2.2218,
+      "step": 6359
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.714510364706531e-06,
+      "loss": 2.2455,
+      "step": 6360
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.712958179564535e-06,
+      "loss": 2.1649,
+      "step": 6361
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7114062730916513e-06,
+      "loss": 2.1557,
+      "step": 6362
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7098546454769683e-06,
+      "loss": 2.2524,
+      "step": 6363
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.708303296909551e-06,
+      "loss": 2.1754,
+      "step": 6364
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7067522275784275e-06,
+      "loss": 2.1927,
+      "step": 6365
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.705201437672585e-06,
+      "loss": 2.1579,
+      "step": 6366
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7036509273809887e-06,
+      "loss": 2.1764,
+      "step": 6367
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7021006968925613e-06,
+      "loss": 2.1006,
+      "step": 6368
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.700550746396193e-06,
+      "loss": 2.18,
+      "step": 6369
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.699001076080742e-06,
+      "loss": 2.1471,
+      "step": 6370
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.697451686135031e-06,
+      "loss": 2.2048,
+      "step": 6371
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6959025767478466e-06,
+      "loss": 2.1815,
+      "step": 6372
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.694353748107944e-06,
+      "loss": 2.2057,
+      "step": 6373
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.692805200404044e-06,
+      "loss": 2.2794,
+      "step": 6374
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6912569338248317e-06,
+      "loss": 2.1863,
+      "step": 6375
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6897089485589584e-06,
+      "loss": 2.1716,
+      "step": 6376
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6881612447950425e-06,
+      "loss": 2.1537,
+      "step": 6377
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.686613822721666e-06,
+      "loss": 2.1954,
+      "step": 6378
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6850666825273762e-06,
+      "loss": 2.2018,
+      "step": 6379
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.683519824400693e-06,
+      "loss": 2.1436,
+      "step": 6380
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6819732485300887e-06,
+      "loss": 2.2141,
+      "step": 6381
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.680426955104014e-06,
+      "loss": 2.1536,
+      "step": 6382
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.678880944310882e-06,
+      "loss": 2.2253,
+      "step": 6383
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.677335216339062e-06,
+      "loss": 2.1584,
+      "step": 6384
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.675789771376904e-06,
+      "loss": 2.2221,
+      "step": 6385
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6742446096127086e-06,
+      "loss": 2.2248,
+      "step": 6386
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6726997312347546e-06,
+      "loss": 2.2478,
+      "step": 6387
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.671155136431279e-06,
+      "loss": 2.2046,
+      "step": 6388
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6696108253904856e-06,
+      "loss": 2.205,
+      "step": 6389
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6680667983005446e-06,
+      "loss": 2.236,
+      "step": 6390
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.66652305534959e-06,
+      "loss": 2.181,
+      "step": 6391
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6649795967257243e-06,
+      "loss": 2.248,
+      "step": 6392
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6634364226170105e-06,
+      "loss": 2.1377,
+      "step": 6393
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.661893533211482e-06,
+      "loss": 2.1856,
+      "step": 6394
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6603509286971342e-06,
+      "loss": 2.1442,
+      "step": 6395
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.658808609261928e-06,
+      "loss": 2.1096,
+      "step": 6396
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6572665750937898e-06,
+      "loss": 2.2356,
+      "step": 6397
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6557248263806175e-06,
+      "loss": 2.2793,
+      "step": 6398
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.65418336331026e-06,
+      "loss": 2.2373,
+      "step": 6399
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6526421860705474e-06,
+      "loss": 2.1733,
+      "step": 6400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6511012948492625e-06,
+      "loss": 2.21,
+      "step": 6401
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.649560689834158e-06,
+      "loss": 2.1727,
+      "step": 6402
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6480203712129583e-06,
+      "loss": 2.184,
+      "step": 6403
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.646480339173337e-06,
+      "loss": 2.19,
+      "step": 6404
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.64494059390295e-06,
+      "loss": 2.2065,
+      "step": 6405
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6434011355894074e-06,
+      "loss": 2.2079,
+      "step": 6406
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.641861964420289e-06,
+      "loss": 2.1841,
+      "step": 6407
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.640323080583137e-06,
+      "loss": 2.1161,
+      "step": 6408
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.63878448426546e-06,
+      "loss": 2.2179,
+      "step": 6409
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.637246175654731e-06,
+      "loss": 2.2076,
+      "step": 6410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6357081549383877e-06,
+      "loss": 2.2714,
+      "step": 6411
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.634170422303835e-06,
+      "loss": 2.1336,
+      "step": 6412
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6326329779384397e-06,
+      "loss": 2.2449,
+      "step": 6413
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6310958220295356e-06,
+      "loss": 2.2132,
+      "step": 6414
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6295589547644195e-06,
+      "loss": 2.1798,
+      "step": 6415
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6280223763303546e-06,
+      "loss": 2.1722,
+      "step": 6416
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.626486086914566e-06,
+      "loss": 2.152,
+      "step": 6417
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6249500867042523e-06,
+      "loss": 2.1487,
+      "step": 6418
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.623414375886564e-06,
+      "loss": 2.1321,
+      "step": 6419
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6218789546486235e-06,
+      "loss": 2.1768,
+      "step": 6420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6203438231775224e-06,
+      "loss": 2.2194,
+      "step": 6421
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.618808981660304e-06,
+      "loss": 2.1551,
+      "step": 6422
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6172744302839925e-06,
+      "loss": 2.122,
+      "step": 6423
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.61574016923556e-06,
+      "loss": 2.183,
+      "step": 6424
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.614206198701958e-06,
+      "loss": 2.2643,
+      "step": 6425
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.612672518870093e-06,
+      "loss": 2.2287,
+      "step": 6426
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6111391299268406e-06,
+      "loss": 2.2362,
+      "step": 6427
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6096060320590393e-06,
+      "loss": 2.2189,
+      "step": 6428
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.608073225453492e-06,
+      "loss": 2.225,
+      "step": 6429
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6065407102969664e-06,
+      "loss": 2.1755,
+      "step": 6430
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6050084867761953e-06,
+      "loss": 2.214,
+      "step": 6431
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6034765550778753e-06,
+      "loss": 2.2015,
+      "step": 6432
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.601944915388668e-06,
+      "loss": 2.2605,
+      "step": 6433
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.600413567895198e-06,
+      "loss": 2.1746,
+      "step": 6434
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5988825127840547e-06,
+      "loss": 2.156,
+      "step": 6435
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5973517502417966e-06,
+      "loss": 2.2507,
+      "step": 6436
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5958212804549387e-06,
+      "loss": 2.1814,
+      "step": 6437
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5942911036099657e-06,
+      "loss": 2.1419,
+      "step": 6438
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5927612198933237e-06,
+      "loss": 2.1489,
+      "step": 6439
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5912316294914232e-06,
+      "loss": 2.2019,
+      "step": 6440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5897023325906458e-06,
+      "loss": 2.2159,
+      "step": 6441
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.588173329377324e-06,
+      "loss": 2.1885,
+      "step": 6442
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5866446200377688e-06,
+      "loss": 2.0864,
+      "step": 6443
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5851162047582477e-06,
+      "loss": 2.2468,
+      "step": 6444
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5835880837249884e-06,
+      "loss": 2.2238,
+      "step": 6445
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.582060257124195e-06,
+      "loss": 2.1687,
+      "step": 6446
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5805327251420205e-06,
+      "loss": 2.2305,
+      "step": 6447
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5790054879645964e-06,
+      "loss": 2.1312,
+      "step": 6448
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5774785457780107e-06,
+      "loss": 2.1967,
+      "step": 6449
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5759518987683154e-06,
+      "loss": 2.1639,
+      "step": 6450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5744255471215284e-06,
+      "loss": 2.1754,
+      "step": 6451
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5728994910236304e-06,
+      "loss": 2.1633,
+      "step": 6452
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.571373730660568e-06,
+      "loss": 2.1293,
+      "step": 6453
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5698482662182494e-06,
+      "loss": 2.2059,
+      "step": 6454
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.568323097882548e-06,
+      "loss": 2.2078,
+      "step": 6455
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5667982258393016e-06,
+      "loss": 2.181,
+      "step": 6456
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5652736502743105e-06,
+      "loss": 2.1421,
+      "step": 6457
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5637493713733376e-06,
+      "loss": 2.1888,
+      "step": 6458
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5622253893221176e-06,
+      "loss": 2.153,
+      "step": 6459
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.560701704306336e-06,
+      "loss": 2.197,
+      "step": 6460
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5591783165116563e-06,
+      "loss": 2.1961,
+      "step": 6461
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.557655226123693e-06,
+      "loss": 2.1743,
+      "step": 6462
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.55613243332803e-06,
+      "loss": 2.2373,
+      "step": 6463
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5546099383102206e-06,
+      "loss": 2.2205,
+      "step": 6464
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5530877412557684e-06,
+      "loss": 2.1992,
+      "step": 6465
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5515658423501573e-06,
+      "loss": 2.1241,
+      "step": 6466
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5500442417788176e-06,
+      "loss": 2.1982,
+      "step": 6467
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5485229397271567e-06,
+      "loss": 2.1453,
+      "step": 6468
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.54700193638054e-06,
+      "loss": 2.1804,
+      "step": 6469
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.545481231924296e-06,
+      "loss": 2.1672,
+      "step": 6470
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5439608265437186e-06,
+      "loss": 2.2438,
+      "step": 6471
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5424407204240653e-06,
+      "loss": 2.1278,
+      "step": 6472
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.540920913750555e-06,
+      "loss": 2.2156,
+      "step": 6473
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.539401406708373e-06,
+      "loss": 2.1645,
+      "step": 6474
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5378821994826654e-06,
+      "loss": 2.1765,
+      "step": 6475
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.536363292258543e-06,
+      "loss": 2.1783,
+      "step": 6476
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5348446852210807e-06,
+      "loss": 2.1352,
+      "step": 6477
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.533326378555314e-06,
+      "loss": 2.1782,
+      "step": 6478
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5318083724462495e-06,
+      "loss": 2.1932,
+      "step": 6479
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5302906670788463e-06,
+      "loss": 2.1691,
+      "step": 6480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.528773262638034e-06,
+      "loss": 2.1596,
+      "step": 6481
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.527256159308703e-06,
+      "loss": 2.2252,
+      "step": 6482
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5257393572757073e-06,
+      "loss": 2.1498,
+      "step": 6483
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.524222856723869e-06,
+      "loss": 2.1884,
+      "step": 6484
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5227066578379624e-06,
+      "loss": 2.1822,
+      "step": 6485
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5211907608027366e-06,
+      "loss": 2.1985,
+      "step": 6486
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5196751658028972e-06,
+      "loss": 2.198,
+      "step": 6487
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.518159873023116e-06,
+      "loss": 2.0856,
+      "step": 6488
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5166448826480274e-06,
+      "loss": 2.1906,
+      "step": 6489
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5151301948622235e-06,
+      "loss": 2.1766,
+      "step": 6490
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.51361580985027e-06,
+      "loss": 2.1847,
+      "step": 6491
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5121017277966875e-06,
+      "loss": 2.222,
+      "step": 6492
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5105879488859635e-06,
+      "loss": 2.15,
+      "step": 6493
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.509074473302546e-06,
+      "loss": 2.1837,
+      "step": 6494
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.507561301230849e-06,
+      "loss": 2.1731,
+      "step": 6495
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.506048432855247e-06,
+      "loss": 2.1237,
+      "step": 6496
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.504535868360078e-06,
+      "loss": 2.1702,
+      "step": 6497
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5030236079296443e-06,
+      "loss": 2.079,
+      "step": 6498
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5015116517482097e-06,
+      "loss": 2.1307,
+      "step": 6499
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5000000000000015e-06,
+      "loss": 2.2222,
+      "step": 6500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4984886528692076e-06,
+      "loss": 2.1684,
+      "step": 6501
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.496977610539988e-06,
+      "loss": 2.1261,
+      "step": 6502
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.49546687319645e-06,
+      "loss": 2.1317,
+      "step": 6503
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.49395644102268e-06,
+      "loss": 2.1186,
+      "step": 6504
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.492446314202713e-06,
+      "loss": 2.1888,
+      "step": 6505
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4909364929205575e-06,
+      "loss": 2.2486,
+      "step": 6506
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.489426977360182e-06,
+      "loss": 2.1554,
+      "step": 6507
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.48791776770551e-06,
+      "loss": 2.2334,
+      "step": 6508
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4864088641404398e-06,
+      "loss": 2.2221,
+      "step": 6509
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.484900266848825e-06,
+      "loss": 2.2323,
+      "step": 6510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4833919760144838e-06,
+      "loss": 2.1804,
+      "step": 6511
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4818839918211963e-06,
+      "loss": 2.1984,
+      "step": 6512
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.480376314452706e-06,
+      "loss": 2.1056,
+      "step": 6513
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4788689440927193e-06,
+      "loss": 2.1364,
+      "step": 6514
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4773618809249045e-06,
+      "loss": 2.147,
+      "step": 6515
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4758551251328923e-06,
+      "loss": 2.1537,
+      "step": 6516
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4743486769002767e-06,
+      "loss": 2.2289,
+      "step": 6517
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4728425364106136e-06,
+      "loss": 2.1412,
+      "step": 6518
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.471336703847422e-06,
+      "loss": 2.2086,
+      "step": 6519
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.469831179394182e-06,
+      "loss": 2.2182,
+      "step": 6520
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4683259632343363e-06,
+      "loss": 2.1994,
+      "step": 6521
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4668210555512974e-06,
+      "loss": 2.2434,
+      "step": 6522
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.465316456528424e-06,
+      "loss": 2.1842,
+      "step": 6523
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4638121663490546e-06,
+      "loss": 2.1508,
+      "step": 6524
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.462308185196481e-06,
+      "loss": 2.1903,
+      "step": 6525
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4608045132539536e-06,
+      "loss": 2.1861,
+      "step": 6526
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4593011507046976e-06,
+      "loss": 2.1985,
+      "step": 6527
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4577980977318866e-06,
+      "loss": 2.2235,
+      "step": 6528
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4562953545186675e-06,
+      "loss": 2.1272,
+      "step": 6529
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4547929212481436e-06,
+      "loss": 2.1753,
+      "step": 6530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4532907981033822e-06,
+      "loss": 2.2356,
+      "step": 6531
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4517889852674114e-06,
+      "loss": 2.1454,
+      "step": 6532
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4502874829232238e-06,
+      "loss": 2.1847,
+      "step": 6533
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.448786291253772e-06,
+      "loss": 2.1803,
+      "step": 6534
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4472854104419717e-06,
+      "loss": 2.1676,
+      "step": 6535
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4457848406707014e-06,
+      "loss": 2.1233,
+      "step": 6536
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4442845821228005e-06,
+      "loss": 2.221,
+      "step": 6537
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.442784634981071e-06,
+      "loss": 2.2047,
+      "step": 6538
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4412849994282744e-06,
+      "loss": 2.1972,
+      "step": 6539
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4397856756471435e-06,
+      "loss": 2.186,
+      "step": 6540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4382866638203578e-06,
+      "loss": 2.2419,
+      "step": 6541
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4367879641305757e-06,
+      "loss": 2.1667,
+      "step": 6542
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4352895767604036e-06,
+      "loss": 2.1419,
+      "step": 6543
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4337915018924147e-06,
+      "loss": 2.2125,
+      "step": 6544
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.432293739709151e-06,
+      "loss": 2.245,
+      "step": 6545
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4307962903931025e-06,
+      "loss": 2.1182,
+      "step": 6546
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4292991541267368e-06,
+      "loss": 2.0804,
+      "step": 6547
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4278023310924676e-06,
+      "loss": 2.2411,
+      "step": 6548
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4263058214726844e-06,
+      "loss": 2.2398,
+      "step": 6549
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.424809625449729e-06,
+      "loss": 2.1011,
+      "step": 6550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.42331374320591e-06,
+      "loss": 2.1026,
+      "step": 6551
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4218181749234954e-06,
+      "loss": 2.1781,
+      "step": 6552
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4203229207847155e-06,
+      "loss": 2.1673,
+      "step": 6553
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.418827980971763e-06,
+      "loss": 2.175,
+      "step": 6554
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4173333556667912e-06,
+      "loss": 2.1209,
+      "step": 6555
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.415839045051916e-06,
+      "loss": 2.1646,
+      "step": 6556
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4143450493092146e-06,
+      "loss": 2.2033,
+      "step": 6557
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.412851368620726e-06,
+      "loss": 2.1766,
+      "step": 6558
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4113580031684487e-06,
+      "loss": 2.1542,
+      "step": 6559
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.40986495313435e-06,
+      "loss": 2.1996,
+      "step": 6560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4083722187003483e-06,
+      "loss": 2.1934,
+      "step": 6561
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4068798000483306e-06,
+      "loss": 2.1712,
+      "step": 6562
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.405387697360143e-06,
+      "loss": 2.1243,
+      "step": 6563
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.403895910817593e-06,
+      "loss": 2.1663,
+      "step": 6564
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.402404440602455e-06,
+      "loss": 2.2142,
+      "step": 6565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4009132868964525e-06,
+      "loss": 2.1863,
+      "step": 6566
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.399422449881284e-06,
+      "loss": 2.2258,
+      "step": 6567
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3979319297386035e-06,
+      "loss": 2.1307,
+      "step": 6568
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.396441726650021e-06,
+      "loss": 2.1609,
+      "step": 6569
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.39495184079712e-06,
+      "loss": 2.1839,
+      "step": 6570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.393462272361432e-06,
+      "loss": 2.2641,
+      "step": 6571
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.391973021524461e-06,
+      "loss": 2.1047,
+      "step": 6572
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3904840884676665e-06,
+      "loss": 2.2036,
+      "step": 6573
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3889954733724708e-06,
+      "loss": 2.1622,
+      "step": 6574
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.387507176420256e-06,
+      "loss": 2.139,
+      "step": 6575
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3860191977923673e-06,
+      "loss": 2.1797,
+      "step": 6576
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3845315376701112e-06,
+      "loss": 2.1761,
+      "step": 6577
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3830441962347528e-06,
+      "loss": 2.1981,
+      "step": 6578
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3815571736675214e-06,
+      "loss": 2.2479,
+      "step": 6579
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.380070470149605e-06,
+      "loss": 2.1981,
+      "step": 6580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3785840858621556e-06,
+      "loss": 2.2363,
+      "step": 6581
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3770980209862814e-06,
+      "loss": 2.1441,
+      "step": 6582
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3756122757030614e-06,
+      "loss": 2.2033,
+      "step": 6583
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3741268501935212e-06,
+      "loss": 2.1393,
+      "step": 6584
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.372641744638662e-06,
+      "loss": 2.1765,
+      "step": 6585
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3711569592194363e-06,
+      "loss": 2.1422,
+      "step": 6586
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3696724941167583e-06,
+      "loss": 2.1585,
+      "step": 6587
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3681883495115114e-06,
+      "loss": 2.1977,
+      "step": 6588
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3667045255845276e-06,
+      "loss": 2.2281,
+      "step": 6589
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3652210225166122e-06,
+      "loss": 2.1846,
+      "step": 6590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3637378404885224e-06,
+      "loss": 2.1957,
+      "step": 6591
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3622549796809807e-06,
+      "loss": 2.2607,
+      "step": 6592
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3607724402746685e-06,
+      "loss": 2.1728,
+      "step": 6593
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3592902224502284e-06,
+      "loss": 2.1825,
+      "step": 6594
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.357808326388265e-06,
+      "loss": 2.1687,
+      "step": 6595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.356326752269342e-06,
+      "loss": 2.1961,
+      "step": 6596
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.354845500273985e-06,
+      "loss": 2.1719,
+      "step": 6597
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.353364570582681e-06,
+      "loss": 2.2188,
+      "step": 6598
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.351883963375875e-06,
+      "loss": 2.2022,
+      "step": 6599
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3504036788339763e-06,
+      "loss": 2.1928,
+      "step": 6600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.348923717137352e-06,
+      "loss": 2.2093,
+      "step": 6601
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3474440784663287e-06,
+      "loss": 2.2439,
+      "step": 6602
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3459647630012026e-06,
+      "loss": 2.1887,
+      "step": 6603
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.344485770922218e-06,
+      "loss": 2.2164,
+      "step": 6604
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3430071024095853e-06,
+      "loss": 2.2582,
+      "step": 6605
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3415287576434807e-06,
+      "loss": 2.2303,
+      "step": 6606
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.34005073680403e-06,
+      "loss": 2.1994,
+      "step": 6607
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.338573040071332e-06,
+      "loss": 2.1066,
+      "step": 6608
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3370956676254327e-06,
+      "loss": 2.1389,
+      "step": 6609
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3356186196463497e-06,
+      "loss": 2.2404,
+      "step": 6610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.334141896314057e-06,
+      "loss": 2.1898,
+      "step": 6611
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3326654978084872e-06,
+      "loss": 2.2054,
+      "step": 6612
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3311894243095363e-06,
+      "loss": 2.1569,
+      "step": 6613
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.329713675997058e-06,
+      "loss": 2.1864,
+      "step": 6614
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3282382530508683e-06,
+      "loss": 2.1477,
+      "step": 6615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3267631556507443e-06,
+      "loss": 2.1848,
+      "step": 6616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.32528838397642e-06,
+      "loss": 2.1676,
+      "step": 6617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.323813938207593e-06,
+      "loss": 2.1532,
+      "step": 6618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.32233981852392e-06,
+      "loss": 2.1998,
+      "step": 6619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.320866025105016e-06,
+      "loss": 2.1816,
+      "step": 6620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.319392558130464e-06,
+      "loss": 2.1987,
+      "step": 6621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3179194177797954e-06,
+      "loss": 2.1674,
+      "step": 6622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3164466042325106e-06,
+      "loss": 2.2103,
+      "step": 6623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3149741176680666e-06,
+      "loss": 2.1356,
+      "step": 6624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3135019582658803e-06,
+      "loss": 2.1992,
+      "step": 6625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.312030126205335e-06,
+      "loss": 2.1746,
+      "step": 6626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3105586216657616e-06,
+      "loss": 2.184,
+      "step": 6627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.309087444826464e-06,
+      "loss": 2.0957,
+      "step": 6628
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3076165958666992e-06,
+      "loss": 2.1767,
+      "step": 6629
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3061460749656844e-06,
+      "loss": 2.1554,
+      "step": 6630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3046758823026018e-06,
+      "loss": 2.2014,
+      "step": 6631
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.303206018056583e-06,
+      "loss": 2.1562,
+      "step": 6632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.301736482406733e-06,
+      "loss": 2.1122,
+      "step": 6633
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3002672755321076e-06,
+      "loss": 2.1438,
+      "step": 6634
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.298798397611725e-06,
+      "loss": 2.1211,
+      "step": 6635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.297329848824565e-06,
+      "loss": 2.163,
+      "step": 6636
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.295861629349564e-06,
+      "loss": 2.2022,
+      "step": 6637
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.294393739365621e-06,
+      "loss": 2.1319,
+      "step": 6638
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2929261790515944e-06,
+      "loss": 2.1481,
+      "step": 6639
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2914589485863015e-06,
+      "loss": 2.2216,
+      "step": 6640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2899920481485192e-06,
+      "loss": 2.2009,
+      "step": 6641
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.288525477916986e-06,
+      "loss": 2.211,
+      "step": 6642
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.287059238070397e-06,
+      "loss": 2.2273,
+      "step": 6643
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.285593328787414e-06,
+      "loss": 2.1716,
+      "step": 6644
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.284127750246646e-06,
+      "loss": 2.2231,
+      "step": 6645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.282662502626678e-06,
+      "loss": 2.1564,
+      "step": 6646
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.281197586106037e-06,
+      "loss": 2.1622,
+      "step": 6647
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2797330008632255e-06,
+      "loss": 2.1643,
+      "step": 6648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2782687470766985e-06,
+      "loss": 2.2014,
+      "step": 6649
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2768048249248648e-06,
+      "loss": 2.2266,
+      "step": 6650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2753412345861065e-06,
+      "loss": 2.1314,
+      "step": 6651
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.27387797623875e-06,
+      "loss": 2.165,
+      "step": 6652
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2724150500610946e-06,
+      "loss": 2.2216,
+      "step": 6653
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2709524562313923e-06,
+      "loss": 2.2226,
+      "step": 6654
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2694901949278554e-06,
+      "loss": 2.1637,
+      "step": 6655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.268028266328655e-06,
+      "loss": 2.2078,
+      "step": 6656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2665666706119237e-06,
+      "loss": 2.1901,
+      "step": 6657
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.265105407955752e-06,
+      "loss": 2.1454,
+      "step": 6658
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.263644478538191e-06,
+      "loss": 2.2139,
+      "step": 6659
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2621838825372496e-06,
+      "loss": 2.1596,
+      "step": 6660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2607236201308974e-06,
+      "loss": 2.1992,
+      "step": 6661
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2592636914970633e-06,
+      "loss": 2.1745,
+      "step": 6662
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2578040968136326e-06,
+      "loss": 2.2243,
+      "step": 6663
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.256344836258459e-06,
+      "loss": 2.1654,
+      "step": 6664
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.254885910009341e-06,
+      "loss": 2.1572,
+      "step": 6665
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2534273182440515e-06,
+      "loss": 2.2263,
+      "step": 6666
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.25196906114031e-06,
+      "loss": 2.2293,
+      "step": 6667
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.250511138875801e-06,
+      "loss": 2.2527,
+      "step": 6668
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.249053551628173e-06,
+      "loss": 2.1635,
+      "step": 6669
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2475962995750224e-06,
+      "loss": 2.222,
+      "step": 6670
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.246139382893915e-06,
+      "loss": 2.2087,
+      "step": 6671
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.24468280176237e-06,
+      "loss": 2.1456,
+      "step": 6672
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2432265563578686e-06,
+      "loss": 2.2048,
+      "step": 6673
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2417706468578495e-06,
+      "loss": 2.2178,
+      "step": 6674
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2403150734397095e-06,
+      "loss": 2.2015,
+      "step": 6675
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2388598362808074e-06,
+      "loss": 2.1855,
+      "step": 6676
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2374049355584583e-06,
+      "loss": 2.0929,
+      "step": 6677
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.235950371449938e-06,
+      "loss": 2.1754,
+      "step": 6678
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2344961441324814e-06,
+      "loss": 2.1538,
+      "step": 6679
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.23304225378328e-06,
+      "loss": 2.2128,
+      "step": 6680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.231588700579488e-06,
+      "loss": 2.1841,
+      "step": 6681
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2301354846982148e-06,
+      "loss": 2.2526,
+      "step": 6682
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.228682606316529e-06,
+      "loss": 2.1446,
+      "step": 6683
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2272300656114648e-06,
+      "loss": 2.159,
+      "step": 6684
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2257778627600044e-06,
+      "loss": 2.1487,
+      "step": 6685
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.224325997939095e-06,
+      "loss": 2.2343,
+      "step": 6686
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.222874471325647e-06,
+      "loss": 2.2218,
+      "step": 6687
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.221423283096517e-06,
+      "loss": 2.1717,
+      "step": 6688
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.219972433428535e-06,
+      "loss": 2.1613,
+      "step": 6689
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.218521922498476e-06,
+      "loss": 2.1689,
+      "step": 6690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.217071750483085e-06,
+      "loss": 2.1315,
+      "step": 6691
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2156219175590623e-06,
+      "loss": 2.1784,
+      "step": 6692
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.214172423903058e-06,
+      "loss": 2.2024,
+      "step": 6693
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.212723269691697e-06,
+      "loss": 2.216,
+      "step": 6694
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2112744551015496e-06,
+      "loss": 2.1885,
+      "step": 6695
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.209825980309151e-06,
+      "loss": 2.1538,
+      "step": 6696
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.208377845490994e-06,
+      "loss": 2.1794,
+      "step": 6697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2069300508235273e-06,
+      "loss": 2.1162,
+      "step": 6698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2054825964831627e-06,
+      "loss": 2.1602,
+      "step": 6699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.204035482646267e-06,
+      "loss": 2.2483,
+      "step": 6700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2025887094891657e-06,
+      "loss": 2.2327,
+      "step": 6701
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.201142277188146e-06,
+      "loss": 2.21,
+      "step": 6702
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1996961859194487e-06,
+      "loss": 2.1743,
+      "step": 6703
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1982504358592777e-06,
+      "loss": 2.2107,
+      "step": 6704
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1968050271837926e-06,
+      "loss": 2.2074,
+      "step": 6705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.19535996006911e-06,
+      "loss": 2.1509,
+      "step": 6706
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.193915234691312e-06,
+      "loss": 2.2179,
+      "step": 6707
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.192470851226428e-06,
+      "loss": 2.1829,
+      "step": 6708
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1910268098504562e-06,
+      "loss": 2.1882,
+      "step": 6709
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1895831107393485e-06,
+      "loss": 2.1725,
+      "step": 6710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1881397540690106e-06,
+      "loss": 2.1735,
+      "step": 6711
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1866967400153184e-06,
+      "loss": 2.1324,
+      "step": 6712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.18525406875409e-06,
+      "loss": 2.2648,
+      "step": 6713
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.183811740461118e-06,
+      "loss": 2.1691,
+      "step": 6714
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1823697553121432e-06,
+      "loss": 2.143,
+      "step": 6715
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1809281134828663e-06,
+      "loss": 2.1919,
+      "step": 6716
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.179486815148948e-06,
+      "loss": 2.1032,
+      "step": 6717
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1780458604860056e-06,
+      "loss": 2.2827,
+      "step": 6718
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1766052496696155e-06,
+      "loss": 2.2073,
+      "step": 6719
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.175164982875311e-06,
+      "loss": 2.2811,
+      "step": 6720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.173725060278585e-06,
+      "loss": 2.1334,
+      "step": 6721
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1722854820548873e-06,
+      "loss": 2.2615,
+      "step": 6722
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1708462483796263e-06,
+      "loss": 2.1802,
+      "step": 6723
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1694073594281663e-06,
+      "loss": 2.1516,
+      "step": 6724
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1679688153758373e-06,
+      "loss": 2.1548,
+      "step": 6725
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1665306163979132e-06,
+      "loss": 2.2043,
+      "step": 6726
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.165092762669643e-06,
+      "loss": 2.1706,
+      "step": 6727
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1636552543662187e-06,
+      "loss": 2.2391,
+      "step": 6728
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1622180916627964e-06,
+      "loss": 2.21,
+      "step": 6729
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1607812747344955e-06,
+      "loss": 2.1841,
+      "step": 6730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1593448037563795e-06,
+      "loss": 2.1854,
+      "step": 6731
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.157908678903487e-06,
+      "loss": 2.2321,
+      "step": 6732
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1564729003507974e-06,
+      "loss": 2.1724,
+      "step": 6733
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1550374682732605e-06,
+      "loss": 2.2008,
+      "step": 6734
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1536023828457793e-06,
+      "loss": 2.3019,
+      "step": 6735
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.152167644243213e-06,
+      "loss": 2.1716,
+      "step": 6736
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1507332526403814e-06,
+      "loss": 2.2325,
+      "step": 6737
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.14929920821206e-06,
+      "loss": 2.1895,
+      "step": 6738
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.147865511132983e-06,
+      "loss": 2.1364,
+      "step": 6739
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.146432161577842e-06,
+      "loss": 2.1611,
+      "step": 6740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1449991597212865e-06,
+      "loss": 2.2016,
+      "step": 6741
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1435665057379233e-06,
+      "loss": 2.2315,
+      "step": 6742
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1421341998023167e-06,
+      "loss": 2.2254,
+      "step": 6743
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.140702242088987e-06,
+      "loss": 2.1133,
+      "step": 6744
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.13927063277242e-06,
+      "loss": 2.1336,
+      "step": 6745
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.137839372027047e-06,
+      "loss": 2.1553,
+      "step": 6746
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1364084600272645e-06,
+      "loss": 2.2133,
+      "step": 6747
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.134977896947425e-06,
+      "loss": 2.0956,
+      "step": 6748
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1335476829618364e-06,
+      "loss": 2.2032,
+      "step": 6749
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.132117818244771e-06,
+      "loss": 2.1815,
+      "step": 6750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1306883029704472e-06,
+      "loss": 2.258,
+      "step": 6751
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1292591373130515e-06,
+      "loss": 2.1356,
+      "step": 6752
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.127830321446722e-06,
+      "loss": 2.2047,
+      "step": 6753
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1264018555455563e-06,
+      "loss": 2.2279,
+      "step": 6754
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.124973739783609e-06,
+      "loss": 2.2302,
+      "step": 6755
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1235459743348874e-06,
+      "loss": 2.164,
+      "step": 6756
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.122118559373366e-06,
+      "loss": 2.1822,
+      "step": 6757
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1206914950729673e-06,
+      "loss": 2.1993,
+      "step": 6758
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.119264781607577e-06,
+      "loss": 2.1975,
+      "step": 6759
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1178384191510344e-06,
+      "loss": 2.1322,
+      "step": 6760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.116412407877138e-06,
+      "loss": 2.1846,
+      "step": 6761
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.114986747959643e-06,
+      "loss": 2.2112,
+      "step": 6762
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1135614395722613e-06,
+      "loss": 2.2475,
+      "step": 6763
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.112136482888663e-06,
+      "loss": 2.1532,
+      "step": 6764
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1107118780824744e-06,
+      "loss": 2.213,
+      "step": 6765
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1092876253272793e-06,
+      "loss": 2.0933,
+      "step": 6766
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1078637247966166e-06,
+      "loss": 2.2018,
+      "step": 6767
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.10644017666399e-06,
+      "loss": 2.2349,
+      "step": 6768
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.105016981102847e-06,
+      "loss": 2.1409,
+      "step": 6769
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.103594138286607e-06,
+      "loss": 2.2272,
+      "step": 6770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1021716483886323e-06,
+      "loss": 2.1605,
+      "step": 6771
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.100749511582254e-06,
+      "loss": 2.2251,
+      "step": 6772
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0993277280407547e-06,
+      "loss": 2.1768,
+      "step": 6773
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.09790629793737e-06,
+      "loss": 2.1554,
+      "step": 6774
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.096485221445301e-06,
+      "loss": 2.1426,
+      "step": 6775
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.095064498737701e-06,
+      "loss": 2.1025,
+      "step": 6776
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0936441299876803e-06,
+      "loss": 2.1718,
+      "step": 6777
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0922241153683064e-06,
+      "loss": 2.18,
+      "step": 6778
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0908044550526034e-06,
+      "loss": 2.1643,
+      "step": 6779
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0893851492135536e-06,
+      "loss": 2.1219,
+      "step": 6780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.087966198024094e-06,
+      "loss": 2.1351,
+      "step": 6781
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0865476016571206e-06,
+      "loss": 2.1863,
+      "step": 6782
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.085129360285484e-06,
+      "loss": 2.1711,
+      "step": 6783
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.083711474081993e-06,
+      "loss": 2.1741,
+      "step": 6784
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0822939432194134e-06,
+      "loss": 2.2105,
+      "step": 6785
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.080876767870466e-06,
+      "loss": 2.1534,
+      "step": 6786
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.079459948207828e-06,
+      "loss": 2.2224,
+      "step": 6787
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.07804348440414e-06,
+      "loss": 2.1853,
+      "step": 6788
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0766273766319873e-06,
+      "loss": 2.1647,
+      "step": 6789
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.075211625063923e-06,
+      "loss": 2.1286,
+      "step": 6790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0737962298724513e-06,
+      "loss": 2.2084,
+      "step": 6791
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0723811912300295e-06,
+      "loss": 2.2224,
+      "step": 6792
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0709665093090824e-06,
+      "loss": 2.1649,
+      "step": 6793
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0695521842819788e-06,
+      "loss": 2.1465,
+      "step": 6794
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0681382163210533e-06,
+      "loss": 2.2174,
+      "step": 6795
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.066724605598594e-06,
+      "loss": 2.1735,
+      "step": 6796
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0653113522868427e-06,
+      "loss": 2.157,
+      "step": 6797
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.063898456558002e-06,
+      "loss": 2.2198,
+      "step": 6798
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0624859185842284e-06,
+      "loss": 2.2353,
+      "step": 6799
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.061073738537635e-06,
+      "loss": 2.2181,
+      "step": 6800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0596619165902916e-06,
+      "loss": 2.1439,
+      "step": 6801
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0582504529142248e-06,
+      "loss": 2.1592,
+      "step": 6802
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.056839347681417e-06,
+      "loss": 2.1685,
+      "step": 6803
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0554286010638076e-06,
+      "loss": 2.192,
+      "step": 6804
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0540182132332886e-06,
+      "loss": 2.1603,
+      "step": 6805
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0526081843617183e-06,
+      "loss": 2.1843,
+      "step": 6806
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0511985146208966e-06,
+      "loss": 2.1015,
+      "step": 6807
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.049789204182596e-06,
+      "loss": 2.2294,
+      "step": 6808
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0483802532185286e-06,
+      "loss": 2.1619,
+      "step": 6809
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.046971661900373e-06,
+      "loss": 2.201,
+      "step": 6810
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.045563430399766e-06,
+      "loss": 2.1967,
+      "step": 6811
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.04415555888829e-06,
+      "loss": 2.1818,
+      "step": 6812
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.042748047537495e-06,
+      "loss": 2.2306,
+      "step": 6813
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.04134089651888e-06,
+      "loss": 2.1282,
+      "step": 6814
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0399341060039023e-06,
+      "loss": 2.1324,
+      "step": 6815
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0385276761639768e-06,
+      "loss": 2.116,
+      "step": 6816
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0371216071704667e-06,
+      "loss": 2.1177,
+      "step": 6817
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.035715899194704e-06,
+      "loss": 2.1442,
+      "step": 6818
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.034310552407967e-06,
+      "loss": 2.2243,
+      "step": 6819
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0329055669814936e-06,
+      "loss": 2.1814,
+      "step": 6820
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0315009430864762e-06,
+      "loss": 2.2013,
+      "step": 6821
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.030096680894065e-06,
+      "loss": 2.1167,
+      "step": 6822
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.028692780575364e-06,
+      "loss": 2.0995,
+      "step": 6823
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.027289242301435e-06,
+      "loss": 2.1899,
+      "step": 6824
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0258860662432946e-06,
+      "loss": 2.1619,
+      "step": 6825
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0244832525719155e-06,
+      "loss": 2.2406,
+      "step": 6826
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0230808014582265e-06,
+      "loss": 2.2295,
+      "step": 6827
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.02167871307311e-06,
+      "loss": 2.1813,
+      "step": 6828
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.020276987587412e-06,
+      "loss": 2.1881,
+      "step": 6829
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0188756251719204e-06,
+      "loss": 2.182,
+      "step": 6830
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0174746259973956e-06,
+      "loss": 2.1835,
+      "step": 6831
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.016073990234536e-06,
+      "loss": 2.1582,
+      "step": 6832
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.014673718054012e-06,
+      "loss": 2.2085,
+      "step": 6833
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0132738096264415e-06,
+      "loss": 2.1298,
+      "step": 6834
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0118742651223944e-06,
+      "loss": 2.2204,
+      "step": 6835
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0104750847124075e-06,
+      "loss": 2.1884,
+      "step": 6836
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0090762685669597e-06,
+      "loss": 2.1527,
+      "step": 6837
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.007677816856498e-06,
+      "loss": 2.1995,
+      "step": 6838
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0062797297514176e-06,
+      "loss": 2.1965,
+      "step": 6839
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0048820074220716e-06,
+      "loss": 2.1648,
+      "step": 6840
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0034846500387674e-06,
+      "loss": 2.2129,
+      "step": 6841
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.002087657771769e-06,
+      "loss": 2.1765,
+      "step": 6842
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0006910307912965e-06,
+      "loss": 2.2148,
+      "step": 6843
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.999294769267523e-06,
+      "loss": 2.1678,
+      "step": 6844
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9978988733705807e-06,
+      "loss": 2.1542,
+      "step": 6845
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.996503343270554e-06,
+      "loss": 2.1854,
+      "step": 6846
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9951081791374843e-06,
+      "loss": 2.1667,
+      "step": 6847
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9937133811413666e-06,
+      "loss": 2.2127,
+      "step": 6848
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9923189494521576e-06,
+      "loss": 2.2053,
+      "step": 6849
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.990924884239758e-06,
+      "loss": 2.1557,
+      "step": 6850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.989531185674037e-06,
+      "loss": 2.1637,
+      "step": 6851
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.988137853924808e-06,
+      "loss": 2.2082,
+      "step": 6852
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.986744889161844e-06,
+      "loss": 2.1759,
+      "step": 6853
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9853522915548777e-06,
+      "loss": 2.1484,
+      "step": 6854
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9839600612735877e-06,
+      "loss": 2.2106,
+      "step": 6855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9825681984876173e-06,
+      "loss": 2.1559,
+      "step": 6856
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9811767033665587e-06,
+      "loss": 2.155,
+      "step": 6857
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.979785576079961e-06,
+      "loss": 2.1387,
+      "step": 6858
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9783948167973306e-06,
+      "loss": 2.1798,
+      "step": 6859
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.977004425688126e-06,
+      "loss": 2.1478,
+      "step": 6860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9756144029217626e-06,
+      "loss": 2.1906,
+      "step": 6861
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.97422474866761e-06,
+      "loss": 2.1685,
+      "step": 6862
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9728354630949935e-06,
+      "loss": 2.1788,
+      "step": 6863
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9714465463731934e-06,
+      "loss": 2.2867,
+      "step": 6864
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.970057998671446e-06,
+      "loss": 2.0995,
+      "step": 6865
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9686698201589395e-06,
+      "loss": 2.1565,
+      "step": 6866
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9672820110048207e-06,
+      "loss": 2.2152,
+      "step": 6867
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9658945713781883e-06,
+      "loss": 2.1129,
+      "step": 6868
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9645075014481024e-06,
+      "loss": 2.158,
+      "step": 6869
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9631208013835677e-06,
+      "loss": 2.1671,
+      "step": 6870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9617344713535503e-06,
+      "loss": 2.193,
+      "step": 6871
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9603485115269743e-06,
+      "loss": 2.1199,
+      "step": 6872
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.958962922072709e-06,
+      "loss": 2.1837,
+      "step": 6873
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9575777031595906e-06,
+      "loss": 2.1505,
+      "step": 6874
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.956192854956397e-06,
+      "loss": 2.1619,
+      "step": 6875
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9548083776318727e-06,
+      "loss": 2.1607,
+      "step": 6876
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.95342427135471e-06,
+      "loss": 2.1571,
+      "step": 6877
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.95204053629356e-06,
+      "loss": 2.1818,
+      "step": 6878
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9506571726170236e-06,
+      "loss": 2.2059,
+      "step": 6879
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9492741804936623e-06,
+      "loss": 2.1491,
+      "step": 6880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9478915600919877e-06,
+      "loss": 2.1337,
+      "step": 6881
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.946509311580469e-06,
+      "loss": 2.1061,
+      "step": 6882
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.945127435127528e-06,
+      "loss": 2.1113,
+      "step": 6883
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9437459309015426e-06,
+      "loss": 2.1588,
+      "step": 6884
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.942364799070845e-06,
+      "loss": 2.1633,
+      "step": 6885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.94098403980372e-06,
+      "loss": 2.1717,
+      "step": 6886
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.939603653268414e-06,
+      "loss": 2.1486,
+      "step": 6887
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.938223639633119e-06,
+      "loss": 2.1509,
+      "step": 6888
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.936843999065985e-06,
+      "loss": 2.1823,
+      "step": 6889
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9354647317351187e-06,
+      "loss": 2.1854,
+      "step": 6890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9340858378085777e-06,
+      "loss": 2.2014,
+      "step": 6891
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.93270731745438e-06,
+      "loss": 2.1666,
+      "step": 6892
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9313291708404885e-06,
+      "loss": 2.2239,
+      "step": 6893
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.929951398134832e-06,
+      "loss": 2.1966,
+      "step": 6894
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.928573999505284e-06,
+      "loss": 2.1558,
+      "step": 6895
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.927196975119678e-06,
+      "loss": 2.2295,
+      "step": 6896
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9258203251458012e-06,
+      "loss": 2.2117,
+      "step": 6897
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9244440497513895e-06,
+      "loss": 2.1303,
+      "step": 6898
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9230681491041425e-06,
+      "loss": 2.1116,
+      "step": 6899
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9216926233717087e-06,
+      "loss": 2.1762,
+      "step": 6900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.920317472721691e-06,
+      "loss": 2.2302,
+      "step": 6901
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9189426973216478e-06,
+      "loss": 2.1369,
+      "step": 6902
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.917568297339091e-06,
+      "loss": 2.1713,
+      "step": 6903
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9161942729414876e-06,
+      "loss": 2.1566,
+      "step": 6904
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9148206242962575e-06,
+      "loss": 2.1452,
+      "step": 6905
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.913447351570776e-06,
+      "loss": 2.1771,
+      "step": 6906
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.912074454932372e-06,
+      "loss": 2.1535,
+      "step": 6907
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.910701934548329e-06,
+      "loss": 2.2189,
+      "step": 6908
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9093297905858833e-06,
+      "loss": 2.1861,
+      "step": 6909
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.90795802321223e-06,
+      "loss": 2.145,
+      "step": 6910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9065866325945099e-06,
+      "loss": 2.212,
+      "step": 6911
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9052156188998284e-06,
+      "loss": 2.2446,
+      "step": 6912
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9038449822952331e-06,
+      "loss": 2.2341,
+      "step": 6913
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9024747229477365e-06,
+      "loss": 2.1802,
+      "step": 6914
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9011048410243011e-06,
+      "loss": 2.1869,
+      "step": 6915
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8997353366918369e-06,
+      "loss": 2.1973,
+      "step": 6916
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8983662101172217e-06,
+      "loss": 2.1229,
+      "step": 6917
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.896997461467272e-06,
+      "loss": 2.1593,
+      "step": 6918
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.895629090908771e-06,
+      "loss": 2.2,
+      "step": 6919
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8942610986084487e-06,
+      "loss": 2.1828,
+      "step": 6920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8928934847329905e-06,
+      "loss": 2.1006,
+      "step": 6921
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8915262494490366e-06,
+      "loss": 2.1932,
+      "step": 6922
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8901593929231804e-06,
+      "loss": 2.1265,
+      "step": 6923
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8887929153219687e-06,
+      "loss": 2.2696,
+      "step": 6924
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.887426816811903e-06,
+      "loss": 2.2099,
+      "step": 6925
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8860610975594384e-06,
+      "loss": 2.1976,
+      "step": 6926
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8846957577309832e-06,
+      "loss": 2.17,
+      "step": 6927
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8833307974929006e-06,
+      "loss": 2.1762,
+      "step": 6928
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8819662170115043e-06,
+      "loss": 2.1779,
+      "step": 6929
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8806020164530702e-06,
+      "loss": 2.2132,
+      "step": 6930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8792381959838147e-06,
+      "loss": 2.1371,
+      "step": 6931
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8778747557699223e-06,
+      "loss": 2.2349,
+      "step": 6932
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8765116959775187e-06,
+      "loss": 2.2691,
+      "step": 6933
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8751490167726888e-06,
+      "loss": 2.1501,
+      "step": 6934
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.873786718321476e-06,
+      "loss": 2.2038,
+      "step": 6935
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8724248007898648e-06,
+      "loss": 2.2077,
+      "step": 6936
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.871063264343807e-06,
+      "loss": 2.1172,
+      "step": 6937
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8697021091491991e-06,
+      "loss": 2.1432,
+      "step": 6938
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8683413353718937e-06,
+      "loss": 2.2534,
+      "step": 6939
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8669809431776991e-06,
+      "loss": 2.1989,
+      "step": 6940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8656209327323704e-06,
+      "loss": 2.0923,
+      "step": 6941
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8642613042016245e-06,
+      "loss": 2.1967,
+      "step": 6942
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.862902057751127e-06,
+      "loss": 2.2061,
+      "step": 6943
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8615431935464984e-06,
+      "loss": 2.1884,
+      "step": 6944
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8601847117533112e-06,
+      "loss": 2.2034,
+      "step": 6945
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8588266125370929e-06,
+      "loss": 2.2067,
+      "step": 6946
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8574688960633236e-06,
+      "loss": 2.1651,
+      "step": 6947
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8561115624974374e-06,
+      "loss": 2.1186,
+      "step": 6948
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8547546120048204e-06,
+      "loss": 2.1882,
+      "step": 6949
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8533980447508138e-06,
+      "loss": 2.2159,
+      "step": 6950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8520418609007107e-06,
+      "loss": 2.1625,
+      "step": 6951
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8506860606197564e-06,
+      "loss": 2.1205,
+      "step": 6952
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8493306440731557e-06,
+      "loss": 2.1781,
+      "step": 6953
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8479756114260562e-06,
+      "loss": 2.1437,
+      "step": 6954
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8466209628435705e-06,
+      "loss": 2.1165,
+      "step": 6955
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8452666984907519e-06,
+      "loss": 2.167,
+      "step": 6956
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8439128185326183e-06,
+      "loss": 2.1852,
+      "step": 6957
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.842559323134136e-06,
+      "loss": 2.2117,
+      "step": 6958
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8412062124602192e-06,
+      "loss": 2.2417,
+      "step": 6959
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8398534866757455e-06,
+      "loss": 2.1339,
+      "step": 6960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8385011459455394e-06,
+      "loss": 2.1317,
+      "step": 6961
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.837149190434378e-06,
+      "loss": 2.1978,
+      "step": 6962
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8357976203069943e-06,
+      "loss": 2.1815,
+      "step": 6963
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8344464357280722e-06,
+      "loss": 2.1663,
+      "step": 6964
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8330956368622498e-06,
+      "loss": 2.2358,
+      "step": 6965
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.831745223874118e-06,
+      "loss": 2.1706,
+      "step": 6966
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8303951969282202e-06,
+      "loss": 2.1523,
+      "step": 6967
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.829045556189053e-06,
+      "loss": 2.2313,
+      "step": 6968
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8276963018210664e-06,
+      "loss": 2.1205,
+      "step": 6969
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8263474339886628e-06,
+      "loss": 2.219,
+      "step": 6970
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.824998952856198e-06,
+      "loss": 2.1895,
+      "step": 6971
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8236508585879781e-06,
+      "loss": 2.1514,
+      "step": 6972
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8223031513482692e-06,
+      "loss": 2.2671,
+      "step": 6973
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8209558313012792e-06,
+      "loss": 2.1798,
+      "step": 6974
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8196088986111798e-06,
+      "loss": 2.0969,
+      "step": 6975
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8182623534420906e-06,
+      "loss": 2.1808,
+      "step": 6976
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8169161959580795e-06,
+      "loss": 2.2176,
+      "step": 6977
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8155704263231777e-06,
+      "loss": 2.1917,
+      "step": 6978
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8142250447013566e-06,
+      "loss": 2.1734,
+      "step": 6979
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8128800512565514e-06,
+      "loss": 2.2229,
+      "step": 6980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8115354461526453e-06,
+      "loss": 2.1526,
+      "step": 6981
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.810191229553473e-06,
+      "loss": 2.1921,
+      "step": 6982
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8088474016228236e-06,
+      "loss": 2.1856,
+      "step": 6983
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.807503962524439e-06,
+      "loss": 2.1336,
+      "step": 6984
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.806160912422012e-06,
+      "loss": 2.1896,
+      "step": 6985
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8048182514791901e-06,
+      "loss": 2.1149,
+      "step": 6986
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8034759798595724e-06,
+      "loss": 2.1293,
+      "step": 6987
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8021340977267104e-06,
+      "loss": 2.1254,
+      "step": 6988
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.800792605244109e-06,
+      "loss": 2.0957,
+      "step": 6989
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.799451502575222e-06,
+      "loss": 2.1927,
+      "step": 6990
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.798110789883465e-06,
+      "loss": 2.1435,
+      "step": 6991
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7967704673321917e-06,
+      "loss": 2.2195,
+      "step": 6992
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7954305350847246e-06,
+      "loss": 2.1796,
+      "step": 6993
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7940909933043243e-06,
+      "loss": 2.1687,
+      "step": 6994
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7927518421542106e-06,
+      "loss": 2.2388,
+      "step": 6995
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7914130817975595e-06,
+      "loss": 2.1285,
+      "step": 6996
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7900747123974882e-06,
+      "loss": 2.1862,
+      "step": 6997
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7887367341170781e-06,
+      "loss": 2.1713,
+      "step": 6998
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7873991471193563e-06,
+      "loss": 2.2023,
+      "step": 6999
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7860619515673034e-06,
+      "loss": 2.2182,
+      "step": 7000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.784725147623853e-06,
+      "loss": 2.1263,
+      "step": 7001
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7833887354518902e-06,
+      "loss": 2.1355,
+      "step": 7002
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7820527152142531e-06,
+      "loss": 2.2486,
+      "step": 7003
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7807170870737317e-06,
+      "loss": 2.1449,
+      "step": 7004
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7793818511930678e-06,
+      "loss": 2.1193,
+      "step": 7005
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7780470077349566e-06,
+      "loss": 2.2114,
+      "step": 7006
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7767125568620442e-06,
+      "loss": 2.2176,
+      "step": 7007
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7753784987369287e-06,
+      "loss": 2.193,
+      "step": 7008
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7740448335221628e-06,
+      "loss": 2.0985,
+      "step": 7009
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7727115613802465e-06,
+      "loss": 2.1694,
+      "step": 7010
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7713786824736406e-06,
+      "loss": 2.1548,
+      "step": 7011
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.770046196964747e-06,
+      "loss": 2.1859,
+      "step": 7012
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7687141050159246e-06,
+      "loss": 2.232,
+      "step": 7013
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7673824067894912e-06,
+      "loss": 2.1797,
+      "step": 7014
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7660511024477018e-06,
+      "loss": 2.2181,
+      "step": 7015
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7647201921527802e-06,
+      "loss": 2.184,
+      "step": 7016
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763389676066885e-06,
+      "loss": 2.1927,
+      "step": 7017
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.762059554352143e-06,
+      "loss": 2.1386,
+      "step": 7018
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760729827170622e-06,
+      "loss": 2.1725,
+      "step": 7019
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7594004946843458e-06,
+      "loss": 2.1908,
+      "step": 7020
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.758071557055291e-06,
+      "loss": 2.2024,
+      "step": 7021
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7567430144453801e-06,
+      "loss": 2.1491,
+      "step": 7022
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7554148670164966e-06,
+      "loss": 2.1814,
+      "step": 7023
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.75408711493047e-06,
+      "loss": 2.1944,
+      "step": 7024
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7527597583490825e-06,
+      "loss": 2.1396,
+      "step": 7025
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.751432797434068e-06,
+      "loss": 2.2107,
+      "step": 7026
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7501062323471136e-06,
+      "loss": 2.1596,
+      "step": 7027
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7487800632498547e-06,
+      "loss": 2.2259,
+      "step": 7028
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.747454290303887e-06,
+      "loss": 2.2183,
+      "step": 7029
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.746128913670746e-06,
+      "loss": 2.202,
+      "step": 7030
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7448039335119272e-06,
+      "loss": 2.1245,
+      "step": 7031
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7434793499888746e-06,
+      "loss": 2.1869,
+      "step": 7032
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7421551632629835e-06,
+      "loss": 2.1222,
+      "step": 7033
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7408313734956074e-06,
+      "loss": 2.1189,
+      "step": 7034
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7395079808480386e-06,
+      "loss": 2.1096,
+      "step": 7035
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.738184985481536e-06,
+      "loss": 2.1701,
+      "step": 7036
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7368623875572948e-06,
+      "loss": 2.1891,
+      "step": 7037
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7355401872364759e-06,
+      "loss": 2.2303,
+      "step": 7038
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.734218384680184e-06,
+      "loss": 2.1508,
+      "step": 7039
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7328969800494727e-06,
+      "loss": 2.1723,
+      "step": 7040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7315759735053562e-06,
+      "loss": 2.1584,
+      "step": 7041
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7302553652087927e-06,
+      "loss": 2.1555,
+      "step": 7042
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7289351553206952e-06,
+      "loss": 2.185,
+      "step": 7043
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.727615344001926e-06,
+      "loss": 2.2087,
+      "step": 7044
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7262959314133015e-06,
+      "loss": 2.1998,
+      "step": 7045
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7249769177155879e-06,
+      "loss": 2.1759,
+      "step": 7046
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.723658303069502e-06,
+      "loss": 2.2685,
+      "step": 7047
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7223400876357144e-06,
+      "loss": 2.1789,
+      "step": 7048
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7210222715748443e-06,
+      "loss": 2.1333,
+      "step": 7049
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7197048550474643e-06,
+      "loss": 2.1825,
+      "step": 7050
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7183878382140978e-06,
+      "loss": 2.1768,
+      "step": 7051
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7170712212352187e-06,
+      "loss": 2.1897,
+      "step": 7052
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7157550042712517e-06,
+      "loss": 2.1896,
+      "step": 7053
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7144391874825784e-06,
+      "loss": 2.2199,
+      "step": 7054
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7131237710295207e-06,
+      "loss": 2.1741,
+      "step": 7055
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7118087550723633e-06,
+      "loss": 2.1373,
+      "step": 7056
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.710494139771336e-06,
+      "loss": 2.174,
+      "step": 7057
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.709179925286617e-06,
+      "loss": 2.1607,
+      "step": 7058
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7078661117783447e-06,
+      "loss": 2.1803,
+      "step": 7059
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7065526994065973e-06,
+      "loss": 2.1235,
+      "step": 7060
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7052396883314154e-06,
+      "loss": 2.1326,
+      "step": 7061
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7039270787127832e-06,
+      "loss": 2.1801,
+      "step": 7062
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7026148707106388e-06,
+      "loss": 2.1601,
+      "step": 7063
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7013030644848698e-06,
+      "loss": 2.2311,
+      "step": 7064
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6999916601953165e-06,
+      "loss": 2.1904,
+      "step": 7065
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6986806580017695e-06,
+      "loss": 2.1579,
+      "step": 7066
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6973700580639707e-06,
+      "loss": 2.2088,
+      "step": 7067
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6960598605416117e-06,
+      "loss": 2.2073,
+      "step": 7068
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6947500655943373e-06,
+      "loss": 2.173,
+      "step": 7069
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6934406733817417e-06,
+      "loss": 2.1824,
+      "step": 7070
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6921316840633678e-06,
+      "loss": 2.2236,
+      "step": 7071
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6908230977987184e-06,
+      "loss": 2.1943,
+      "step": 7072
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6895149147472344e-06,
+      "loss": 2.1406,
+      "step": 7073
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6882071350683165e-06,
+      "loss": 2.1591,
+      "step": 7074
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6868997589213138e-06,
+      "loss": 2.1343,
+      "step": 7075
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6855927864655241e-06,
+      "loss": 2.1962,
+      "step": 7076
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6842862178602026e-06,
+      "loss": 2.2173,
+      "step": 7077
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6829800532645447e-06,
+      "loss": 2.1664,
+      "step": 7078
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6816742928377072e-06,
+      "loss": 2.1787,
+      "step": 7079
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.680368936738792e-06,
+      "loss": 2.2563,
+      "step": 7080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.679063985126852e-06,
+      "loss": 2.2037,
+      "step": 7081
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6777594381608936e-06,
+      "loss": 2.2396,
+      "step": 7082
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.676455295999867e-06,
+      "loss": 2.191,
+      "step": 7083
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6751515588026828e-06,
+      "loss": 2.195,
+      "step": 7084
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6738482267281963e-06,
+      "loss": 2.167,
+      "step": 7085
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6725452999352137e-06,
+      "loss": 2.248,
+      "step": 7086
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.671242778582493e-06,
+      "loss": 2.1733,
+      "step": 7087
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6699406628287423e-06,
+      "loss": 2.1753,
+      "step": 7088
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6686389528326214e-06,
+      "loss": 2.1569,
+      "step": 7089
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6673376487527382e-06,
+      "loss": 2.2356,
+      "step": 7090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6660367507476539e-06,
+      "loss": 2.2009,
+      "step": 7091
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6647362589758787e-06,
+      "loss": 2.1923,
+      "step": 7092
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6634361735958731e-06,
+      "loss": 2.1967,
+      "step": 7093
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6621364947660472e-06,
+      "loss": 2.2305,
+      "step": 7094
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6608372226447678e-06,
+      "loss": 2.1717,
+      "step": 7095
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6595383573903412e-06,
+      "loss": 2.1535,
+      "step": 7096
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.658239899161036e-06,
+      "loss": 2.2253,
+      "step": 7097
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6569418481150596e-06,
+      "loss": 2.2078,
+      "step": 7098
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6556442044105797e-06,
+      "loss": 2.1974,
+      "step": 7099
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6543469682057105e-06,
+      "loss": 2.1922,
+      "step": 7100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.653050139658512e-06,
+      "loss": 2.1782,
+      "step": 7101
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6517537189270043e-06,
+      "loss": 2.1806,
+      "step": 7102
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6504577061691468e-06,
+      "loss": 2.2304,
+      "step": 7103
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6491621015428588e-06,
+      "loss": 2.0982,
+      "step": 7104
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6478669052060048e-06,
+      "loss": 2.1489,
+      "step": 7105
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6465721173164e-06,
+      "loss": 2.1603,
+      "step": 7106
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6452777380318112e-06,
+      "loss": 2.0732,
+      "step": 7107
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.643983767509954e-06,
+      "loss": 2.1924,
+      "step": 7108
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6426902059084942e-06,
+      "loss": 2.1501,
+      "step": 7109
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6413970533850498e-06,
+      "loss": 2.1772,
+      "step": 7110
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6401043100971864e-06,
+      "loss": 2.137,
+      "step": 7111
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6388119762024213e-06,
+      "loss": 2.183,
+      "step": 7112
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6375200518582208e-06,
+      "loss": 2.2374,
+      "step": 7113
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6362285372220016e-06,
+      "loss": 2.2293,
+      "step": 7114
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6349374324511347e-06,
+      "loss": 2.1476,
+      "step": 7115
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6336467377029308e-06,
+      "loss": 2.2352,
+      "step": 7116
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6323564531346642e-06,
+      "loss": 2.2488,
+      "step": 7117
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6310665789035468e-06,
+      "loss": 2.16,
+      "step": 7118
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6297771151667462e-06,
+      "loss": 2.1017,
+      "step": 7119
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6284880620813847e-06,
+      "loss": 2.1945,
+      "step": 7120
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.627199419804522e-06,
+      "loss": 2.1778,
+      "step": 7121
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6259111884931817e-06,
+      "loss": 2.1186,
+      "step": 7122
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6246233683043279e-06,
+      "loss": 2.1912,
+      "step": 7123
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6233359593948777e-06,
+      "loss": 2.2376,
+      "step": 7124
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6220489619216988e-06,
+      "loss": 2.2296,
+      "step": 7125
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6207623760416074e-06,
+      "loss": 2.1871,
+      "step": 7126
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6194762019113703e-06,
+      "loss": 2.1959,
+      "step": 7127
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6181904396877041e-06,
+      "loss": 2.171,
+      "step": 7128
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6169050895272743e-06,
+      "loss": 2.1586,
+      "step": 7129
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6156201515866971e-06,
+      "loss": 2.1863,
+      "step": 7130
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6143356260225385e-06,
+      "loss": 2.1868,
+      "step": 7131
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6130515129913144e-06,
+      "loss": 2.1353,
+      "step": 7132
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6117678126494895e-06,
+      "loss": 2.1327,
+      "step": 7133
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6104845251534772e-06,
+      "loss": 2.171,
+      "step": 7134
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6092016506596481e-06,
+      "loss": 2.215,
+      "step": 7135
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6079191893243102e-06,
+      "loss": 2.1318,
+      "step": 7136
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6066371413037286e-06,
+      "loss": 2.221,
+      "step": 7137
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.605355506754121e-06,
+      "loss": 2.234,
+      "step": 7138
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6040742858316443e-06,
+      "loss": 2.1958,
+      "step": 7139
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6027934786924187e-06,
+      "loss": 2.2225,
+      "step": 7140
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6015130854924999e-06,
+      "loss": 2.1687,
+      "step": 7141
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.600233106387904e-06,
+      "loss": 2.1742,
+      "step": 7142
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.598953541534592e-06,
+      "loss": 2.1887,
+      "step": 7143
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.597674391088474e-06,
+      "loss": 2.1376,
+      "step": 7144
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.596395655205411e-06,
+      "loss": 2.1745,
+      "step": 7145
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5951173340412134e-06,
+      "loss": 2.1952,
+      "step": 7146
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5938394277516412e-06,
+      "loss": 2.1385,
+      "step": 7147
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5925619364924016e-06,
+      "loss": 2.1815,
+      "step": 7148
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.591284860419155e-06,
+      "loss": 2.2167,
+      "step": 7149
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5900081996875083e-06,
+      "loss": 2.1795,
+      "step": 7150
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5887319544530182e-06,
+      "loss": 2.1785,
+      "step": 7151
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.587456124871191e-06,
+      "loss": 2.1124,
+      "step": 7152
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5861807110974869e-06,
+      "loss": 2.2034,
+      "step": 7153
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5849057132873063e-06,
+      "loss": 2.247,
+      "step": 7154
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5836311315960051e-06,
+      "loss": 2.18,
+      "step": 7155
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.582356966178888e-06,
+      "loss": 2.023,
+      "step": 7156
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.581083217191206e-06,
+      "loss": 2.1375,
+      "step": 7157
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5798098847881664e-06,
+      "loss": 2.163,
+      "step": 7158
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5785369691249147e-06,
+      "loss": 2.1542,
+      "step": 7159
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5772644703565564e-06,
+      "loss": 2.1724,
+      "step": 7160
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5759923886381402e-06,
+      "loss": 2.2119,
+      "step": 7161
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5747207241246654e-06,
+      "loss": 2.1802,
+      "step": 7162
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5734494769710817e-06,
+      "loss": 2.2562,
+      "step": 7163
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5721786473322825e-06,
+      "loss": 2.1697,
+      "step": 7164
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5709082353631188e-06,
+      "loss": 2.2004,
+      "step": 7165
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5696382412183853e-06,
+      "loss": 2.2281,
+      "step": 7166
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5683686650528267e-06,
+      "loss": 2.2312,
+      "step": 7167
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.567099507021137e-06,
+      "loss": 2.2169,
+      "step": 7168
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5658307672779594e-06,
+      "loss": 2.1142,
+      "step": 7169
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5645624459778858e-06,
+      "loss": 2.2031,
+      "step": 7170
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.563294543275457e-06,
+      "loss": 2.1566,
+      "step": 7171
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5620270593251635e-06,
+      "loss": 2.1353,
+      "step": 7172
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.560759994281445e-06,
+      "loss": 2.1898,
+      "step": 7173
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5594933482986885e-06,
+      "loss": 2.194,
+      "step": 7174
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5582271215312294e-06,
+      "loss": 2.1625,
+      "step": 7175
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.556961314133359e-06,
+      "loss": 2.1554,
+      "step": 7176
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5556959262593058e-06,
+      "loss": 2.1576,
+      "step": 7177
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.554430958063259e-06,
+      "loss": 2.237,
+      "step": 7178
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5531664096993454e-06,
+      "loss": 2.1364,
+      "step": 7179
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.551902281321651e-06,
+      "loss": 2.195,
+      "step": 7180
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5506385730842062e-06,
+      "loss": 2.1899,
+      "step": 7181
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5493752851409844e-06,
+      "loss": 2.2217,
+      "step": 7182
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5481124176459195e-06,
+      "loss": 2.2144,
+      "step": 7183
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5468499707528856e-06,
+      "loss": 2.1951,
+      "step": 7184
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5455879446157084e-06,
+      "loss": 2.1621,
+      "step": 7185
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5443263393881619e-06,
+      "loss": 2.1398,
+      "step": 7186
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5430651552239684e-06,
+      "loss": 2.1785,
+      "step": 7187
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5418043922768e-06,
+      "loss": 2.1155,
+      "step": 7188
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.540544050700276e-06,
+      "loss": 2.1482,
+      "step": 7189
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5392841306479667e-06,
+      "loss": 2.1831,
+      "step": 7190
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5380246322733883e-06,
+      "loss": 2.1424,
+      "step": 7191
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5367655557300066e-06,
+      "loss": 2.1725,
+      "step": 7192
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5355069011712376e-06,
+      "loss": 2.1793,
+      "step": 7193
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5342486687504432e-06,
+      "loss": 2.2354,
+      "step": 7194
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5329908586209347e-06,
+      "loss": 2.2146,
+      "step": 7195
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.531733470935976e-06,
+      "loss": 2.1244,
+      "step": 7196
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5304765058487725e-06,
+      "loss": 2.2198,
+      "step": 7197
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.529219963512481e-06,
+      "loss": 2.2132,
+      "step": 7198
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5279638440802118e-06,
+      "loss": 2.1388,
+      "step": 7199
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5267081477050132e-06,
+      "loss": 2.2464,
+      "step": 7200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5254528745398943e-06,
+      "loss": 2.1549,
+      "step": 7201
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5241980247378008e-06,
+      "loss": 2.1957,
+      "step": 7202
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5229435984516355e-06,
+      "loss": 2.1397,
+      "step": 7203
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.521689595834246e-06,
+      "loss": 2.2184,
+      "step": 7204
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5204360170384286e-06,
+      "loss": 2.0897,
+      "step": 7205
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.519182862216929e-06,
+      "loss": 2.225,
+      "step": 7206
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5179301315224364e-06,
+      "loss": 2.1676,
+      "step": 7207
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5166778251075964e-06,
+      "loss": 2.1746,
+      "step": 7208
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5154259431249978e-06,
+      "loss": 2.2082,
+      "step": 7209
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.514174485727178e-06,
+      "loss": 2.1855,
+      "step": 7210
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5129234530666232e-06,
+      "loss": 2.1403,
+      "step": 7211
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5116728452957686e-06,
+      "loss": 2.1183,
+      "step": 7212
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5104226625669943e-06,
+      "loss": 2.1646,
+      "step": 7213
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5091729050326376e-06,
+      "loss": 2.1353,
+      "step": 7214
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5079235728449714e-06,
+      "loss": 2.1503,
+      "step": 7215
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5066746661562254e-06,
+      "loss": 2.2228,
+      "step": 7216
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5054261851185753e-06,
+      "loss": 2.161,
+      "step": 7217
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5041781298841424e-06,
+      "loss": 2.1969,
+      "step": 7218
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5029305006050038e-06,
+      "loss": 2.1498,
+      "step": 7219
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5016832974331725e-06,
+      "loss": 2.1234,
+      "step": 7220
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5004365205206235e-06,
+      "loss": 2.1667,
+      "step": 7221
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4991901700192657e-06,
+      "loss": 2.2169,
+      "step": 7222
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4979442460809684e-06,
+      "loss": 2.1965,
+      "step": 7223
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.496698748857543e-06,
+      "loss": 2.1938,
+      "step": 7224
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4954536785007456e-06,
+      "loss": 2.2029,
+      "step": 7225
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4942090351622884e-06,
+      "loss": 2.1938,
+      "step": 7226
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.492964818993826e-06,
+      "loss": 2.105,
+      "step": 7227
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.491721030146963e-06,
+      "loss": 2.1592,
+      "step": 7228
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4904776687732503e-06,
+      "loss": 2.1356,
+      "step": 7229
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.489234735024188e-06,
+      "loss": 2.1798,
+      "step": 7230
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4879922290512244e-06,
+      "loss": 2.1938,
+      "step": 7231
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4867501510057548e-06,
+      "loss": 2.2221,
+      "step": 7232
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4855085010391217e-06,
+      "loss": 2.1302,
+      "step": 7233
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.484267279302618e-06,
+      "loss": 2.1643,
+      "step": 7234
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4830264859474814e-06,
+      "loss": 2.1674,
+      "step": 7235
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4817861211248996e-06,
+      "loss": 2.1837,
+      "step": 7236
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.480546184986007e-06,
+      "loss": 2.2519,
+      "step": 7237
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4793066776818843e-06,
+      "loss": 2.2352,
+      "step": 7238
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4780675993635668e-06,
+      "loss": 2.0945,
+      "step": 7239
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4768289501820265e-06,
+      "loss": 2.1109,
+      "step": 7240
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4755907302881927e-06,
+      "loss": 2.1271,
+      "step": 7241
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4743529398329393e-06,
+      "loss": 2.1672,
+      "step": 7242
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.473115578967083e-06,
+      "loss": 2.1461,
+      "step": 7243
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4718786478413983e-06,
+      "loss": 2.1838,
+      "step": 7244
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4706421466065952e-06,
+      "loss": 2.1984,
+      "step": 7245
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.469406075413342e-06,
+      "loss": 2.17,
+      "step": 7246
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.46817043441225e-06,
+      "loss": 2.1807,
+      "step": 7247
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4669352237538763e-06,
+      "loss": 2.1758,
+      "step": 7248
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4657004435887296e-06,
+      "loss": 2.1565,
+      "step": 7249
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4644660940672628e-06,
+      "loss": 2.1585,
+      "step": 7250
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.463232175339878e-06,
+      "loss": 2.1815,
+      "step": 7251
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4619986875569247e-06,
+      "loss": 2.1631,
+      "step": 7252
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.460765630868699e-06,
+      "loss": 2.159,
+      "step": 7253
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.459533005425446e-06,
+      "loss": 2.108,
+      "step": 7254
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4583008113773567e-06,
+      "loss": 2.1995,
+      "step": 7255
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4570690488745687e-06,
+      "loss": 2.1547,
+      "step": 7256
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4558377180671734e-06,
+      "loss": 2.1366,
+      "step": 7257
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4546068191051988e-06,
+      "loss": 2.1025,
+      "step": 7258
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4533763521386319e-06,
+      "loss": 2.1896,
+      "step": 7259
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4521463173173966e-06,
+      "loss": 2.1461,
+      "step": 7260
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4509167147913693e-06,
+      "loss": 2.1624,
+      "step": 7261
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4496875447103781e-06,
+      "loss": 2.1553,
+      "step": 7262
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4484588072241873e-06,
+      "loss": 2.2342,
+      "step": 7263
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4472305024825189e-06,
+      "loss": 2.1766,
+      "step": 7264
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4460026306350378e-06,
+      "loss": 2.1684,
+      "step": 7265
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4447751918313552e-06,
+      "loss": 2.2005,
+      "step": 7266
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4435481862210315e-06,
+      "loss": 2.1616,
+      "step": 7267
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4423216139535735e-06,
+      "loss": 2.2475,
+      "step": 7268
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4410954751784352e-06,
+      "loss": 2.1442,
+      "step": 7269
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4398697700450181e-06,
+      "loss": 2.2195,
+      "step": 7270
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4386444987026705e-06,
+      "loss": 2.1547,
+      "step": 7271
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4374196613006874e-06,
+      "loss": 2.1796,
+      "step": 7272
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4361952579883127e-06,
+      "loss": 2.1924,
+      "step": 7273
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4349712889147355e-06,
+      "loss": 2.1961,
+      "step": 7274
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.433747754229093e-06,
+      "loss": 2.2271,
+      "step": 7275
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4325246540804672e-06,
+      "loss": 2.1849,
+      "step": 7276
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4313019886178942e-06,
+      "loss": 2.2087,
+      "step": 7277
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4300797579903476e-06,
+      "loss": 2.1909,
+      "step": 7278
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.428857962346752e-06,
+      "loss": 2.1774,
+      "step": 7279
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4276366018359845e-06,
+      "loss": 2.182,
+      "step": 7280
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4264156766068577e-06,
+      "loss": 2.1565,
+      "step": 7281
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4251951868081438e-06,
+      "loss": 2.1237,
+      "step": 7282
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4239751325885499e-06,
+      "loss": 2.1885,
+      "step": 7283
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4227555140967402e-06,
+      "loss": 2.1977,
+      "step": 7284
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4215363314813208e-06,
+      "loss": 2.2238,
+      "step": 7285
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.420317584890844e-06,
+      "loss": 2.1086,
+      "step": 7286
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4190992744738135e-06,
+      "loss": 2.1529,
+      "step": 7287
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4178814003786706e-06,
+      "loss": 2.2124,
+      "step": 7288
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4166639627538153e-06,
+      "loss": 2.1717,
+      "step": 7289
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4154469617475864e-06,
+      "loss": 2.1872,
+      "step": 7290
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4142303975082723e-06,
+      "loss": 2.1797,
+      "step": 7291
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4130142701841076e-06,
+      "loss": 2.1643,
+      "step": 7292
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4117985799232735e-06,
+      "loss": 2.1577,
+      "step": 7293
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4105833268738966e-06,
+      "loss": 2.2151,
+      "step": 7294
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4093685111840567e-06,
+      "loss": 2.1543,
+      "step": 7295
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4081541330017706e-06,
+      "loss": 2.1539,
+      "step": 7296
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4069401924750082e-06,
+      "loss": 2.1646,
+      "step": 7297
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4057266897516842e-06,
+      "loss": 2.1917,
+      "step": 7298
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4045136249796588e-06,
+      "loss": 2.223,
+      "step": 7299
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4033009983067454e-06,
+      "loss": 2.1137,
+      "step": 7300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4020888098806924e-06,
+      "loss": 2.1698,
+      "step": 7301
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4008770598492072e-06,
+      "loss": 2.1754,
+      "step": 7302
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3996657483599318e-06,
+      "loss": 2.136,
+      "step": 7303
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3984548755604655e-06,
+      "loss": 2.0994,
+      "step": 7304
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3972444415983495e-06,
+      "loss": 2.109,
+      "step": 7305
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3960344466210669e-06,
+      "loss": 2.1727,
+      "step": 7306
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3948248907760565e-06,
+      "loss": 2.1543,
+      "step": 7307
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3936157742106977e-06,
+      "loss": 2.1572,
+      "step": 7308
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3924070970723176e-06,
+      "loss": 2.1652,
+      "step": 7309
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3911988595081894e-06,
+      "loss": 2.1661,
+      "step": 7310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3899910616655338e-06,
+      "loss": 2.1501,
+      "step": 7311
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3887837036915169e-06,
+      "loss": 2.1258,
+      "step": 7312
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3875767857332512e-06,
+      "loss": 2.153,
+      "step": 7313
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3863703079377971e-06,
+      "loss": 2.2089,
+      "step": 7314
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3851642704521596e-06,
+      "loss": 2.1225,
+      "step": 7315
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3839586734232907e-06,
+      "loss": 2.1638,
+      "step": 7316
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3827535169980888e-06,
+      "loss": 2.0885,
+      "step": 7317
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3815488013233986e-06,
+      "loss": 2.1456,
+      "step": 7318
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3803445265460096e-06,
+      "loss": 2.1783,
+      "step": 7319
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3791406928126638e-06,
+      "loss": 2.1045,
+      "step": 7320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3779373002700391e-06,
+      "loss": 2.1621,
+      "step": 7321
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3767343490647668e-06,
+      "loss": 2.2317,
+      "step": 7322
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3755318393434259e-06,
+      "loss": 2.1566,
+      "step": 7323
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3743297712525334e-06,
+      "loss": 2.1846,
+      "step": 7324
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.373128144938563e-06,
+      "loss": 2.1938,
+      "step": 7325
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3719269605479241e-06,
+      "loss": 2.1683,
+      "step": 7326
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3707262182269814e-06,
+      "loss": 2.1458,
+      "step": 7327
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3695259181220405e-06,
+      "loss": 2.1756,
+      "step": 7328
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.368326060379354e-06,
+      "loss": 2.166,
+      "step": 7329
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3671266451451209e-06,
+      "loss": 2.1324,
+      "step": 7330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3659276725654863e-06,
+      "loss": 2.1839,
+      "step": 7331
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3647291427865417e-06,
+      "loss": 2.1979,
+      "step": 7332
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3635310559543235e-06,
+      "loss": 2.0514,
+      "step": 7333
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3623334122148164e-06,
+      "loss": 2.1505,
+      "step": 7334
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3611362117139481e-06,
+      "loss": 2.1909,
+      "step": 7335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3599394545975952e-06,
+      "loss": 2.1767,
+      "step": 7336
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3587431410115765e-06,
+      "loss": 2.2312,
+      "step": 7337
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3575472711016634e-06,
+      "loss": 2.1323,
+      "step": 7338
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.356351845013566e-06,
+      "loss": 2.1801,
+      "step": 7339
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3551568628929434e-06,
+      "loss": 2.1544,
+      "step": 7340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3539623248854012e-06,
+      "loss": 2.2036,
+      "step": 7341
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3527682311364886e-06,
+      "loss": 2.1801,
+      "step": 7342
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.351574581791707e-06,
+      "loss": 2.1817,
+      "step": 7343
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3503813769964923e-06,
+      "loss": 2.2133,
+      "step": 7344
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.349188616896238e-06,
+      "loss": 2.2105,
+      "step": 7345
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3479963016362768e-06,
+      "loss": 2.1488,
+      "step": 7346
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3468044313618883e-06,
+      "loss": 2.1695,
+      "step": 7347
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3456130062183003e-06,
+      "loss": 2.1949,
+      "step": 7348
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3444220263506797e-06,
+      "loss": 2.1234,
+      "step": 7349
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3432314919041478e-06,
+      "loss": 2.2115,
+      "step": 7350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3420414030237667e-06,
+      "loss": 2.2391,
+      "step": 7351
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3408517598545446e-06,
+      "loss": 2.1991,
+      "step": 7352
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3396625625414362e-06,
+      "loss": 2.2355,
+      "step": 7353
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3384738112293415e-06,
+      "loss": 2.1928,
+      "step": 7354
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3372855060631067e-06,
+      "loss": 2.1041,
+      "step": 7355
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3360976471875226e-06,
+      "loss": 2.1832,
+      "step": 7356
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3349102347473264e-06,
+      "loss": 2.1318,
+      "step": 7357
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.333723268887201e-06,
+      "loss": 2.1825,
+      "step": 7358
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3325367497517739e-06,
+      "loss": 2.2309,
+      "step": 7359
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3313506774856177e-06,
+      "loss": 2.2645,
+      "step": 7360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3301650522332566e-06,
+      "loss": 2.1779,
+      "step": 7361
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3289798741391486e-06,
+      "loss": 2.1355,
+      "step": 7362
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.327795143347711e-06,
+      "loss": 2.1968,
+      "step": 7363
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3266108600032928e-06,
+      "loss": 2.2182,
+      "step": 7364
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3254270242502004e-06,
+      "loss": 2.2066,
+      "step": 7365
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3242436362326804e-06,
+      "loss": 2.1937,
+      "step": 7366
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3230606960949204e-06,
+      "loss": 2.1509,
+      "step": 7367
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3218782039810634e-06,
+      "loss": 2.1332,
+      "step": 7368
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3206961600351897e-06,
+      "loss": 2.1021,
+      "step": 7369
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3195145644013286e-06,
+      "loss": 2.1387,
+      "step": 7370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3183334172234536e-06,
+      "loss": 2.1466,
+      "step": 7371
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.317152718645484e-06,
+      "loss": 2.1665,
+      "step": 7372
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3159724688112846e-06,
+      "loss": 2.1695,
+      "step": 7373
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.314792667864665e-06,
+      "loss": 2.152,
+      "step": 7374
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3136133159493803e-06,
+      "loss": 2.1563,
+      "step": 7375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.312434413209131e-06,
+      "loss": 2.1958,
+      "step": 7376
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3112559597875628e-06,
+      "loss": 2.1258,
+      "step": 7377
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3100779558282673e-06,
+      "loss": 2.2187,
+      "step": 7378
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3089004014747797e-06,
+      "loss": 2.1623,
+      "step": 7379
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3077232968705805e-06,
+      "loss": 2.225,
+      "step": 7380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3065466421591006e-06,
+      "loss": 2.1737,
+      "step": 7381
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3053704374837063e-06,
+      "loss": 2.1062,
+      "step": 7382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3041946829877178e-06,
+      "loss": 2.188,
+      "step": 7383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3030193788143991e-06,
+      "loss": 2.1872,
+      "step": 7384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.301844525106951e-06,
+      "loss": 2.1583,
+      "step": 7385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3006701220085338e-06,
+      "loss": 2.215,
+      "step": 7386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.299496169662237e-06,
+      "loss": 2.1016,
+      "step": 7387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2983226682111094e-06,
+      "loss": 2.1955,
+      "step": 7388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2971496177981362e-06,
+      "loss": 2.208,
+      "step": 7389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2959770185662502e-06,
+      "loss": 2.1852,
+      "step": 7390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2948048706583284e-06,
+      "loss": 2.1795,
+      "step": 7391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2936331742171943e-06,
+      "loss": 2.19,
+      "step": 7392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2924619293856155e-06,
+      "loss": 2.1851,
+      "step": 7393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2912911363063048e-06,
+      "loss": 2.1163,
+      "step": 7394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2901207951219186e-06,
+      "loss": 2.1795,
+      "step": 7395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2889509059750605e-06,
+      "loss": 2.1847,
+      "step": 7396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.287781469008278e-06,
+      "loss": 2.1626,
+      "step": 7397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2866124843640614e-06,
+      "loss": 2.1576,
+      "step": 7398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2854439521848526e-06,
+      "loss": 2.2104,
+      "step": 7399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2842758726130283e-06,
+      "loss": 2.2011,
+      "step": 7400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2831082457909206e-06,
+      "loss": 2.1676,
+      "step": 7401
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2819410718607972e-06,
+      "loss": 2.1229,
+      "step": 7402
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2807743509648745e-06,
+      "loss": 2.1888,
+      "step": 7403
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2796080832453183e-06,
+      "loss": 2.1946,
+      "step": 7404
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2784422688442294e-06,
+      "loss": 2.1634,
+      "step": 7405
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2772769079036639e-06,
+      "loss": 2.1279,
+      "step": 7406
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2761120005656125e-06,
+      "loss": 2.1832,
+      "step": 7407
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2749475469720196e-06,
+      "loss": 2.1642,
+      "step": 7408
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2737835472647686e-06,
+      "loss": 2.194,
+      "step": 7409
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2726200015856893e-06,
+      "loss": 2.2255,
+      "step": 7410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2714569100765567e-06,
+      "loss": 2.2502,
+      "step": 7411
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2702942728790897e-06,
+      "loss": 2.1571,
+      "step": 7412
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2691320901349518e-06,
+      "loss": 2.1492,
+      "step": 7413
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2679703619857525e-06,
+      "loss": 2.137,
+      "step": 7414
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2668090885730439e-06,
+      "loss": 2.1719,
+      "step": 7415
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2656482700383238e-06,
+      "loss": 2.1915,
+      "step": 7416
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2644879065230343e-06,
+      "loss": 2.1351,
+      "step": 7417
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2633279981685608e-06,
+      "loss": 2.1438,
+      "step": 7418
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2621685451162397e-06,
+      "loss": 2.0974,
+      "step": 7419
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2610095475073415e-06,
+      "loss": 2.1105,
+      "step": 7420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2598510054830888e-06,
+      "loss": 2.1391,
+      "step": 7421
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2586929191846453e-06,
+      "loss": 2.1363,
+      "step": 7422
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.25753528875312e-06,
+      "loss": 2.1676,
+      "step": 7423
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2563781143295705e-06,
+      "loss": 2.1985,
+      "step": 7424
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2552213960549891e-06,
+      "loss": 2.1531,
+      "step": 7425
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2540651340703231e-06,
+      "loss": 2.1847,
+      "step": 7426
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2529093285164579e-06,
+      "loss": 2.1212,
+      "step": 7427
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2517539795342248e-06,
+      "loss": 2.1795,
+      "step": 7428
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2505990872644008e-06,
+      "loss": 2.2266,
+      "step": 7429
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2494446518477022e-06,
+      "loss": 2.1632,
+      "step": 7430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.248290673424798e-06,
+      "loss": 2.1916,
+      "step": 7431
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2471371521362946e-06,
+      "loss": 2.1622,
+      "step": 7432
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2459840881227458e-06,
+      "loss": 2.1818,
+      "step": 7433
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2448314815246487e-06,
+      "loss": 2.2347,
+      "step": 7434
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2436793324824448e-06,
+      "loss": 2.1551,
+      "step": 7435
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.24252764113652e-06,
+      "loss": 2.1349,
+      "step": 7436
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.241376407627205e-06,
+      "loss": 2.1599,
+      "step": 7437
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.240225632094773e-06,
+      "loss": 2.1641,
+      "step": 7438
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2390753146794438e-06,
+      "loss": 2.2475,
+      "step": 7439
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2379254555213788e-06,
+      "loss": 2.2181,
+      "step": 7440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2367760547606844e-06,
+      "loss": 2.218,
+      "step": 7441
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2356271125374153e-06,
+      "loss": 2.1836,
+      "step": 7442
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.234478628991561e-06,
+      "loss": 2.1513,
+      "step": 7443
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2333306042630672e-06,
+      "loss": 2.1573,
+      "step": 7444
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2321830384918116e-06,
+      "loss": 2.1625,
+      "step": 7445
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2310359318176229e-06,
+      "loss": 2.1715,
+      "step": 7446
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2298892843802756e-06,
+      "loss": 2.2139,
+      "step": 7447
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2287430963194807e-06,
+      "loss": 2.1921,
+      "step": 7448
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2275973677749015e-06,
+      "loss": 2.1711,
+      "step": 7449
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.22645209888614e-06,
+      "loss": 2.138,
+      "step": 7450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2253072897927437e-06,
+      "loss": 2.123,
+      "step": 7451
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2241629406342048e-06,
+      "loss": 2.1775,
+      "step": 7452
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.223019051549958e-06,
+      "loss": 2.1525,
+      "step": 7453
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2218756226793827e-06,
+      "loss": 2.1007,
+      "step": 7454
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2207326541618024e-06,
+      "loss": 2.1738,
+      "step": 7455
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2195901461364851e-06,
+      "loss": 2.112,
+      "step": 7456
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.218448098742641e-06,
+      "loss": 2.162,
+      "step": 7457
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.217306512119425e-06,
+      "loss": 2.131,
+      "step": 7458
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.216165386405937e-06,
+      "loss": 2.1945,
+      "step": 7459
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2150247217412186e-06,
+      "loss": 2.1536,
+      "step": 7460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2138845182642555e-06,
+      "loss": 2.136,
+      "step": 7461
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2127447761139821e-06,
+      "loss": 2.1562,
+      "step": 7462
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2116054954292688e-06,
+      "loss": 2.195,
+      "step": 7463
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2104666763489326e-06,
+      "loss": 2.1756,
+      "step": 7464
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.20932831901174e-06,
+      "loss": 2.1544,
+      "step": 7465
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2081904235563908e-06,
+      "loss": 2.2214,
+      "step": 7466
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2070529901215388e-06,
+      "loss": 2.1957,
+      "step": 7467
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2059160188457724e-06,
+      "loss": 2.1506,
+      "step": 7468
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2047795098676317e-06,
+      "loss": 2.1726,
+      "step": 7469
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.203643463325596e-06,
+      "loss": 2.1407,
+      "step": 7470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2025078793580885e-06,
+      "loss": 2.165,
+      "step": 7471
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2013727581034783e-06,
+      "loss": 2.1534,
+      "step": 7472
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2002380997000717e-06,
+      "loss": 2.216,
+      "step": 7473
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.199103904286129e-06,
+      "loss": 2.2081,
+      "step": 7474
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1979701719998454e-06,
+      "loss": 2.2169,
+      "step": 7475
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1968369029793642e-06,
+      "loss": 2.2025,
+      "step": 7476
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1957040973627698e-06,
+      "loss": 2.1534,
+      "step": 7477
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1945717552880919e-06,
+      "loss": 2.173,
+      "step": 7478
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.193439876893301e-06,
+      "loss": 2.1148,
+      "step": 7479
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1923084623163172e-06,
+      "loss": 2.1744,
+      "step": 7480
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1911775116949958e-06,
+      "loss": 2.152,
+      "step": 7481
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1900470251671415e-06,
+      "loss": 2.1942,
+      "step": 7482
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1889170028705e-06,
+      "loss": 2.2398,
+      "step": 7483
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.18778744494276e-06,
+      "loss": 2.1454,
+      "step": 7484
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1866583515215597e-06,
+      "loss": 2.2448,
+      "step": 7485
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.185529722744469e-06,
+      "loss": 2.1789,
+      "step": 7486
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1844015587490138e-06,
+      "loss": 2.1615,
+      "step": 7487
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1832738596726518e-06,
+      "loss": 2.1471,
+      "step": 7488
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1821466256527942e-06,
+      "loss": 2.2259,
+      "step": 7489
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1810198568267906e-06,
+      "loss": 2.1523,
+      "step": 7490
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1798935533319305e-06,
+      "loss": 2.2006,
+      "step": 7491
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.178767715305455e-06,
+      "loss": 2.1703,
+      "step": 7492
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1776423428845423e-06,
+      "loss": 2.1757,
+      "step": 7493
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1765174362063152e-06,
+      "loss": 2.1611,
+      "step": 7494
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1753929954078414e-06,
+      "loss": 2.1875,
+      "step": 7495
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1742690206261293e-06,
+      "loss": 2.1661,
+      "step": 7496
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1731455119981327e-06,
+      "loss": 2.2283,
+      "step": 7497
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1720224696607474e-06,
+      "loss": 2.2018,
+      "step": 7498
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1708998937508126e-06,
+      "loss": 2.208,
+      "step": 7499
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1697777844051105e-06,
+      "loss": 2.1556,
+      "step": 7500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1686561417603677e-06,
+      "loss": 2.171,
+      "step": 7501
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1675349659532514e-06,
+      "loss": 2.1653,
+      "step": 7502
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1664142571203751e-06,
+      "loss": 2.1794,
+      "step": 7503
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1652940153982917e-06,
+      "loss": 2.1359,
+      "step": 7504
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.164174240923503e-06,
+      "loss": 2.2292,
+      "step": 7505
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1630549338324454e-06,
+      "loss": 2.1469,
+      "step": 7506
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1619360942615065e-06,
+      "loss": 2.154,
+      "step": 7507
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.160817722347014e-06,
+      "loss": 2.156,
+      "step": 7508
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.159699818225234e-06,
+      "loss": 2.1692,
+      "step": 7509
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1585823820323845e-06,
+      "loss": 2.0952,
+      "step": 7510
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1574654139046171e-06,
+      "loss": 2.2324,
+      "step": 7511
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1563489139780344e-06,
+      "loss": 2.1347,
+      "step": 7512
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1552328823886776e-06,
+      "loss": 2.1639,
+      "step": 7513
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.154117319272532e-06,
+      "loss": 2.1384,
+      "step": 7514
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1530022247655253e-06,
+      "loss": 2.1598,
+      "step": 7515
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1518875990035278e-06,
+      "loss": 2.1868,
+      "step": 7516
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1507734421223544e-06,
+      "loss": 2.1917,
+      "step": 7517
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1496597542577603e-06,
+      "loss": 2.149,
+      "step": 7518
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1485465355454467e-06,
+      "loss": 2.2229,
+      "step": 7519
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1474337861210543e-06,
+      "loss": 2.1357,
+      "step": 7520
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1463215061201694e-06,
+      "loss": 2.155,
+      "step": 7521
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1452096956783181e-06,
+      "loss": 2.1446,
+      "step": 7522
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1440983549309753e-06,
+      "loss": 2.1662,
+      "step": 7523
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1429874840135492e-06,
+      "loss": 2.1666,
+      "step": 7524
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1418770830614012e-06,
+      "loss": 2.1701,
+      "step": 7525
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1407671522098262e-06,
+      "loss": 2.1448,
+      "step": 7526
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.139657691594066e-06,
+      "loss": 2.1349,
+      "step": 7527
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1385487013493095e-06,
+      "loss": 2.1612,
+      "step": 7528
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1374401816106778e-06,
+      "loss": 2.2073,
+      "step": 7529
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.136332132513245e-06,
+      "loss": 2.1658,
+      "step": 7530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.135224554192022e-06,
+      "loss": 2.1541,
+      "step": 7531
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1341174467819637e-06,
+      "loss": 2.1621,
+      "step": 7532
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1330108104179682e-06,
+      "loss": 2.1774,
+      "step": 7533
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1319046452348758e-06,
+      "loss": 2.1768,
+      "step": 7534
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1307989513674695e-06,
+      "loss": 2.1845,
+      "step": 7535
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.129693728950474e-06,
+      "loss": 2.1543,
+      "step": 7536
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1285889781185576e-06,
+      "loss": 2.1762,
+      "step": 7537
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1274846990063314e-06,
+      "loss": 2.2579,
+      "step": 7538
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1263808917483476e-06,
+      "loss": 2.2179,
+      "step": 7539
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1252775564791023e-06,
+      "loss": 2.1973,
+      "step": 7540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1241746933330338e-06,
+      "loss": 2.1353,
+      "step": 7541
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1230723024445212e-06,
+      "loss": 2.1537,
+      "step": 7542
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1219703839478907e-06,
+      "loss": 2.1823,
+      "step": 7543
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.120868937977404e-06,
+      "loss": 2.1269,
+      "step": 7544
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1197679646672698e-06,
+      "loss": 2.1837,
+      "step": 7545
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1186674641516415e-06,
+      "loss": 2.1492,
+      "step": 7546
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1175674365646067e-06,
+      "loss": 2.1504,
+      "step": 7547
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1164678820402059e-06,
+      "loss": 2.2598,
+      "step": 7548
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1153688007124109e-06,
+      "loss": 2.1804,
+      "step": 7549
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1142701927151456e-06,
+      "loss": 2.1844,
+      "step": 7550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1131720581822703e-06,
+      "loss": 2.1659,
+      "step": 7551
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.11207439724759e-06,
+      "loss": 2.1262,
+      "step": 7552
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1109772100448512e-06,
+      "loss": 2.1559,
+      "step": 7553
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1098804967077425e-06,
+      "loss": 2.2115,
+      "step": 7554
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1087842573698953e-06,
+      "loss": 2.1819,
+      "step": 7555
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1076884921648834e-06,
+      "loss": 2.1503,
+      "step": 7556
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1065932012262215e-06,
+      "loss": 2.1182,
+      "step": 7557
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1054983846873684e-06,
+      "loss": 2.1432,
+      "step": 7558
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1044040426817237e-06,
+      "loss": 2.174,
+      "step": 7559
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1033101753426285e-06,
+      "loss": 2.1537,
+      "step": 7560
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1022167828033715e-06,
+      "loss": 2.1736,
+      "step": 7561
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1011238651971744e-06,
+      "loss": 2.1855,
+      "step": 7562
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1000314226572083e-06,
+      "loss": 2.1806,
+      "step": 7563
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0989394553165833e-06,
+      "loss": 2.1481,
+      "step": 7564
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.097847963308351e-06,
+      "loss": 2.2099,
+      "step": 7565
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0967569467655104e-06,
+      "loss": 2.1706,
+      "step": 7566
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0956664058209936e-06,
+      "loss": 2.0815,
+      "step": 7567
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0945763406076837e-06,
+      "loss": 2.1585,
+      "step": 7568
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0934867512584013e-06,
+      "loss": 2.1775,
+      "step": 7569
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0923976379059059e-06,
+      "loss": 2.1938,
+      "step": 7570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0913090006829085e-06,
+      "loss": 2.1716,
+      "step": 7571
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.09022083972205e-06,
+      "loss": 2.1887,
+      "step": 7572
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0891331551559237e-06,
+      "loss": 2.1243,
+      "step": 7573
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0880459471170597e-06,
+      "loss": 2.1848,
+      "step": 7574
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0869592157379305e-06,
+      "loss": 2.1695,
+      "step": 7575
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0858729611509516e-06,
+      "loss": 2.1778,
+      "step": 7576
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0847871834884798e-06,
+      "loss": 2.1509,
+      "step": 7577
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0837018828828133e-06,
+      "loss": 2.1448,
+      "step": 7578
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0826170594661933e-06,
+      "loss": 2.1126,
+      "step": 7579
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0815327133708015e-06,
+      "loss": 2.1759,
+      "step": 7580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.080448844728763e-06,
+      "loss": 2.1334,
+      "step": 7581
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0793654536721432e-06,
+      "loss": 2.1779,
+      "step": 7582
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0782825403329488e-06,
+      "loss": 2.1795,
+      "step": 7583
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.077200104843134e-06,
+      "loss": 2.1926,
+      "step": 7584
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.076118147334585e-06,
+      "loss": 2.1916,
+      "step": 7585
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0750366679391393e-06,
+      "loss": 2.1991,
+      "step": 7586
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0739556667885692e-06,
+      "loss": 2.1368,
+      "step": 7587
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0728751440145907e-06,
+      "loss": 2.1383,
+      "step": 7588
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0717950997488662e-06,
+      "loss": 2.2118,
+      "step": 7589
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0707155341229902e-06,
+      "loss": 2.2275,
+      "step": 7590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0696364472685112e-06,
+      "loss": 2.1387,
+      "step": 7591
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0685578393169054e-06,
+      "loss": 2.1462,
+      "step": 7592
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0674797103996033e-06,
+      "loss": 2.1688,
+      "step": 7593
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0664020606479702e-06,
+      "loss": 2.1667,
+      "step": 7594
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.065324890193314e-06,
+      "loss": 2.2514,
+      "step": 7595
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.064248199166884e-06,
+      "loss": 2.1525,
+      "step": 7596
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0631719876998736e-06,
+      "loss": 2.1601,
+      "step": 7597
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0620962559234144e-06,
+      "loss": 2.2269,
+      "step": 7598
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0610210039685815e-06,
+      "loss": 2.2035,
+      "step": 7599
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0599462319663906e-06,
+      "loss": 2.1693,
+      "step": 7600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0588719400478004e-06,
+      "loss": 2.1867,
+      "step": 7601
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0577981283437095e-06,
+      "loss": 2.165,
+      "step": 7602
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0567247969849576e-06,
+      "loss": 2.1318,
+      "step": 7603
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0556519461023301e-06,
+      "loss": 2.2334,
+      "step": 7604
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0545795758265476e-06,
+      "loss": 2.2037,
+      "step": 7605
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.053507686288276e-06,
+      "loss": 2.1843,
+      "step": 7606
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.052436277618122e-06,
+      "loss": 2.112,
+      "step": 7607
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0513653499466315e-06,
+      "loss": 2.146,
+      "step": 7608
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0502949034042985e-06,
+      "loss": 2.1442,
+      "step": 7609
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.049224938121548e-06,
+      "loss": 2.2331,
+      "step": 7610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0481554542287565e-06,
+      "loss": 2.1462,
+      "step": 7611
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.047086451856235e-06,
+      "loss": 2.2231,
+      "step": 7612
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0460179311342394e-06,
+      "loss": 2.1813,
+      "step": 7613
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0449498921929669e-06,
+      "loss": 2.1466,
+      "step": 7614
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.04388233516255e-06,
+      "loss": 2.1699,
+      "step": 7615
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0428152601730718e-06,
+      "loss": 2.2038,
+      "step": 7616
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0417486673545508e-06,
+      "loss": 2.1619,
+      "step": 7617
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0406825568369478e-06,
+      "loss": 2.1966,
+      "step": 7618
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0396169287501652e-06,
+      "loss": 2.1153,
+      "step": 7619
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0385517832240472e-06,
+      "loss": 2.1689,
+      "step": 7620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0374871203883774e-06,
+      "loss": 2.1564,
+      "step": 7621
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.036422940372883e-06,
+      "loss": 2.1408,
+      "step": 7622
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0353592433072302e-06,
+      "loss": 2.107,
+      "step": 7623
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0342960293210281e-06,
+      "loss": 2.2329,
+      "step": 7624
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0332332985438248e-06,
+      "loss": 2.1889,
+      "step": 7625
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0321710511051108e-06,
+      "loss": 2.1381,
+      "step": 7626
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0311092871343209e-06,
+      "loss": 2.161,
+      "step": 7627
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0300480067608232e-06,
+      "loss": 2.184,
+      "step": 7628
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0289872101139359e-06,
+      "loss": 2.1513,
+      "step": 7629
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0279268973229089e-06,
+      "loss": 2.2026,
+      "step": 7630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.026867068516943e-06,
+      "loss": 2.1678,
+      "step": 7631
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0258077238251735e-06,
+      "loss": 2.1463,
+      "step": 7632
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0247488633766756e-06,
+      "loss": 2.1151,
+      "step": 7633
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0236904873004722e-06,
+      "loss": 2.1221,
+      "step": 7634
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0226325957255207e-06,
+      "loss": 2.2037,
+      "step": 7635
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0215751887807228e-06,
+      "loss": 2.1918,
+      "step": 7636
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.020518266594921e-06,
+      "loss": 2.1787,
+      "step": 7637
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0194618292968972e-06,
+      "loss": 2.1634,
+      "step": 7638
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0184058770153748e-06,
+      "loss": 2.1186,
+      "step": 7639
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0173504098790188e-06,
+      "loss": 2.1548,
+      "step": 7640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.016295428016435e-06,
+      "loss": 2.1669,
+      "step": 7641
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0152409315561696e-06,
+      "loss": 2.197,
+      "step": 7642
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0141869206267097e-06,
+      "loss": 2.1763,
+      "step": 7643
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0131333953564825e-06,
+      "loss": 2.1605,
+      "step": 7644
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0120803558738585e-06,
+      "loss": 2.1379,
+      "step": 7645
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0110278023071445e-06,
+      "loss": 2.1886,
+      "step": 7646
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0099757347845957e-06,
+      "loss": 2.2203,
+      "step": 7647
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0089241534343986e-06,
+      "loss": 2.1581,
+      "step": 7648
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0078730583846879e-06,
+      "loss": 2.1046,
+      "step": 7649
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.006822449763537e-06,
+      "loss": 2.1679,
+      "step": 7650
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0057723276989551e-06,
+      "loss": 2.1722,
+      "step": 7651
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0047226923189024e-06,
+      "loss": 2.1729,
+      "step": 7652
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.003673543751268e-06,
+      "loss": 2.1901,
+      "step": 7653
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0026248821238915e-06,
+      "loss": 2.2214,
+      "step": 7654
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0015767075645472e-06,
+      "loss": 2.1842,
+      "step": 7655
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0005290202009533e-06,
+      "loss": 2.1699,
+      "step": 7656
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.994818201607665e-07,
+      "loss": 2.131,
+      "step": 7657
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.984351075715848e-07,
+      "loss": 2.1671,
+      "step": 7658
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.973888825609474e-07,
+      "loss": 2.195,
+      "step": 7659
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.963431452563331e-07,
+      "loss": 2.1947,
+      "step": 7660
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.952978957851622e-07,
+      "loss": 2.2305,
+      "step": 7661
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.942531342747953e-07,
+      "loss": 2.1916,
+      "step": 7662
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.932088608525336e-07,
+      "loss": 2.1115,
+      "step": 7663
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.921650756456164e-07,
+      "loss": 2.1929,
+      "step": 7664
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.911217787812305e-07,
+      "loss": 2.1945,
+      "step": 7665
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.900789703864933e-07,
+      "loss": 2.1213,
+      "step": 7666
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.890366505884725e-07,
+      "loss": 2.1832,
+      "step": 7667
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.879948195141681e-07,
+      "loss": 2.1928,
+      "step": 7668
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.869534772905242e-07,
+      "loss": 2.1786,
+      "step": 7669
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.859126240444284e-07,
+      "loss": 2.1549,
+      "step": 7670
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.848722599027012e-07,
+      "loss": 2.217,
+      "step": 7671
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.838323849921123e-07,
+      "loss": 2.2468,
+      "step": 7672
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.82792999439362e-07,
+      "loss": 2.1771,
+      "step": 7673
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.81754103371101e-07,
+      "loss": 2.1817,
+      "step": 7674
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.807156969139136e-07,
+      "loss": 2.1527,
+      "step": 7675
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.79677780194327e-07,
+      "loss": 2.1792,
+      "step": 7676
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.786403533388072e-07,
+      "loss": 2.195,
+      "step": 7677
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.77603416473763e-07,
+      "loss": 2.207,
+      "step": 7678
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.765669697255413e-07,
+      "loss": 2.1452,
+      "step": 7679
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.7553101322043e-07,
+      "loss": 2.1312,
+      "step": 7680
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.744955470846567e-07,
+      "loss": 2.1544,
+      "step": 7681
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.734605714443906e-07,
+      "loss": 2.0961,
+      "step": 7682
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.724260864257401e-07,
+      "loss": 2.2149,
+      "step": 7683
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.713920921547532e-07,
+      "loss": 2.1324,
+      "step": 7684
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.70358588757422e-07,
+      "loss": 2.2136,
+      "step": 7685
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.69325576359672e-07,
+      "loss": 2.13,
+      "step": 7686
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.682930550873742e-07,
+      "loss": 2.1987,
+      "step": 7687
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.67261025066339e-07,
+      "loss": 2.1829,
+      "step": 7688
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.662294864223132e-07,
+      "loss": 2.1313,
+      "step": 7689
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.651984392809916e-07,
+      "loss": 2.1885,
+      "step": 7690
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.641678837679985e-07,
+      "loss": 2.145,
+      "step": 7691
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.631378200089082e-07,
+      "loss": 2.1807,
+      "step": 7692
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.621082481292309e-07,
+      "loss": 2.1443,
+      "step": 7693
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.610791682544123e-07,
+      "loss": 2.1607,
+      "step": 7694
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.600505805098486e-07,
+      "loss": 2.1946,
+      "step": 7695
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.590224850208645e-07,
+      "loss": 2.1088,
+      "step": 7696
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.57994881912735e-07,
+      "loss": 2.2105,
+      "step": 7697
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.569677713106673e-07,
+      "loss": 2.1448,
+      "step": 7698
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.559411533398139e-07,
+      "loss": 2.1965,
+      "step": 7699
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.549150281252633e-07,
+      "loss": 2.1937,
+      "step": 7700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.538893957920464e-07,
+      "loss": 2.1593,
+      "step": 7701
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.528642564651341e-07,
+      "loss": 2.1986,
+      "step": 7702
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.518396102694355e-07,
+      "loss": 2.1579,
+      "step": 7703
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.508154573298012e-07,
+      "loss": 2.1222,
+      "step": 7704
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.497917977710208e-07,
+      "loss": 2.1525,
+      "step": 7705
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.487686317178241e-07,
+      "loss": 2.2117,
+      "step": 7706
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.477459592948796e-07,
+      "loss": 2.1217,
+      "step": 7707
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.467237806268009e-07,
+      "loss": 2.1393,
+      "step": 7708
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.457020958381324e-07,
+      "loss": 2.1786,
+      "step": 7709
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.446809050533679e-07,
+      "loss": 2.1769,
+      "step": 7710
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.436602083969326e-07,
+      "loss": 2.1984,
+      "step": 7711
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.426400059931956e-07,
+      "loss": 2.2277,
+      "step": 7712
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.41620297966469e-07,
+      "loss": 2.16,
+      "step": 7713
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.406010844409957e-07,
+      "loss": 2.165,
+      "step": 7714
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.395823655409686e-07,
+      "loss": 2.1535,
+      "step": 7715
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.385641413905139e-07,
+      "loss": 2.1797,
+      "step": 7716
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.375464121136984e-07,
+      "loss": 2.207,
+      "step": 7717
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.365291778345303e-07,
+      "loss": 2.2112,
+      "step": 7718
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.355124386769559e-07,
+      "loss": 2.1417,
+      "step": 7719
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.344961947648624e-07,
+      "loss": 2.1693,
+      "step": 7720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.334804462220748e-07,
+      "loss": 2.1524,
+      "step": 7721
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.3246519317236e-07,
+      "loss": 2.1612,
+      "step": 7722
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.314504357394233e-07,
+      "loss": 2.146,
+      "step": 7723
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.304361740469103e-07,
+      "loss": 2.1986,
+      "step": 7724
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.294224082184045e-07,
+      "loss": 2.1581,
+      "step": 7725
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.284091383774313e-07,
+      "loss": 2.2005,
+      "step": 7726
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.273963646474527e-07,
+      "loss": 2.0876,
+      "step": 7727
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.263840871518759e-07,
+      "loss": 2.2298,
+      "step": 7728
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.253723060140407e-07,
+      "loss": 2.0874,
+      "step": 7729
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.243610213572285e-07,
+      "loss": 2.1746,
+      "step": 7730
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.233502333046662e-07,
+      "loss": 2.1606,
+      "step": 7731
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.223399419795093e-07,
+      "loss": 2.1539,
+      "step": 7732
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.213301475048642e-07,
+      "loss": 2.2218,
+      "step": 7733
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.203208500037664e-07,
+      "loss": 2.169,
+      "step": 7734
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.193120495991986e-07,
+      "loss": 2.1823,
+      "step": 7735
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.183037464140804e-07,
+      "loss": 2.1912,
+      "step": 7736
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.17295940571269e-07,
+      "loss": 2.1707,
+      "step": 7737
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.162886321935632e-07,
+      "loss": 2.2031,
+      "step": 7738
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.152818214037007e-07,
+      "loss": 2.1241,
+      "step": 7739
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.142755083243577e-07,
+      "loss": 2.167,
+      "step": 7740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.132696930781509e-07,
+      "loss": 2.1835,
+      "step": 7741
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.122643757876354e-07,
+      "loss": 2.1998,
+      "step": 7742
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.112595565753063e-07,
+      "loss": 2.1836,
+      "step": 7743
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.10255235563598e-07,
+      "loss": 2.1605,
+      "step": 7744
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.09251412874882e-07,
+      "loss": 2.1931,
+      "step": 7745
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.08248088631476e-07,
+      "loss": 2.1392,
+      "step": 7746
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.072452629556272e-07,
+      "loss": 2.1798,
+      "step": 7747
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.06242935969528e-07,
+      "loss": 2.1272,
+      "step": 7748
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.052411077953099e-07,
+      "loss": 2.2035,
+      "step": 7749
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.042397785550405e-07,
+      "loss": 2.1541,
+      "step": 7750
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.032389483707332e-07,
+      "loss": 2.1803,
+      "step": 7751
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.022386173643305e-07,
+      "loss": 2.1845,
+      "step": 7752
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.012387856577238e-07,
+      "loss": 2.1587,
+      "step": 7753
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.002394533727382e-07,
+      "loss": 2.1704,
+      "step": 7754
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.992406206311394e-07,
+      "loss": 2.1294,
+      "step": 7755
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.982422875546332e-07,
+      "loss": 2.2182,
+      "step": 7756
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.972444542648595e-07,
+      "loss": 2.1783,
+      "step": 7757
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.962471208834056e-07,
+      "loss": 2.2012,
+      "step": 7758
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.95250287531792e-07,
+      "loss": 2.1415,
+      "step": 7759
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.942539543314799e-07,
+      "loss": 2.2022,
+      "step": 7760
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.932581214038693e-07,
+      "loss": 2.1777,
+      "step": 7761
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.922627888703e-07,
+      "loss": 2.14,
+      "step": 7762
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.912679568520494e-07,
+      "loss": 2.162,
+      "step": 7763
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.902736254703347e-07,
+      "loss": 2.1886,
+      "step": 7764
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.892797948463134e-07,
+      "loss": 2.1559,
+      "step": 7765
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.882864651010798e-07,
+      "loss": 2.1755,
+      "step": 7766
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.872936363556678e-07,
+      "loss": 2.1287,
+      "step": 7767
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.863013087310502e-07,
+      "loss": 2.1912,
+      "step": 7768
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.853094823481423e-07,
+      "loss": 2.1133,
+      "step": 7769
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.843181573277904e-07,
+      "loss": 2.1905,
+      "step": 7770
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.833273337907888e-07,
+      "loss": 2.0974,
+      "step": 7771
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.823370118578628e-07,
+      "loss": 2.1263,
+      "step": 7772
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.813471916496824e-07,
+      "loss": 2.1876,
+      "step": 7773
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.803578732868545e-07,
+      "loss": 2.1956,
+      "step": 7774
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.793690568899216e-07,
+      "loss": 2.1313,
+      "step": 7775
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.783807425793722e-07,
+      "loss": 2.1826,
+      "step": 7776
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.773929304756246e-07,
+      "loss": 2.1001,
+      "step": 7777
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.764056206990446e-07,
+      "loss": 2.1638,
+      "step": 7778
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.754188133699316e-07,
+      "loss": 2.2373,
+      "step": 7779
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.744325086085248e-07,
+      "loss": 2.1456,
+      "step": 7780
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.734467065350022e-07,
+      "loss": 2.1417,
+      "step": 7781
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.72461407269482e-07,
+      "loss": 2.2063,
+      "step": 7782
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.714766109320188e-07,
+      "loss": 2.1916,
+      "step": 7783
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.704923176426072e-07,
+      "loss": 2.1601,
+      "step": 7784
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.695085275211812e-07,
+      "loss": 2.1763,
+      "step": 7785
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.685252406876116e-07,
+      "loss": 2.0839,
+      "step": 7786
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.675424572617092e-07,
+      "loss": 2.1306,
+      "step": 7787
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.665601773632226e-07,
+      "loss": 2.18,
+      "step": 7788
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.655784011118424e-07,
+      "loss": 2.2179,
+      "step": 7789
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.645971286271903e-07,
+      "loss": 2.1915,
+      "step": 7790
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.636163600288372e-07,
+      "loss": 2.1543,
+      "step": 7791
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.626360954362817e-07,
+      "loss": 2.1383,
+      "step": 7792
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.616563349689672e-07,
+      "loss": 2.1812,
+      "step": 7793
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.606770787462776e-07,
+      "loss": 2.1721,
+      "step": 7794
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.596983268875281e-07,
+      "loss": 2.2102,
+      "step": 7795
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.587200795119793e-07,
+      "loss": 2.2247,
+      "step": 7796
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.577423367388271e-07,
+      "loss": 2.1726,
+      "step": 7797
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.567650986872061e-07,
+      "loss": 2.1989,
+      "step": 7798
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.557883654761906e-07,
+      "loss": 2.2102,
+      "step": 7799
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.54812137224792e-07,
+      "loss": 2.0968,
+      "step": 7800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.5383641405196e-07,
+      "loss": 2.1167,
+      "step": 7801
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.528611960765853e-07,
+      "loss": 2.1417,
+      "step": 7802
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.518864834174939e-07,
+      "loss": 2.1513,
+      "step": 7803
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.509122761934519e-07,
+      "loss": 2.1611,
+      "step": 7804
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.499385745231631e-07,
+      "loss": 2.1282,
+      "step": 7805
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.489653785252711e-07,
+      "loss": 2.1778,
+      "step": 7806
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.479926883183559e-07,
+      "loss": 2.1728,
+      "step": 7807
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.470205040209362e-07,
+      "loss": 2.1803,
+      "step": 7808
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.460488257514731e-07,
+      "loss": 2.114,
+      "step": 7809
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.450776536283594e-07,
+      "loss": 2.1025,
+      "step": 7810
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.441069877699287e-07,
+      "loss": 2.2033,
+      "step": 7811
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.431368282944585e-07,
+      "loss": 2.182,
+      "step": 7812
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.421671753201538e-07,
+      "loss": 2.1927,
+      "step": 7813
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.411980289651689e-07,
+      "loss": 2.1192,
+      "step": 7814
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.402293893475872e-07,
+      "loss": 2.141,
+      "step": 7815
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.392612565854374e-07,
+      "loss": 2.1338,
+      "step": 7816
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.382936307966838e-07,
+      "loss": 2.2071,
+      "step": 7817
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.373265120992252e-07,
+      "loss": 2.1754,
+      "step": 7818
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.363599006109057e-07,
+      "loss": 2.1487,
+      "step": 7819
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.353937964495029e-07,
+      "loss": 2.1999,
+      "step": 7820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.344281997327331e-07,
+      "loss": 2.1246,
+      "step": 7821
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.334631105782515e-07,
+      "loss": 2.0783,
+      "step": 7822
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.324985291036513e-07,
+      "loss": 2.1238,
+      "step": 7823
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.315344554264643e-07,
+      "loss": 2.1448,
+      "step": 7824
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.305708896641596e-07,
+      "loss": 2.2108,
+      "step": 7825
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.296078319341444e-07,
+      "loss": 2.1975,
+      "step": 7826
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.286452823537649e-07,
+      "loss": 2.1492,
+      "step": 7827
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.276832410403051e-07,
+      "loss": 2.1691,
+      "step": 7828
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.267217081109863e-07,
+      "loss": 2.1964,
+      "step": 7829
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.25760683682968e-07,
+      "loss": 2.1899,
+      "step": 7830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.248001678733475e-07,
+      "loss": 2.131,
+      "step": 7831
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.238401607991647e-07,
+      "loss": 2.1488,
+      "step": 7832
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.228806625773878e-07,
+      "loss": 2.1824,
+      "step": 7833
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.21921673324933e-07,
+      "loss": 2.141,
+      "step": 7834
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.209631931586499e-07,
+      "loss": 2.1419,
+      "step": 7835
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.200052221953231e-07,
+      "loss": 2.2331,
+      "step": 7836
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.190477605516828e-07,
+      "loss": 2.2253,
+      "step": 7837
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.180908083443884e-07,
+      "loss": 2.1885,
+      "step": 7838
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.171343656900455e-07,
+      "loss": 2.172,
+      "step": 7839
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.161784327051919e-07,
+      "loss": 2.1072,
+      "step": 7840
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.152230095063051e-07,
+      "loss": 2.2036,
+      "step": 7841
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.142680962098016e-07,
+      "loss": 2.1965,
+      "step": 7842
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.133136929320329e-07,
+      "loss": 2.1461,
+      "step": 7843
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.123597997892918e-07,
+      "loss": 2.2029,
+      "step": 7844
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.114064168978064e-07,
+      "loss": 2.2511,
+      "step": 7845
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.104535443737438e-07,
+      "loss": 2.1986,
+      "step": 7846
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.095011823332089e-07,
+      "loss": 2.2244,
+      "step": 7847
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.085493308922432e-07,
+      "loss": 2.1552,
+      "step": 7848
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.075979901668269e-07,
+      "loss": 2.1743,
+      "step": 7849
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.066471602728804e-07,
+      "loss": 2.1483,
+      "step": 7850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.056968413262555e-07,
+      "loss": 2.148,
+      "step": 7851
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.047470334427504e-07,
+      "loss": 2.1556,
+      "step": 7852
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.037977367380922e-07,
+      "loss": 2.2047,
+      "step": 7853
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.028489513279503e-07,
+      "loss": 2.2538,
+      "step": 7854
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.019006773279348e-07,
+      "loss": 2.2029,
+      "step": 7855
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.009529148535855e-07,
+      "loss": 2.2218,
+      "step": 7856
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.000056640203885e-07,
+      "loss": 2.1649,
+      "step": 7857
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.990589249437591e-07,
+      "loss": 2.2336,
+      "step": 7858
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.98112697739058e-07,
+      "loss": 2.133,
+      "step": 7859
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.971669825215789e-07,
+      "loss": 2.2419,
+      "step": 7860
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.962217794065547e-07,
+      "loss": 2.1199,
+      "step": 7861
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.952770885091548e-07,
+      "loss": 2.1046,
+      "step": 7862
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.94332909944488e-07,
+      "loss": 2.1537,
+      "step": 7863
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.933892438275987e-07,
+      "loss": 2.1282,
+      "step": 7864
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.924460902734698e-07,
+      "loss": 2.2503,
+      "step": 7865
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.91503449397022e-07,
+      "loss": 2.1412,
+      "step": 7866
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.90561321313113e-07,
+      "loss": 2.1408,
+      "step": 7867
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.89619706136539e-07,
+      "loss": 2.1599,
+      "step": 7868
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.88678603982031e-07,
+      "loss": 2.1367,
+      "step": 7869
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.877380149642628e-07,
+      "loss": 2.1818,
+      "step": 7870
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.867979391978398e-07,
+      "loss": 2.151,
+      "step": 7871
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.858583767973071e-07,
+      "loss": 2.2019,
+      "step": 7872
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.849193278771489e-07,
+      "loss": 2.2355,
+      "step": 7873
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.839807925517834e-07,
+      "loss": 2.2004,
+      "step": 7874
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.830427709355726e-07,
+      "loss": 2.1472,
+      "step": 7875
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.821052631428061e-07,
+      "loss": 2.0945,
+      "step": 7876
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.811682692877204e-07,
+      "loss": 2.1993,
+      "step": 7877
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.802317894844835e-07,
+      "loss": 2.1739,
+      "step": 7878
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.792958238472037e-07,
+      "loss": 2.2065,
+      "step": 7879
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.783603724899258e-07,
+      "loss": 2.1696,
+      "step": 7880
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.774254355266287e-07,
+      "loss": 2.1488,
+      "step": 7881
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.76491013071235e-07,
+      "loss": 2.2049,
+      "step": 7882
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.755571052376004e-07,
+      "loss": 2.2005,
+      "step": 7883
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.746237121395184e-07,
+      "loss": 2.212,
+      "step": 7884
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.736908338907195e-07,
+      "loss": 2.212,
+      "step": 7885
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.727584706048735e-07,
+      "loss": 2.2502,
+      "step": 7886
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.71826622395585e-07,
+      "loss": 2.1362,
+      "step": 7887
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.708952893763972e-07,
+      "loss": 2.199,
+      "step": 7888
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.699644716607896e-07,
+      "loss": 2.1631,
+      "step": 7889
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.690341693621805e-07,
+      "loss": 2.1189,
+      "step": 7890
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.681043825939238e-07,
+      "loss": 2.1135,
+      "step": 7891
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.671751114693104e-07,
+      "loss": 2.1474,
+      "step": 7892
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.662463561015726e-07,
+      "loss": 2.1579,
+      "step": 7893
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.653181166038715e-07,
+      "loss": 2.1683,
+      "step": 7894
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.643903930893154e-07,
+      "loss": 2.1368,
+      "step": 7895
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.63463185670939e-07,
+      "loss": 2.1966,
+      "step": 7896
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.625364944617242e-07,
+      "loss": 2.2151,
+      "step": 7897
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.61610319574585e-07,
+      "loss": 2.1761,
+      "step": 7898
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.606846611223695e-07,
+      "loss": 2.1595,
+      "step": 7899
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.597595192178702e-07,
+      "loss": 2.18,
+      "step": 7900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.588348939738116e-07,
+      "loss": 2.1129,
+      "step": 7901
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.579107855028562e-07,
+      "loss": 2.1802,
+      "step": 7902
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.569871939176037e-07,
+      "loss": 2.1749,
+      "step": 7903
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.560641193305912e-07,
+      "loss": 2.1649,
+      "step": 7904
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.551415618542928e-07,
+      "loss": 2.1447,
+      "step": 7905
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.542195216011188e-07,
+      "loss": 2.1693,
+      "step": 7906
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.532979986834177e-07,
+      "loss": 2.1496,
+      "step": 7907
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.523769932134739e-07,
+      "loss": 2.1461,
+      "step": 7908
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.514565053035083e-07,
+      "loss": 2.1972,
+      "step": 7909
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.505365350656813e-07,
+      "loss": 2.1545,
+      "step": 7910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.496170826120869e-07,
+      "loss": 2.17,
+      "step": 7911
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.486981480547567e-07,
+      "loss": 2.1598,
+      "step": 7912
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.477797315056645e-07,
+      "loss": 2.1834,
+      "step": 7913
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.468618330767114e-07,
+      "loss": 2.1547,
+      "step": 7914
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.459444528797438e-07,
+      "loss": 2.111,
+      "step": 7915
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.450275910265415e-07,
+      "loss": 2.1657,
+      "step": 7916
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.441112476288187e-07,
+      "loss": 2.1997,
+      "step": 7917
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.43195422798233e-07,
+      "loss": 2.1595,
+      "step": 7918
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.422801166463706e-07,
+      "loss": 2.1426,
+      "step": 7919
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.413653292847617e-07,
+      "loss": 2.1923,
+      "step": 7920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.4045106082487e-07,
+      "loss": 2.1873,
+      "step": 7921
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.395373113780962e-07,
+      "loss": 2.1982,
+      "step": 7922
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.386240810557771e-07,
+      "loss": 2.1996,
+      "step": 7923
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.377113699691879e-07,
+      "loss": 2.2159,
+      "step": 7924
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.367991782295392e-07,
+      "loss": 2.1346,
+      "step": 7925
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.358875059479792e-07,
+      "loss": 2.2074,
+      "step": 7926
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.349763532355919e-07,
+      "loss": 2.149,
+      "step": 7927
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.34065720203399e-07,
+      "loss": 2.212,
+      "step": 7928
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.33155606962358e-07,
+      "loss": 2.1461,
+      "step": 7929
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.322460136233622e-07,
+      "loss": 2.1285,
+      "step": 7930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.31336940297247e-07,
+      "loss": 2.188,
+      "step": 7931
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.304283870947748e-07,
+      "loss": 2.1759,
+      "step": 7932
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.295203541266549e-07,
+      "loss": 2.1946,
+      "step": 7933
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.286128415035249e-07,
+      "loss": 2.1835,
+      "step": 7934
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.277058493359629e-07,
+      "loss": 2.0902,
+      "step": 7935
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.267993777344856e-07,
+      "loss": 2.161,
+      "step": 7936
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.258934268095402e-07,
+      "loss": 2.2505,
+      "step": 7937
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.249879966715174e-07,
+      "loss": 2.1006,
+      "step": 7938
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.240830874307392e-07,
+      "loss": 2.1737,
+      "step": 7939
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.23178699197467e-07,
+      "loss": 2.2107,
+      "step": 7940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.222748320818984e-07,
+      "loss": 2.206,
+      "step": 7941
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.213714861941628e-07,
+      "loss": 2.1512,
+      "step": 7942
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.204686616443352e-07,
+      "loss": 2.1514,
+      "step": 7943
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.195663585424195e-07,
+      "loss": 2.2111,
+      "step": 7944
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.186645769983591e-07,
+      "loss": 2.0957,
+      "step": 7945
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.177633171220339e-07,
+      "loss": 2.1857,
+      "step": 7946
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.168625790232586e-07,
+      "loss": 2.1927,
+      "step": 7947
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.159623628117856e-07,
+      "loss": 2.1107,
+      "step": 7948
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.150626685973045e-07,
+      "loss": 2.2301,
+      "step": 7949
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.141634964894389e-07,
+      "loss": 2.1567,
+      "step": 7950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.132648465977515e-07,
+      "loss": 2.1873,
+      "step": 7951
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.123667190317396e-07,
+      "loss": 2.1575,
+      "step": 7952
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.114691139008356e-07,
+      "loss": 2.1108,
+      "step": 7953
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.105720313144143e-07,
+      "loss": 2.1931,
+      "step": 7954
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.096754713817771e-07,
+      "loss": 2.153,
+      "step": 7955
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.087794342121724e-07,
+      "loss": 2.1744,
+      "step": 7956
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.078839199147741e-07,
+      "loss": 2.0869,
+      "step": 7957
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.069889285987025e-07,
+      "loss": 2.1715,
+      "step": 7958
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.060944603730086e-07,
+      "loss": 2.1647,
+      "step": 7959
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.052005153466779e-07,
+      "loss": 2.1819,
+      "step": 7960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.043070936286395e-07,
+      "loss": 2.1609,
+      "step": 7961
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.034141953277484e-07,
+      "loss": 2.1773,
+      "step": 7962
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.025218205528061e-07,
+      "loss": 2.2366,
+      "step": 7963
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.01629969412545e-07,
+      "loss": 2.1467,
+      "step": 7964
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.007386420156332e-07,
+      "loss": 2.1093,
+      "step": 7965
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.99847838470677e-07,
+      "loss": 2.1489,
+      "step": 7966
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.989575588862174e-07,
+      "loss": 2.2056,
+      "step": 7967
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.980678033707333e-07,
+      "loss": 2.1292,
+      "step": 7968
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.971785720326385e-07,
+      "loss": 2.2234,
+      "step": 7969
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.962898649802824e-07,
+      "loss": 2.0803,
+      "step": 7970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.954016823219517e-07,
+      "loss": 2.1888,
+      "step": 7971
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.945140241658688e-07,
+      "loss": 2.1918,
+      "step": 7972
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.936268906201915e-07,
+      "loss": 2.1387,
+      "step": 7973
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.927402817930168e-07,
+      "loss": 2.1906,
+      "step": 7974
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.918541977923709e-07,
+      "loss": 2.1939,
+      "step": 7975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.909686387262255e-07,
+      "loss": 2.1661,
+      "step": 7976
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.9008360470248e-07,
+      "loss": 2.1138,
+      "step": 7977
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.891990958289724e-07,
+      "loss": 2.1743,
+      "step": 7978
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.883151122134812e-07,
+      "loss": 2.1664,
+      "step": 7979
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.874316539637127e-07,
+      "loss": 2.1885,
+      "step": 7980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.865487211873167e-07,
+      "loss": 2.1438,
+      "step": 7981
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.856663139918751e-07,
+      "loss": 2.0808,
+      "step": 7982
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.847844324849062e-07,
+      "loss": 2.2005,
+      "step": 7983
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.839030767738653e-07,
+      "loss": 2.208,
+      "step": 7984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.830222469661419e-07,
+      "loss": 2.1783,
+      "step": 7985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.821419431690629e-07,
+      "loss": 2.1893,
+      "step": 7986
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.81262165489891e-07,
+      "loss": 2.1496,
+      "step": 7987
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.803829140358237e-07,
+      "loss": 2.144,
+      "step": 7988
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.795041889139958e-07,
+      "loss": 2.1734,
+      "step": 7989
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.786259902314768e-07,
+      "loss": 2.1194,
+      "step": 7990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.777483180952732e-07,
+      "loss": 2.1583,
+      "step": 7991
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.768711726123261e-07,
+      "loss": 2.1759,
+      "step": 7992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.759945538895119e-07,
+      "loss": 2.0985,
+      "step": 7993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.751184620336471e-07,
+      "loss": 2.2133,
+      "step": 7994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.742428971514786e-07,
+      "loss": 2.1755,
+      "step": 7995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.733678593496901e-07,
+      "loss": 2.1944,
+      "step": 7996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.724933487349061e-07,
+      "loss": 2.1828,
+      "step": 7997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.716193654136788e-07,
+      "loss": 2.2063,
+      "step": 7998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.707459094925045e-07,
+      "loss": 2.1666,
+      "step": 7999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.698729810778065e-07,
+      "loss": 2.0343,
+      "step": 8000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.69000580275953e-07,
+      "loss": 2.1597,
+      "step": 8001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.681287071932408e-07,
+      "loss": 2.1798,
+      "step": 8002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.672573619359063e-07,
+      "loss": 2.2043,
+      "step": 8003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.663865446101192e-07,
+      "loss": 2.1736,
+      "step": 8004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.655162553219862e-07,
+      "loss": 2.1828,
+      "step": 8005
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.646464941775499e-07,
+      "loss": 2.23,
+      "step": 8006
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.637772612827881e-07,
+      "loss": 2.2191,
+      "step": 8007
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.629085567436133e-07,
+      "loss": 2.1781,
+      "step": 8008
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.620403806658754e-07,
+      "loss": 2.1783,
+      "step": 8009
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.611727331553585e-07,
+      "loss": 2.1966,
+      "step": 8010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.603056143177817e-07,
+      "loss": 2.1771,
+      "step": 8011
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.594390242588044e-07,
+      "loss": 2.1897,
+      "step": 8012
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.585729630840149e-07,
+      "loss": 2.1805,
+      "step": 8013
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.577074308989406e-07,
+      "loss": 2.1141,
+      "step": 8014
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.568424278090446e-07,
+      "loss": 2.1989,
+      "step": 8015
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.559779539197231e-07,
+      "loss": 2.2026,
+      "step": 8016
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.551140093363135e-07,
+      "loss": 2.1845,
+      "step": 8017
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.542505941640803e-07,
+      "loss": 2.1913,
+      "step": 8018
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.533877085082307e-07,
+      "loss": 2.1515,
+      "step": 8019
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.52525352473905e-07,
+      "loss": 2.1641,
+      "step": 8020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.516635261661775e-07,
+      "loss": 2.131,
+      "step": 8021
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.508022296900601e-07,
+      "loss": 2.1727,
+      "step": 8022
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.499414631504969e-07,
+      "loss": 2.2218,
+      "step": 8023
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.490812266523716e-07,
+      "loss": 2.1907,
+      "step": 8024
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.482215203005016e-07,
+      "loss": 2.147,
+      "step": 8025
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.47362344199639e-07,
+      "loss": 2.1367,
+      "step": 8026
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.465036984544721e-07,
+      "loss": 2.1081,
+      "step": 8027
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.456455831696234e-07,
+      "loss": 2.1694,
+      "step": 8028
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.447879984496525e-07,
+      "loss": 2.1931,
+      "step": 8029
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.439309443990532e-07,
+      "loss": 2.2339,
+      "step": 8030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.43074421122255e-07,
+      "loss": 2.1814,
+      "step": 8031
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.422184287236227e-07,
+      "loss": 2.1816,
+      "step": 8032
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.413629673074562e-07,
+      "loss": 2.1926,
+      "step": 8033
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.405080369779898e-07,
+      "loss": 2.1449,
+      "step": 8034
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.396536378393975e-07,
+      "loss": 2.1612,
+      "step": 8035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.387997699957815e-07,
+      "loss": 2.1905,
+      "step": 8036
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.379464335511859e-07,
+      "loss": 2.1691,
+      "step": 8037
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.370936286095842e-07,
+      "loss": 2.1296,
+      "step": 8038
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.362413552748908e-07,
+      "loss": 2.2094,
+      "step": 8039
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.353896136509524e-07,
+      "loss": 2.1093,
+      "step": 8040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.345384038415486e-07,
+      "loss": 2.1705,
+      "step": 8041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.336877259504004e-07,
+      "loss": 2.1948,
+      "step": 8042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.328375800811559e-07,
+      "loss": 2.0994,
+      "step": 8043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.319879663374068e-07,
+      "loss": 2.1806,
+      "step": 8044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.311388848226741e-07,
+      "loss": 2.1091,
+      "step": 8045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.302903356404161e-07,
+      "loss": 2.1664,
+      "step": 8046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.294423188940263e-07,
+      "loss": 2.1996,
+      "step": 8047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.28594834686832e-07,
+      "loss": 2.1863,
+      "step": 8048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.277478831220979e-07,
+      "loss": 2.1509,
+      "step": 8049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.269014643030214e-07,
+      "loss": 2.1938,
+      "step": 8050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.260555783327366e-07,
+      "loss": 2.2032,
+      "step": 8051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.252102253143122e-07,
+      "loss": 2.1894,
+      "step": 8052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.243654053507515e-07,
+      "loss": 2.2377,
+      "step": 8053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.235211185449919e-07,
+      "loss": 2.1671,
+      "step": 8054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.226773649999113e-07,
+      "loss": 2.1263,
+      "step": 8055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.218341448183141e-07,
+      "loss": 2.2598,
+      "step": 8056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.209914581029474e-07,
+      "loss": 2.1847,
+      "step": 8057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.201493049564883e-07,
+      "loss": 2.192,
+      "step": 8058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.193076854815494e-07,
+      "loss": 2.1817,
+      "step": 8059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.184665997806832e-07,
+      "loss": 2.181,
+      "step": 8060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.17626047956369e-07,
+      "loss": 2.1883,
+      "step": 8061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.167860301110284e-07,
+      "loss": 2.0796,
+      "step": 8062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.159465463470149e-07,
+      "loss": 2.1757,
+      "step": 8063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.151075967666165e-07,
+      "loss": 2.1713,
+      "step": 8064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.142691814720575e-07,
+      "loss": 2.1746,
+      "step": 8065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.134313005654929e-07,
+      "loss": 2.1129,
+      "step": 8066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.1259395414902e-07,
+      "loss": 2.118,
+      "step": 8067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.117571423246655e-07,
+      "loss": 2.2367,
+      "step": 8068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.109208651943921e-07,
+      "loss": 2.255,
+      "step": 8069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.100851228600974e-07,
+      "loss": 2.2175,
+      "step": 8070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.092499154236148e-07,
+      "loss": 2.2428,
+      "step": 8071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.084152429867113e-07,
+      "loss": 2.165,
+      "step": 8072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.075811056510894e-07,
+      "loss": 2.114,
+      "step": 8073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.067475035183862e-07,
+      "loss": 2.1356,
+      "step": 8074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.059144366901737e-07,
+      "loss": 2.1851,
+      "step": 8075
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.050819052679585e-07,
+      "loss": 2.141,
+      "step": 8076
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.042499093531806e-07,
+      "loss": 2.1437,
+      "step": 8077
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.034184490472195e-07,
+      "loss": 2.1358,
+      "step": 8078
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.025875244513824e-07,
+      "loss": 2.1944,
+      "step": 8079
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.017571356669183e-07,
+      "loss": 2.2114,
+      "step": 8080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.009272827950042e-07,
+      "loss": 2.1897,
+      "step": 8081
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.000979659367579e-07,
+      "loss": 2.1329,
+      "step": 8082
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.992691851932292e-07,
+      "loss": 2.1932,
+      "step": 8083
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.98440940665399e-07,
+      "loss": 2.2115,
+      "step": 8084
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9761323245419e-07,
+      "loss": 2.1663,
+      "step": 8085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.967860606604553e-07,
+      "loss": 2.1235,
+      "step": 8086
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.959594253849821e-07,
+      "loss": 2.2372,
+      "step": 8087
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.951333267284942e-07,
+      "loss": 2.1631,
+      "step": 8088
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.943077647916496e-07,
+      "loss": 2.1634,
+      "step": 8089
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.934827396750392e-07,
+      "loss": 2.1636,
+      "step": 8090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.926582514791912e-07,
+      "loss": 2.161,
+      "step": 8091
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.918343003045656e-07,
+      "loss": 2.1513,
+      "step": 8092
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.910108862515596e-07,
+      "loss": 2.2335,
+      "step": 8093
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.901880094205037e-07,
+      "loss": 2.1615,
+      "step": 8094
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.893656699116618e-07,
+      "loss": 2.19,
+      "step": 8095
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.885438678252342e-07,
+      "loss": 2.1929,
+      "step": 8096
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.877226032613542e-07,
+      "loss": 2.1371,
+      "step": 8097
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.869018763200929e-07,
+      "loss": 2.1531,
+      "step": 8098
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.860816871014496e-07,
+      "loss": 2.1257,
+      "step": 8099
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.852620357053651e-07,
+      "loss": 2.1793,
+      "step": 8100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.844429222317111e-07,
+      "loss": 2.2181,
+      "step": 8101
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.836243467802915e-07,
+      "loss": 2.1803,
+      "step": 8102
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.828063094508507e-07,
+      "loss": 2.1656,
+      "step": 8103
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.819888103430598e-07,
+      "loss": 2.1552,
+      "step": 8104
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.811718495565327e-07,
+      "loss": 2.1714,
+      "step": 8105
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.803554271908124e-07,
+      "loss": 2.169,
+      "step": 8106
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.795395433453765e-07,
+      "loss": 2.1602,
+      "step": 8107
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.787241981196384e-07,
+      "loss": 2.1938,
+      "step": 8108
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.779093916129464e-07,
+      "loss": 2.1858,
+      "step": 8109
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.770951239245803e-07,
+      "loss": 2.1362,
+      "step": 8110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.762813951537582e-07,
+      "loss": 2.1482,
+      "step": 8111
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.754682053996291e-07,
+      "loss": 2.1861,
+      "step": 8112
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.746555547612781e-07,
+      "loss": 2.1646,
+      "step": 8113
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.738434433377244e-07,
+      "loss": 2.1688,
+      "step": 8114
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.730318712279203e-07,
+      "loss": 2.1858,
+      "step": 8115
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.722208385307559e-07,
+      "loss": 2.1224,
+      "step": 8116
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.714103453450498e-07,
+      "loss": 2.2185,
+      "step": 8117
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.706003917695619e-07,
+      "loss": 2.1598,
+      "step": 8118
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.697909779029786e-07,
+      "loss": 2.2095,
+      "step": 8119
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.689821038439264e-07,
+      "loss": 2.1661,
+      "step": 8120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.681737696909656e-07,
+      "loss": 2.2032,
+      "step": 8121
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.673659755425859e-07,
+      "loss": 2.1261,
+      "step": 8122
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.665587214972173e-07,
+      "loss": 2.1669,
+      "step": 8123
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.657520076532208e-07,
+      "loss": 2.2313,
+      "step": 8124
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.649458341088915e-07,
+      "loss": 2.127,
+      "step": 8125
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.641402009624591e-07,
+      "loss": 2.2288,
+      "step": 8126
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.633351083120886e-07,
+      "loss": 2.2008,
+      "step": 8127
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.625305562558764e-07,
+      "loss": 2.2129,
+      "step": 8128
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.617265448918563e-07,
+      "loss": 2.1819,
+      "step": 8129
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.609230743179939e-07,
+      "loss": 2.152,
+      "step": 8130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.601201446321891e-07,
+      "loss": 2.19,
+      "step": 8131
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.593177559322776e-07,
+      "loss": 2.1498,
+      "step": 8132
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.585159083160274e-07,
+      "loss": 2.1258,
+      "step": 8133
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.577146018811419e-07,
+      "loss": 2.1689,
+      "step": 8134
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.569138367252553e-07,
+      "loss": 2.1703,
+      "step": 8135
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.561136129459432e-07,
+      "loss": 2.2197,
+      "step": 8136
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.553139306407062e-07,
+      "loss": 2.1238,
+      "step": 8137
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.545147899069836e-07,
+      "loss": 2.1086,
+      "step": 8138
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.537161908421512e-07,
+      "loss": 2.1646,
+      "step": 8139
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.529181335435124e-07,
+      "loss": 2.2176,
+      "step": 8140
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.521206181083111e-07,
+      "loss": 2.1646,
+      "step": 8141
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.51323644633719e-07,
+      "loss": 2.1277,
+      "step": 8142
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.505272132168471e-07,
+      "loss": 2.1668,
+      "step": 8143
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.497313239547374e-07,
+      "loss": 2.1206,
+      "step": 8144
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.489359769443675e-07,
+      "loss": 2.1419,
+      "step": 8145
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.48141172282648e-07,
+      "loss": 2.1702,
+      "step": 8146
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.473469100664208e-07,
+      "loss": 2.1861,
+      "step": 8147
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.46553190392467e-07,
+      "loss": 2.1648,
+      "step": 8148
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.457600133574987e-07,
+      "loss": 2.1577,
+      "step": 8149
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.449673790581611e-07,
+      "loss": 2.2156,
+      "step": 8150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.44175287591035e-07,
+      "loss": 2.2013,
+      "step": 8151
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.433837390526341e-07,
+      "loss": 2.2163,
+      "step": 8152
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.425927335394054e-07,
+      "loss": 2.1547,
+      "step": 8153
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.418022711477333e-07,
+      "loss": 2.1459,
+      "step": 8154
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.410123519739302e-07,
+      "loss": 2.1734,
+      "step": 8155
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.402229761142464e-07,
+      "loss": 2.1906,
+      "step": 8156
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.394341436648653e-07,
+      "loss": 2.146,
+      "step": 8157
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.386458547219026e-07,
+      "loss": 2.2141,
+      "step": 8158
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.378581093814112e-07,
+      "loss": 2.2308,
+      "step": 8159
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.370709077393721e-07,
+      "loss": 2.1766,
+      "step": 8160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.362842498917081e-07,
+      "loss": 2.144,
+      "step": 8161
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.354981359342659e-07,
+      "loss": 2.2147,
+      "step": 8162
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.347125659628344e-07,
+      "loss": 2.1826,
+      "step": 8163
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.339275400731331e-07,
+      "loss": 2.1248,
+      "step": 8164
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.331430583608122e-07,
+      "loss": 2.1803,
+      "step": 8165
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.323591209214612e-07,
+      "loss": 2.1541,
+      "step": 8166
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.315757278505995e-07,
+      "loss": 2.1681,
+      "step": 8167
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.307928792436812e-07,
+      "loss": 2.1513,
+      "step": 8168
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.300105751960943e-07,
+      "loss": 2.1621,
+      "step": 8169
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.292288158031595e-07,
+      "loss": 2.1882,
+      "step": 8170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.28447601160132e-07,
+      "loss": 2.179,
+      "step": 8171
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.276669313622013e-07,
+      "loss": 2.1802,
+      "step": 8172
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.268868065044886e-07,
+      "loss": 2.2291,
+      "step": 8173
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2610722668205e-07,
+      "loss": 2.1602,
+      "step": 8174
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.253281919898751e-07,
+      "loss": 2.1847,
+      "step": 8175
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.245497025228874e-07,
+      "loss": 2.0916,
+      "step": 8176
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.237717583759421e-07,
+      "loss": 2.1529,
+      "step": 8177
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.229943596438297e-07,
+      "loss": 2.1321,
+      "step": 8178
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.222175064212764e-07,
+      "loss": 2.1855,
+      "step": 8179
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.214411988029355e-07,
+      "loss": 2.1385,
+      "step": 8180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.206654368834002e-07,
+      "loss": 2.2055,
+      "step": 8181
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.198902207571955e-07,
+      "loss": 2.1858,
+      "step": 8182
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.191155505187756e-07,
+      "loss": 2.2007,
+      "step": 8183
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.183414262625364e-07,
+      "loss": 2.1448,
+      "step": 8184
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.175678480827972e-07,
+      "loss": 2.1732,
+      "step": 8185
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.167948160738206e-07,
+      "loss": 2.183,
+      "step": 8186
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.160223303297967e-07,
+      "loss": 2.1675,
+      "step": 8187
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.152503909448503e-07,
+      "loss": 2.2063,
+      "step": 8188
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.144789980130404e-07,
+      "loss": 2.1843,
+      "step": 8189
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.137081516283582e-07,
+      "loss": 2.1697,
+      "step": 8190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.129378518847295e-07,
+      "loss": 2.1136,
+      "step": 8191
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.121680988760125e-07,
+      "loss": 2.1137,
+      "step": 8192
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.113988926960001e-07,
+      "loss": 2.163,
+      "step": 8193
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.106302334384172e-07,
+      "loss": 2.1549,
+      "step": 8194
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.098621211969224e-07,
+      "loss": 2.1817,
+      "step": 8195
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.090945560651073e-07,
+      "loss": 2.1481,
+      "step": 8196
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.083275381364999e-07,
+      "loss": 2.2026,
+      "step": 8197
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.075610675045567e-07,
+      "loss": 2.1428,
+      "step": 8198
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0679514426267e-07,
+      "loss": 2.2069,
+      "step": 8199
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.06029768504166e-07,
+      "loss": 2.1991,
+      "step": 8200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.052649403223015e-07,
+      "loss": 2.198,
+      "step": 8201
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.045006598102725e-07,
+      "loss": 2.1128,
+      "step": 8202
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.037369270611997e-07,
+      "loss": 2.1919,
+      "step": 8203
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.029737421681446e-07,
+      "loss": 2.1306,
+      "step": 8204
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.022111052240985e-07,
+      "loss": 2.1736,
+      "step": 8205
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.014490163219854e-07,
+      "loss": 2.1336,
+      "step": 8206
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.006874755546654e-07,
+      "loss": 2.1626,
+      "step": 8207
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.99926483014927e-07,
+      "loss": 2.2039,
+      "step": 8208
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.991660387954967e-07,
+      "loss": 2.1616,
+      "step": 8209
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.984061429890324e-07,
+      "loss": 2.2343,
+      "step": 8210
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.976467956881254e-07,
+      "loss": 2.1586,
+      "step": 8211
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.968879969852985e-07,
+      "loss": 2.1608,
+      "step": 8212
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.961297469730097e-07,
+      "loss": 2.1997,
+      "step": 8213
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9537204574365e-07,
+      "loss": 2.1925,
+      "step": 8214
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.946148933895423e-07,
+      "loss": 2.1682,
+      "step": 8215
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.938582900029437e-07,
+      "loss": 2.1614,
+      "step": 8216
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.931022356760439e-07,
+      "loss": 2.1583,
+      "step": 8217
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.92346730500966e-07,
+      "loss": 2.1288,
+      "step": 8218
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.915917745697645e-07,
+      "loss": 2.1679,
+      "step": 8219
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.908373679744316e-07,
+      "loss": 2.1867,
+      "step": 8220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.900835108068863e-07,
+      "loss": 2.1674,
+      "step": 8221
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.893302031589864e-07,
+      "loss": 2.0928,
+      "step": 8222
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.885774451225178e-07,
+      "loss": 2.1842,
+      "step": 8223
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.878252367892033e-07,
+      "loss": 2.0749,
+      "step": 8224
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.87073578250698e-07,
+      "loss": 2.1578,
+      "step": 8225
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.863224695985858e-07,
+      "loss": 2.1675,
+      "step": 8226
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.855719109243917e-07,
+      "loss": 2.187,
+      "step": 8227
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.848219023195644e-07,
+      "loss": 2.1599,
+      "step": 8228
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.840724438754929e-07,
+      "loss": 2.2075,
+      "step": 8229
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.833235356834959e-07,
+      "loss": 2.2029,
+      "step": 8230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.825751778348259e-07,
+      "loss": 2.2087,
+      "step": 8231
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.818273704206678e-07,
+      "loss": 2.1465,
+      "step": 8232
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.810801135321391e-07,
+      "loss": 2.2008,
+      "step": 8233
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.803334072602917e-07,
+      "loss": 2.139,
+      "step": 8234
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.795872516961087e-07,
+      "loss": 2.1569,
+      "step": 8235
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.788416469305068e-07,
+      "loss": 2.1268,
+      "step": 8236
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.780965930543369e-07,
+      "loss": 2.1236,
+      "step": 8237
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.773520901583801e-07,
+      "loss": 2.1737,
+      "step": 8238
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.766081383333521e-07,
+      "loss": 2.1199,
+      "step": 8239
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.758647376699033e-07,
+      "loss": 2.1627,
+      "step": 8240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.751218882586106e-07,
+      "loss": 2.1528,
+      "step": 8241
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.743795901899928e-07,
+      "loss": 2.2014,
+      "step": 8242
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7363784355449303e-07,
+      "loss": 2.2384,
+      "step": 8243
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.728966484424913e-07,
+      "loss": 2.1629,
+      "step": 8244
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.72156004944303e-07,
+      "loss": 2.1843,
+      "step": 8245
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.714159131501689e-07,
+      "loss": 2.1478,
+      "step": 8246
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7067637315027005e-07,
+      "loss": 2.1241,
+      "step": 8247
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.699373850347161e-07,
+      "loss": 2.1646,
+      "step": 8248
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.691989488935511e-07,
+      "loss": 2.1799,
+      "step": 8249
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6846106481675035e-07,
+      "loss": 2.1431,
+      "step": 8250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.677237328942236e-07,
+      "loss": 2.2079,
+      "step": 8251
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6698695321581165e-07,
+      "loss": 2.1542,
+      "step": 8252
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.662507258712895e-07,
+      "loss": 2.1996,
+      "step": 8253
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.655150509503642e-07,
+      "loss": 2.1515,
+      "step": 8254
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.647799285426757e-07,
+      "loss": 2.1394,
+      "step": 8255
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.640453587377958e-07,
+      "loss": 2.2073,
+      "step": 8256
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6331134162522994e-07,
+      "loss": 2.1358,
+      "step": 8257
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.625778772944156e-07,
+      "loss": 2.1865,
+      "step": 8258
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6184496583472293e-07,
+      "loss": 2.1767,
+      "step": 8259
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6111260733545714e-07,
+      "loss": 2.1947,
+      "step": 8260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6038080188585135e-07,
+      "loss": 2.1822,
+      "step": 8261
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5964954957507414e-07,
+      "loss": 2.1601,
+      "step": 8262
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5891885049222815e-07,
+      "loss": 2.1902,
+      "step": 8263
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.581887047263445e-07,
+      "loss": 2.1766,
+      "step": 8264
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5745911236639186e-07,
+      "loss": 2.167,
+      "step": 8265
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.567300735012653e-07,
+      "loss": 2.1932,
+      "step": 8266
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5600158821979933e-07,
+      "loss": 2.1329,
+      "step": 8267
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.552736566107563e-07,
+      "loss": 2.1291,
+      "step": 8268
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5454627876283295e-07,
+      "loss": 2.1573,
+      "step": 8269
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.538194547646574e-07,
+      "loss": 2.1342,
+      "step": 8270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5309318470479144e-07,
+      "loss": 2.189,
+      "step": 8271
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.523674686717283e-07,
+      "loss": 2.1731,
+      "step": 8272
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.51642306753895e-07,
+      "loss": 2.1331,
+      "step": 8273
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5091769903964965e-07,
+      "loss": 2.111,
+      "step": 8274
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.501936456172845e-07,
+      "loss": 2.1644,
+      "step": 8275
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.494701465750217e-07,
+      "loss": 2.215,
+      "step": 8276
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.487472020010181e-07,
+      "loss": 2.1555,
+      "step": 8277
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.480248119833641e-07,
+      "loss": 2.2016,
+      "step": 8278
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.473029766100784e-07,
+      "loss": 2.1527,
+      "step": 8279
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4658169596911493e-07,
+      "loss": 2.1623,
+      "step": 8280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4586097014836017e-07,
+      "loss": 2.1608,
+      "step": 8281
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4514079923563103e-07,
+      "loss": 2.1459,
+      "step": 8282
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.444211833186807e-07,
+      "loss": 2.1852,
+      "step": 8283
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4370212248518895e-07,
+      "loss": 2.2263,
+      "step": 8284
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4298361682277355e-07,
+      "loss": 2.2245,
+      "step": 8285
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4226566641898173e-07,
+      "loss": 2.1756,
+      "step": 8286
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.415482713612934e-07,
+      "loss": 2.164,
+      "step": 8287
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4083143173712207e-07,
+      "loss": 2.1991,
+      "step": 8288
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.401151476338095e-07,
+      "loss": 2.1837,
+      "step": 8289
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3939941913863525e-07,
+      "loss": 2.1871,
+      "step": 8290
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.38684246338808e-07,
+      "loss": 2.1425,
+      "step": 8291
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.379696293214697e-07,
+      "loss": 2.2023,
+      "step": 8292
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.372555681736934e-07,
+      "loss": 2.1309,
+      "step": 8293
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3654206298248625e-07,
+      "loss": 2.2129,
+      "step": 8294
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3582911383478646e-07,
+      "loss": 2.191,
+      "step": 8295
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3511672081746393e-07,
+      "loss": 2.2008,
+      "step": 8296
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.344048840173226e-07,
+      "loss": 2.1885,
+      "step": 8297
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.33693603521097e-07,
+      "loss": 2.1441,
+      "step": 8298
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.32982879415455e-07,
+      "loss": 2.1471,
+      "step": 8299
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.322727117869951e-07,
+      "loss": 2.1802,
+      "step": 8300
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.315631007222515e-07,
+      "loss": 2.1569,
+      "step": 8301
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.308540463076849e-07,
+      "loss": 2.1485,
+      "step": 8302
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.301455486296946e-07,
+      "loss": 2.1641,
+      "step": 8303
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.29437607774606e-07,
+      "loss": 2.1682,
+      "step": 8304
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2873022382868115e-07,
+      "loss": 2.1476,
+      "step": 8305
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.280233968781139e-07,
+      "loss": 2.179,
+      "step": 8306
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.273171270090254e-07,
+      "loss": 2.0849,
+      "step": 8307
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.266114143074751e-07,
+      "loss": 2.154,
+      "step": 8308
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2590625885945205e-07,
+      "loss": 2.1365,
+      "step": 8309
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2520166075087635e-07,
+      "loss": 2.2062,
+      "step": 8310
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.24497620067601e-07,
+      "loss": 2.1359,
+      "step": 8311
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.237941368954124e-07,
+      "loss": 2.1958,
+      "step": 8312
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2309121132002695e-07,
+      "loss": 2.1112,
+      "step": 8313
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2238884342709397e-07,
+      "loss": 2.2013,
+      "step": 8314
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2168703330219494e-07,
+      "loss": 2.1851,
+      "step": 8315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2098578103084376e-07,
+      "loss": 2.2066,
+      "step": 8316
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.202850866984853e-07,
+      "loss": 2.1814,
+      "step": 8317
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.195849503904975e-07,
+      "loss": 2.1193,
+      "step": 8318
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.188853721921893e-07,
+      "loss": 2.2196,
+      "step": 8319
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1818635218880186e-07,
+      "loss": 2.1922,
+      "step": 8320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1748789046551055e-07,
+      "loss": 2.2102,
+      "step": 8321
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1678998710741936e-07,
+      "loss": 2.1583,
+      "step": 8322
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.160926421995648e-07,
+      "loss": 2.1272,
+      "step": 8323
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.153958558269189e-07,
+      "loss": 2.23,
+      "step": 8324
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.146996280743798e-07,
+      "loss": 2.2023,
+      "step": 8325
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.140039590267836e-07,
+      "loss": 2.1803,
+      "step": 8326
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.13308848768893e-07,
+      "loss": 2.2315,
+      "step": 8327
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1261429738540694e-07,
+      "loss": 2.1241,
+      "step": 8328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.119203049609538e-07,
+      "loss": 2.2123,
+      "step": 8329
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.112268715800943e-07,
+      "loss": 2.1489,
+      "step": 8330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.10533997327322e-07,
+      "loss": 2.1368,
+      "step": 8331
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0984168228705934e-07,
+      "loss": 2.1898,
+      "step": 8332
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.091499265436649e-07,
+      "loss": 2.192,
+      "step": 8333
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.084587301814269e-07,
+      "loss": 2.1786,
+      "step": 8334
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0776809328456455e-07,
+      "loss": 2.2676,
+      "step": 8335
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0707801593723006e-07,
+      "loss": 2.1628,
+      "step": 8336
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.063884982235078e-07,
+      "loss": 2.1843,
+      "step": 8337
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.056995402274122e-07,
+      "loss": 2.1674,
+      "step": 8338
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0501114203289395e-07,
+      "loss": 2.2152,
+      "step": 8339
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.043233037238281e-07,
+      "loss": 2.1576,
+      "step": 8340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0363602538402823e-07,
+      "loss": 2.1605,
+      "step": 8341
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.029493070972362e-07,
+      "loss": 2.2492,
+      "step": 8342
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.022631489471257e-07,
+      "loss": 2.2184,
+      "step": 8343
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0157755101730645e-07,
+      "loss": 2.181,
+      "step": 8344
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0089251339131164e-07,
+      "loss": 2.1918,
+      "step": 8345
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.002080361526156e-07,
+      "loss": 2.1244,
+      "step": 8346
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9952411938461557e-07,
+      "loss": 2.1309,
+      "step": 8347
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9884076317064813e-07,
+      "loss": 2.2099,
+      "step": 8348
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9815796759397783e-07,
+      "loss": 2.1389,
+      "step": 8349
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9747573273779816e-07,
+      "loss": 2.0965,
+      "step": 8350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.967940586852409e-07,
+      "loss": 2.164,
+      "step": 8351
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.961129455193641e-07,
+      "loss": 2.1986,
+      "step": 8352
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.954323933231602e-07,
+      "loss": 2.173,
+      "step": 8353
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.947524021795518e-07,
+      "loss": 2.1614,
+      "step": 8354
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9407297217139427e-07,
+      "loss": 2.1243,
+      "step": 8355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9339410338147363e-07,
+      "loss": 2.094,
+      "step": 8356
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9271579589250817e-07,
+      "loss": 2.1751,
+      "step": 8357
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.920380497871473e-07,
+      "loss": 2.1587,
+      "step": 8358
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.913608651479733e-07,
+      "loss": 2.1099,
+      "step": 8359
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.90684242057498e-07,
+      "loss": 2.1984,
+      "step": 8360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9000818059816593e-07,
+      "loss": 2.1535,
+      "step": 8361
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.89332680852354e-07,
+      "loss": 2.1637,
+      "step": 8362
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.88657742902368e-07,
+      "loss": 2.1495,
+      "step": 8363
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.879833668304506e-07,
+      "loss": 2.193,
+      "step": 8364
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8730955271876813e-07,
+      "loss": 2.166,
+      "step": 8365
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.866363006494256e-07,
+      "loss": 2.2569,
+      "step": 8366
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.859636107044573e-07,
+      "loss": 2.2007,
+      "step": 8367
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.85291482965825e-07,
+      "loss": 2.1103,
+      "step": 8368
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.846199175154297e-07,
+      "loss": 2.1675,
+      "step": 8369
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8394891443509554e-07,
+      "loss": 2.1263,
+      "step": 8370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.832784738065853e-07,
+      "loss": 2.149,
+      "step": 8371
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8260859571158883e-07,
+      "loss": 2.1545,
+      "step": 8372
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8193928023172897e-07,
+      "loss": 2.1545,
+      "step": 8373
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.812705274485595e-07,
+      "loss": 2.1729,
+      "step": 8374
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8060233744356634e-07,
+      "loss": 2.1124,
+      "step": 8375
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7993471029816653e-07,
+      "loss": 2.1364,
+      "step": 8376
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.792676460937078e-07,
+      "loss": 2.0855,
+      "step": 8377
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7860114491147017e-07,
+      "loss": 2.1456,
+      "step": 8378
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.779352068326653e-07,
+      "loss": 2.1795,
+      "step": 8379
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.772698319384349e-07,
+      "loss": 2.1748,
+      "step": 8380
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7660502030985203e-07,
+      "loss": 2.2279,
+      "step": 8381
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.759407720279257e-07,
+      "loss": 2.1664,
+      "step": 8382
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.752770871735878e-07,
+      "loss": 2.1219,
+      "step": 8383
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7461396582771035e-07,
+      "loss": 2.1188,
+      "step": 8384
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.739514080710899e-07,
+      "loss": 2.1827,
+      "step": 8385
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.732894139844578e-07,
+      "loss": 2.0904,
+      "step": 8386
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7262798364847753e-07,
+      "loss": 2.2256,
+      "step": 8387
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7196711714373947e-07,
+      "loss": 2.1626,
+      "step": 8388
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.713068145507709e-07,
+      "loss": 2.1896,
+      "step": 8389
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7064707595002636e-07,
+      "loss": 2.2083,
+      "step": 8390
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6998790142189324e-07,
+      "loss": 2.2032,
+      "step": 8391
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.693292910466906e-07,
+      "loss": 2.1728,
+      "step": 8392
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6867124490466697e-07,
+      "loss": 2.2279,
+      "step": 8393
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.680137630760039e-07,
+      "loss": 2.1553,
+      "step": 8394
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6735684564081385e-07,
+      "loss": 2.1678,
+      "step": 8395
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6670049267913954e-07,
+      "loss": 2.1482,
+      "step": 8396
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6604470427095587e-07,
+      "loss": 2.2562,
+      "step": 8397
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6538948049616886e-07,
+      "loss": 2.1533,
+      "step": 8398
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6473482143461523e-07,
+      "loss": 2.1269,
+      "step": 8399
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6408072716606346e-07,
+      "loss": 2.1444,
+      "step": 8400
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6342719777021194e-07,
+      "loss": 2.1722,
+      "step": 8401
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.627742333266937e-07,
+      "loss": 2.1805,
+      "step": 8402
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.621218339150684e-07,
+      "loss": 2.2208,
+      "step": 8403
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.614699996148285e-07,
+      "loss": 2.0818,
+      "step": 8404
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.608187305054006e-07,
+      "loss": 2.1931,
+      "step": 8405
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.601680266661367e-07,
+      "loss": 2.188,
+      "step": 8406
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5951788817632615e-07,
+      "loss": 2.1938,
+      "step": 8407
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5886831511518336e-07,
+      "loss": 2.2031,
+      "step": 8408
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5821930756185894e-07,
+      "loss": 2.2434,
+      "step": 8409
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.575708655954324e-07,
+      "loss": 2.2299,
+      "step": 8410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.569229892949133e-07,
+      "loss": 2.2204,
+      "step": 8411
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.562756787392452e-07,
+      "loss": 2.2117,
+      "step": 8412
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.556289340072977e-07,
+      "loss": 2.1361,
+      "step": 8413
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5498275517787783e-07,
+      "loss": 2.1498,
+      "step": 8414
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5433714232971927e-07,
+      "loss": 2.2066,
+      "step": 8415
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5369209554148854e-07,
+      "loss": 2.1294,
+      "step": 8416
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.530476148917816e-07,
+      "loss": 2.1344,
+      "step": 8417
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.524037004591274e-07,
+      "loss": 2.1006,
+      "step": 8418
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5176035232198367e-07,
+      "loss": 2.1759,
+      "step": 8419
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.511175705587433e-07,
+      "loss": 2.1243,
+      "step": 8420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5047535524772467e-07,
+      "loss": 2.1882,
+      "step": 8421
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.498337064671803e-07,
+      "loss": 2.1707,
+      "step": 8422
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.491926242952931e-07,
+      "loss": 2.209,
+      "step": 8423
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4855210881017675e-07,
+      "loss": 2.1988,
+      "step": 8424
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.479121600898777e-07,
+      "loss": 2.1941,
+      "step": 8425
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.472727782123697e-07,
+      "loss": 2.1996,
+      "step": 8426
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4663396325556154e-07,
+      "loss": 2.182,
+      "step": 8427
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.459957152972887e-07,
+      "loss": 2.1297,
+      "step": 8428
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4535803441532125e-07,
+      "loss": 2.1341,
+      "step": 8429
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4472092068735917e-07,
+      "loss": 2.1723,
+      "step": 8430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4408437419103047e-07,
+      "loss": 2.1473,
+      "step": 8431
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.434483950038986e-07,
+      "loss": 2.1609,
+      "step": 8432
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.428129832034549e-07,
+      "loss": 2.1574,
+      "step": 8433
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.421781388671225e-07,
+      "loss": 2.2089,
+      "step": 8434
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.415438620722555e-07,
+      "loss": 2.1944,
+      "step": 8435
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.409101528961378e-07,
+      "loss": 2.1523,
+      "step": 8436
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.402770114159859e-07,
+      "loss": 2.1869,
+      "step": 8437
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.396444377089453e-07,
+      "loss": 2.1503,
+      "step": 8438
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3901243185209375e-07,
+      "loss": 2.1747,
+      "step": 8439
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3838099392243915e-07,
+      "loss": 2.1322,
+      "step": 8440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3775012399692055e-07,
+      "loss": 2.167,
+      "step": 8441
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.371198221524069e-07,
+      "loss": 2.2095,
+      "step": 8442
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.364900884656991e-07,
+      "loss": 2.2023,
+      "step": 8443
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.358609230135268e-07,
+      "loss": 2.1586,
+      "step": 8444
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.352323258725554e-07,
+      "loss": 2.2403,
+      "step": 8445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3460429711937417e-07,
+      "loss": 2.1173,
+      "step": 8446
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3397683683050685e-07,
+      "loss": 2.2556,
+      "step": 8447
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3334994508241013e-07,
+      "loss": 2.1647,
+      "step": 8448
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.327236219514657e-07,
+      "loss": 2.1352,
+      "step": 8449
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.320978675139919e-07,
+      "loss": 2.1141,
+      "step": 8450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3147268184623216e-07,
+      "loss": 2.1563,
+      "step": 8451
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3084806502436617e-07,
+      "loss": 2.1548,
+      "step": 8452
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3022401712450025e-07,
+      "loss": 2.1724,
+      "step": 8453
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2960053822267245e-07,
+      "loss": 2.083,
+      "step": 8454
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.289776283948526e-07,
+      "loss": 2.1389,
+      "step": 8455
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.283552877169399e-07,
+      "loss": 2.1414,
+      "step": 8456
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.277335162647649e-07,
+      "loss": 2.1512,
+      "step": 8457
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.271123141140886e-07,
+      "loss": 2.1885,
+      "step": 8458
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.264916813406022e-07,
+      "loss": 2.2007,
+      "step": 8459
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.258716180199278e-07,
+      "loss": 2.1906,
+      "step": 8460
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.252521242276191e-07,
+      "loss": 2.1622,
+      "step": 8461
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.246332000391583e-07,
+      "loss": 2.1578,
+      "step": 8462
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.240148455299619e-07,
+      "loss": 2.1775,
+      "step": 8463
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.233970607753717e-07,
+      "loss": 2.1989,
+      "step": 8464
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.227798458506637e-07,
+      "loss": 2.1681,
+      "step": 8465
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2216320083104434e-07,
+      "loss": 2.1955,
+      "step": 8466
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2154712579164913e-07,
+      "loss": 2.1791,
+      "step": 8467
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2093162080754634e-07,
+      "loss": 2.1306,
+      "step": 8468
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.20316685953731e-07,
+      "loss": 2.2284,
+      "step": 8469
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1970232130513365e-07,
+      "loss": 2.1615,
+      "step": 8470
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1908852693661116e-07,
+      "loss": 2.211,
+      "step": 8471
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1847530292295313e-07,
+      "loss": 2.1719,
+      "step": 8472
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1786264933887977e-07,
+      "loss": 2.1213,
+      "step": 8473
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.172505662590386e-07,
+      "loss": 2.1979,
+      "step": 8474
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.166390537580122e-07,
+      "loss": 2.1417,
+      "step": 8475
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.160281119103109e-07,
+      "loss": 2.1964,
+      "step": 8476
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1541774079037635e-07,
+      "loss": 2.1519,
+      "step": 8477
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.148079404725801e-07,
+      "loss": 2.1766,
+      "step": 8478
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1419871103122447e-07,
+      "loss": 2.1546,
+      "step": 8479
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.135900525405428e-07,
+      "loss": 2.1879,
+      "step": 8480
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1298196507469737e-07,
+      "loss": 2.153,
+      "step": 8481
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.123744487077829e-07,
+      "loss": 2.156,
+      "step": 8482
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1176750351382235e-07,
+      "loss": 2.1347,
+      "step": 8483
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1116112956677045e-07,
+      "loss": 2.2487,
+      "step": 8484
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.105553269405115e-07,
+      "loss": 2.2047,
+      "step": 8485
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0995009570886305e-07,
+      "loss": 2.1339,
+      "step": 8486
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.093454359455672e-07,
+      "loss": 2.1611,
+      "step": 8487
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0874134772430344e-07,
+      "loss": 2.2096,
+      "step": 8488
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.08137831118675e-07,
+      "loss": 2.1294,
+      "step": 8489
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0753488620222037e-07,
+      "loss": 2.1597,
+      "step": 8490
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.069325130484069e-07,
+      "loss": 2.1734,
+      "step": 8491
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0633071173062966e-07,
+      "loss": 2.1405,
+      "step": 8492
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.057294823222184e-07,
+      "loss": 2.1492,
+      "step": 8493
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.051288248964307e-07,
+      "loss": 2.1939,
+      "step": 8494
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0452873952645455e-07,
+      "loss": 2.1594,
+      "step": 8495
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0392922628540875e-07,
+      "loss": 2.1658,
+      "step": 8496
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0333028524634156e-07,
+      "loss": 2.2377,
+      "step": 8497
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.027319164822329e-07,
+      "loss": 2.1584,
+      "step": 8498
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0213412006599216e-07,
+      "loss": 2.1697,
+      "step": 8499
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.015368960704584e-07,
+      "loss": 2.2209,
+      "step": 8500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0094024456840176e-07,
+      "loss": 2.1858,
+      "step": 8501
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.003441656325229e-07,
+      "loss": 2.1771,
+      "step": 8502
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9974865933545207e-07,
+      "loss": 2.2367,
+      "step": 8503
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.99153725749749e-07,
+      "loss": 2.1518,
+      "step": 8504
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9855936494790516e-07,
+      "loss": 2.1967,
+      "step": 8505
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9796557700234317e-07,
+      "loss": 2.16,
+      "step": 8506
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9737236198541077e-07,
+      "loss": 2.144,
+      "step": 8507
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.967797199693928e-07,
+      "loss": 2.1717,
+      "step": 8508
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.961876510264999e-07,
+      "loss": 2.1604,
+      "step": 8509
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9559615522887275e-07,
+      "loss": 2.1638,
+      "step": 8510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9500523264858473e-07,
+      "loss": 2.1444,
+      "step": 8511
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9441488335763656e-07,
+      "loss": 2.1644,
+      "step": 8512
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.938251074279619e-07,
+      "loss": 2.1929,
+      "step": 8513
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9323590493142206e-07,
+      "loss": 2.132,
+      "step": 8514
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9264727593981024e-07,
+      "loss": 2.1965,
+      "step": 8515
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.920592205248496e-07,
+      "loss": 2.1408,
+      "step": 8516
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.914717387581917e-07,
+      "loss": 2.1988,
+      "step": 8517
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.908848307114198e-07,
+      "loss": 2.1542,
+      "step": 8518
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9029849645604735e-07,
+      "loss": 2.2214,
+      "step": 8519
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8971273606351656e-07,
+      "loss": 2.145,
+      "step": 8520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.891275496052015e-07,
+      "loss": 2.1955,
+      "step": 8521
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8854293715240455e-07,
+      "loss": 2.2174,
+      "step": 8522
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.879588987763593e-07,
+      "loss": 2.194,
+      "step": 8523
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8737543454822993e-07,
+      "loss": 2.1428,
+      "step": 8524
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.867925445391079e-07,
+      "loss": 2.1443,
+      "step": 8525
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.862102288200186e-07,
+      "loss": 2.1919,
+      "step": 8526
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.856284874619142e-07,
+      "loss": 2.178,
+      "step": 8527
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.850473205356774e-07,
+      "loss": 2.16,
+      "step": 8528
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.844667281121244e-07,
+      "loss": 2.1117,
+      "step": 8529
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.838867102619952e-07,
+      "loss": 2.1629,
+      "step": 8530
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.833072670559661e-07,
+      "loss": 2.2249,
+      "step": 8531
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8272839856463783e-07,
+      "loss": 2.1148,
+      "step": 8532
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.821501048585462e-07,
+      "loss": 2.2096,
+      "step": 8533
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.815723860081537e-07,
+      "loss": 2.1656,
+      "step": 8534
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8099524208385297e-07,
+      "loss": 2.1722,
+      "step": 8535
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.804186731559677e-07,
+      "loss": 2.1555,
+      "step": 8536
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.798426792947517e-07,
+      "loss": 2.1628,
+      "step": 8537
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.792672605703867e-07,
+      "loss": 2.1501,
+      "step": 8538
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.78692417052987e-07,
+      "loss": 2.1989,
+      "step": 8539
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7811814881259503e-07,
+      "loss": 2.1725,
+      "step": 8540
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.775444559191837e-07,
+      "loss": 2.1265,
+      "step": 8541
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7697133844265535e-07,
+      "loss": 2.1513,
+      "step": 8542
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.763987964528425e-07,
+      "loss": 2.1933,
+      "step": 8543
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.758268300195094e-07,
+      "loss": 2.1053,
+      "step": 8544
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.752554392123463e-07,
+      "loss": 2.1861,
+      "step": 8545
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.746846241009765e-07,
+      "loss": 2.1222,
+      "step": 8546
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7411438475495155e-07,
+      "loss": 2.1611,
+      "step": 8547
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.735447212437531e-07,
+      "loss": 2.1296,
+      "step": 8548
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.72975633636795e-07,
+      "loss": 2.1569,
+      "step": 8549
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.724071220034158e-07,
+      "loss": 2.1797,
+      "step": 8550
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7183918641288943e-07,
+      "loss": 2.1237,
+      "step": 8551
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.712718269344161e-07,
+      "loss": 2.1614,
+      "step": 8552
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.707050436371267e-07,
+      "loss": 2.1016,
+      "step": 8553
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.701388365900831e-07,
+      "loss": 2.1049,
+      "step": 8554
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6957320586227354e-07,
+      "loss": 2.162,
+      "step": 8555
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.690081515226206e-07,
+      "loss": 2.1212,
+      "step": 8556
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.684436736399737e-07,
+      "loss": 2.1898,
+      "step": 8557
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6787977228311336e-07,
+      "loss": 2.1794,
+      "step": 8558
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6731644752074846e-07,
+      "loss": 2.182,
+      "step": 8559
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6675369942151864e-07,
+      "loss": 2.2245,
+      "step": 8560
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6619152805399286e-07,
+      "loss": 2.1583,
+      "step": 8561
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.656299334866702e-07,
+      "loss": 2.1755,
+      "step": 8562
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.650689157879799e-07,
+      "loss": 2.1739,
+      "step": 8563
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6450847502627883e-07,
+      "loss": 2.2232,
+      "step": 8564
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.639486112698564e-07,
+      "loss": 2.1794,
+      "step": 8565
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6338932458692847e-07,
+      "loss": 2.1716,
+      "step": 8566
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6283061504564553e-07,
+      "loss": 2.2046,
+      "step": 8567
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.622724827140816e-07,
+      "loss": 2.1475,
+      "step": 8568
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.617149276602454e-07,
+      "loss": 2.2693,
+      "step": 8569
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.611579499520722e-07,
+      "loss": 2.1993,
+      "step": 8570
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.606015496574277e-07,
+      "loss": 2.2196,
+      "step": 8571
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.600457268441092e-07,
+      "loss": 2.1977,
+      "step": 8572
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.594904815798399e-07,
+      "loss": 2.1188,
+      "step": 8573
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.589358139322767e-07,
+      "loss": 2.1709,
+      "step": 8574
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.583817239690034e-07,
+      "loss": 2.1647,
+      "step": 8575
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.578282117575343e-07,
+      "loss": 2.1503,
+      "step": 8576
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5727527736531256e-07,
+      "loss": 2.2012,
+      "step": 8577
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5672292085971276e-07,
+      "loss": 2.2174,
+      "step": 8578
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.561711423080365e-07,
+      "loss": 2.2028,
+      "step": 8579
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.556199417775174e-07,
+      "loss": 2.1876,
+      "step": 8580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.550693193353171e-07,
+      "loss": 2.0944,
+      "step": 8581
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5451927504852757e-07,
+      "loss": 2.205,
+      "step": 8582
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.539698089841691e-07,
+      "loss": 2.2071,
+      "step": 8583
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.534209212091937e-07,
+      "loss": 2.2116,
+      "step": 8584
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5287261179048117e-07,
+      "loss": 2.2079,
+      "step": 8585
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.523248807948403e-07,
+      "loss": 2.1562,
+      "step": 8586
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5177772828901327e-07,
+      "loss": 2.2197,
+      "step": 8587
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5123115433966615e-07,
+      "loss": 2.0812,
+      "step": 8588
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5068515901339794e-07,
+      "loss": 2.1291,
+      "step": 8589
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5013974237673824e-07,
+      "loss": 2.1165,
+      "step": 8590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.49594904496141e-07,
+      "loss": 2.1634,
+      "step": 8591
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4905064543799706e-07,
+      "loss": 2.2351,
+      "step": 8592
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.485069652686195e-07,
+      "loss": 2.1266,
+      "step": 8593
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.479638640542564e-07,
+      "loss": 2.148,
+      "step": 8594
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.474213418610816e-07,
+      "loss": 2.1888,
+      "step": 8595
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4687939875519984e-07,
+      "loss": 2.1546,
+      "step": 8596
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.463380348026467e-07,
+      "loss": 2.0933,
+      "step": 8597
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.457972500693834e-07,
+      "loss": 2.1374,
+      "step": 8598
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4525704462130485e-07,
+      "loss": 2.2047,
+      "step": 8599
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.447174185242324e-07,
+      "loss": 2.1056,
+      "step": 8600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4417837184391833e-07,
+      "loss": 2.1364,
+      "step": 8601
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4363990464604357e-07,
+      "loss": 2.1448,
+      "step": 8602
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4310201699621896e-07,
+      "loss": 2.224,
+      "step": 8603
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4256470895998363e-07,
+      "loss": 2.1698,
+      "step": 8604
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.420279806028092e-07,
+      "loss": 2.1864,
+      "step": 8605
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.414918319900922e-07,
+      "loss": 2.1924,
+      "step": 8606
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4095626318716146e-07,
+      "loss": 2.1363,
+      "step": 8607
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.404212742592743e-07,
+      "loss": 2.1707,
+      "step": 8608
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3988686527161686e-07,
+      "loss": 2.1563,
+      "step": 8609
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3935303628930705e-07,
+      "loss": 2.153,
+      "step": 8610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.388197873773879e-07,
+      "loss": 2.2012,
+      "step": 8611
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3828711860083676e-07,
+      "loss": 2.1397,
+      "step": 8612
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3775503002455514e-07,
+      "loss": 2.1742,
+      "step": 8613
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3722352171337836e-07,
+      "loss": 2.1602,
+      "step": 8614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.366925937320691e-07,
+      "loss": 2.2054,
+      "step": 8615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.361622461453178e-07,
+      "loss": 2.1761,
+      "step": 8616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3563247901774666e-07,
+      "loss": 2.1844,
+      "step": 8617
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.351032924139063e-07,
+      "loss": 2.1171,
+      "step": 8618
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3457468639827563e-07,
+      "loss": 2.1778,
+      "step": 8619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3404666103526542e-07,
+      "loss": 2.1848,
+      "step": 8620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3351921638921193e-07,
+      "loss": 2.1783,
+      "step": 8621
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3299235252438434e-07,
+      "loss": 2.1474,
+      "step": 8622
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3246606950497851e-07,
+      "loss": 2.1822,
+      "step": 8623
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.319403673951204e-07,
+      "loss": 2.1474,
+      "step": 8624
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.314152462588659e-07,
+      "loss": 2.1756,
+      "step": 8625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3089070616019838e-07,
+      "loss": 2.1854,
+      "step": 8626
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3036674716303277e-07,
+      "loss": 2.0743,
+      "step": 8627
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2984336933121076e-07,
+      "loss": 2.1265,
+      "step": 8628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2932057272850416e-07,
+      "loss": 2.1618,
+      "step": 8629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.287983574186159e-07,
+      "loss": 2.2097,
+      "step": 8630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2827672346517448e-07,
+      "loss": 2.1394,
+      "step": 8631
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2775567093174022e-07,
+      "loss": 2.153,
+      "step": 8632
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2723519988180232e-07,
+      "loss": 2.1813,
+      "step": 8633
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2671531037877724e-07,
+      "loss": 2.218,
+      "step": 8634
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2619600248601327e-07,
+      "loss": 2.1837,
+      "step": 8635
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2567727626678527e-07,
+      "loss": 2.1706,
+      "step": 8636
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2515913178429937e-07,
+      "loss": 2.1475,
+      "step": 8637
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2464156910168954e-07,
+      "loss": 2.173,
+      "step": 8638
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2412458828201977e-07,
+      "loss": 2.1932,
+      "step": 8639
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2360818938828189e-07,
+      "loss": 2.1663,
+      "step": 8640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2309237248339776e-07,
+      "loss": 2.1157,
+      "step": 8641
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2257713763021826e-07,
+      "loss": 2.1839,
+      "step": 8642
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.220624848915226e-07,
+      "loss": 2.1477,
+      "step": 8643
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2154841433002062e-07,
+      "loss": 2.1409,
+      "step": 8644
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.210349260083494e-07,
+      "loss": 2.1204,
+      "step": 8645
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2052201998907673e-07,
+      "loss": 2.1507,
+      "step": 8646
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.200096963346976e-07,
+      "loss": 2.2053,
+      "step": 8647
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1949795510763872e-07,
+      "loss": 2.1401,
+      "step": 8648
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.189867963702519e-07,
+      "loss": 2.1434,
+      "step": 8649
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1847622018482283e-07,
+      "loss": 2.1278,
+      "step": 8650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1796622661356238e-07,
+      "loss": 2.1346,
+      "step": 8651
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.174568157186102e-07,
+      "loss": 2.1412,
+      "step": 8652
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1694798756204005e-07,
+      "loss": 2.1103,
+      "step": 8653
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1643974220584729e-07,
+      "loss": 2.1624,
+      "step": 8654
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1593207971196296e-07,
+      "loss": 2.1734,
+      "step": 8655
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.154250001422431e-07,
+      "loss": 2.1872,
+      "step": 8656
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1491850355847332e-07,
+      "loss": 2.0962,
+      "step": 8657
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1441259002236924e-07,
+      "loss": 2.1992,
+      "step": 8658
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1390725959557546e-07,
+      "loss": 2.173,
+      "step": 8659
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.134025123396638e-07,
+      "loss": 2.1193,
+      "step": 8660
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1289834831613675e-07,
+      "loss": 2.2205,
+      "step": 8661
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.123947675864252e-07,
+      "loss": 2.1569,
+      "step": 8662
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1189177021188888e-07,
+      "loss": 2.1333,
+      "step": 8663
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1138935625381663e-07,
+      "loss": 2.172,
+      "step": 8664
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1088752577342607e-07,
+      "loss": 2.1852,
+      "step": 8665
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.103862788318628e-07,
+      "loss": 2.1931,
+      "step": 8666
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.098856154902029e-07,
+      "loss": 2.1162,
+      "step": 8667
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0938553580945208e-07,
+      "loss": 2.1371,
+      "step": 8668
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0888603985054156e-07,
+      "loss": 2.2137,
+      "step": 8669
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.083871276743338e-07,
+      "loss": 2.1248,
+      "step": 8670
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0788879934162064e-07,
+      "loss": 2.2041,
+      "step": 8671
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0739105491312028e-07,
+      "loss": 2.156,
+      "step": 8672
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.068938944494836e-07,
+      "loss": 2.2173,
+      "step": 8673
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0639731801128603e-07,
+      "loss": 2.1394,
+      "step": 8674
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0590132565903475e-07,
+      "loss": 2.185,
+      "step": 8675
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.054059174531653e-07,
+      "loss": 2.2456,
+      "step": 8676
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0491109345404102e-07,
+      "loss": 2.1588,
+      "step": 8677
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0441685372195487e-07,
+      "loss": 2.1418,
+      "step": 8678
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.039231983171286e-07,
+      "loss": 2.1972,
+      "step": 8679
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0343012729971244e-07,
+      "loss": 2.1204,
+      "step": 8680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0293764072978618e-07,
+      "loss": 2.1347,
+      "step": 8681
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0244573866735673e-07,
+      "loss": 2.2087,
+      "step": 8682
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0195442117236176e-07,
+      "loss": 2.16,
+      "step": 8683
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0146368830466668e-07,
+      "loss": 2.2103,
+      "step": 8684
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0097354012406535e-07,
+      "loss": 2.1908,
+      "step": 8685
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0048397669028164e-07,
+      "loss": 2.1629,
+      "step": 8686
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9999499806296674e-07,
+      "loss": 2.1817,
+      "step": 8687
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.995066043017013e-07,
+      "loss": 2.155,
+      "step": 8688
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.99018795465995e-07,
+      "loss": 2.1836,
+      "step": 8689
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9853157161528468e-07,
+      "loss": 2.1776,
+      "step": 8690
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.98044932808939e-07,
+      "loss": 2.1538,
+      "step": 8691
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9755887910625103e-07,
+      "loss": 2.1466,
+      "step": 8692
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9707341056644737e-07,
+      "loss": 2.1834,
+      "step": 8693
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9658852724868005e-07,
+      "loss": 2.2116,
+      "step": 8694
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.961042292120291e-07,
+      "loss": 2.172,
+      "step": 8695
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9562051651550784e-07,
+      "loss": 2.16,
+      "step": 8696
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9513738921805192e-07,
+      "loss": 2.2113,
+      "step": 8697
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9465484737853092e-07,
+      "loss": 2.1069,
+      "step": 8698
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9417289105574054e-07,
+      "loss": 2.145,
+      "step": 8699
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9369152030840553e-07,
+      "loss": 2.2021,
+      "step": 8700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9321073519518007e-07,
+      "loss": 2.1824,
+      "step": 8701
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.927305357746462e-07,
+      "loss": 2.0748,
+      "step": 8702
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9225092210531425e-07,
+      "loss": 2.1979,
+      "step": 8703
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.917718942456237e-07,
+      "loss": 2.173,
+      "step": 8704
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9129345225394335e-07,
+      "loss": 2.1557,
+      "step": 8705
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9081559618856938e-07,
+      "loss": 2.1356,
+      "step": 8706
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.903383261077274e-07,
+      "loss": 2.2042,
+      "step": 8707
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8986164206957037e-07,
+      "loss": 2.2275,
+      "step": 8708
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8938554413218292e-07,
+      "loss": 2.1849,
+      "step": 8709
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8891003235357307e-07,
+      "loss": 2.1602,
+      "step": 8710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8843510679168341e-07,
+      "loss": 2.1966,
+      "step": 8711
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8796076750438096e-07,
+      "loss": 2.2113,
+      "step": 8712
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.874870145494617e-07,
+      "loss": 2.1282,
+      "step": 8713
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8701384798465284e-07,
+      "loss": 2.1058,
+      "step": 8714
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8654126786760597e-07,
+      "loss": 2.1914,
+      "step": 8715
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8606927425590616e-07,
+      "loss": 2.1285,
+      "step": 8716
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8559786720706185e-07,
+      "loss": 2.1938,
+      "step": 8717
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8512704677851489e-07,
+      "loss": 2.1515,
+      "step": 8718
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.846568130276316e-07,
+      "loss": 2.1893,
+      "step": 8719
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.841871660117095e-07,
+      "loss": 2.1902,
+      "step": 8720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8371810578797277e-07,
+      "loss": 2.198,
+      "step": 8721
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.832496324135763e-07,
+      "loss": 2.1597,
+      "step": 8722
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.827817459456005e-07,
+      "loss": 2.1107,
+      "step": 8723
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8231444644105755e-07,
+      "loss": 2.1857,
+      "step": 8724
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8184773395688527e-07,
+      "loss": 2.2336,
+      "step": 8725
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8138160854995145e-07,
+      "loss": 2.1733,
+      "step": 8726
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8091607027705293e-07,
+      "loss": 2.1373,
+      "step": 8727
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.804511191949121e-07,
+      "loss": 2.1541,
+      "step": 8728
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7998675536018474e-07,
+      "loss": 2.1999,
+      "step": 8729
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7952297882945e-07,
+      "loss": 2.1674,
+      "step": 8730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7905978965921778e-07,
+      "loss": 2.1722,
+      "step": 8731
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.785971879059273e-07,
+      "loss": 2.1398,
+      "step": 8732
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7813517362594347e-07,
+      "loss": 2.2184,
+      "step": 8733
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7767374687556405e-07,
+      "loss": 2.1101,
+      "step": 8734
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7721290771100964e-07,
+      "loss": 2.193,
+      "step": 8735
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7675265618843361e-07,
+      "loss": 2.1283,
+      "step": 8736
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7629299236391616e-07,
+      "loss": 2.0733,
+      "step": 8737
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.758339162934658e-07,
+      "loss": 2.1447,
+      "step": 8738
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7537542803302e-07,
+      "loss": 2.1523,
+      "step": 8739
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7491752763844294e-07,
+      "loss": 2.1798,
+      "step": 8740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.744602151655289e-07,
+      "loss": 2.1304,
+      "step": 8741
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.740034906700011e-07,
+      "loss": 2.1153,
+      "step": 8742
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7354735420750835e-07,
+      "loss": 2.1913,
+      "step": 8743
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7309180583363062e-07,
+      "loss": 2.1504,
+      "step": 8744
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7263684560387518e-07,
+      "loss": 2.2106,
+      "step": 8745
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7218247357367656e-07,
+      "loss": 2.1823,
+      "step": 8746
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.717286897983994e-07,
+      "loss": 2.1457,
+      "step": 8747
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7127549433333557e-07,
+      "loss": 2.193,
+      "step": 8748
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7082288723370587e-07,
+      "loss": 2.2006,
+      "step": 8749
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7037086855465902e-07,
+      "loss": 2.1045,
+      "step": 8750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.699194383512709e-07,
+      "loss": 2.1803,
+      "step": 8751
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6946859667854977e-07,
+      "loss": 2.2015,
+      "step": 8752
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.690183435914261e-07,
+      "loss": 2.147,
+      "step": 8753
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6856867914476492e-07,
+      "loss": 2.1941,
+      "step": 8754
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6811960339335298e-07,
+      "loss": 2.205,
+      "step": 8755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6767111639191202e-07,
+      "loss": 2.1558,
+      "step": 8756
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.672232181950878e-07,
+      "loss": 2.0902,
+      "step": 8757
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6677590885745388e-07,
+      "loss": 2.1879,
+      "step": 8758
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6632918843351554e-07,
+      "loss": 2.1438,
+      "step": 8759
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6588305697770313e-07,
+      "loss": 2.1159,
+      "step": 8760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6543751454437708e-07,
+      "loss": 2.1429,
+      "step": 8761
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6499256118782503e-07,
+      "loss": 2.1135,
+      "step": 8762
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.645481969622631e-07,
+      "loss": 2.2124,
+      "step": 8763
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6410442192183574e-07,
+      "loss": 2.2335,
+      "step": 8764
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6366123612061636e-07,
+      "loss": 2.2105,
+      "step": 8765
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6321863961260452e-07,
+      "loss": 2.199,
+      "step": 8766
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6277663245173047e-07,
+      "loss": 2.1977,
+      "step": 8767
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6233521469185054e-07,
+      "loss": 2.2137,
+      "step": 8768
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.618943863867506e-07,
+      "loss": 2.1388,
+      "step": 8769
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6145414759014433e-07,
+      "loss": 2.1717,
+      "step": 8770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6101449835567273e-07,
+      "loss": 2.167,
+      "step": 8771
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6057543873690685e-07,
+      "loss": 2.1484,
+      "step": 8772
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6013696878734385e-07,
+      "loss": 2.1959,
+      "step": 8773
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.596990885604105e-07,
+      "loss": 2.1688,
+      "step": 8774
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5926179810946185e-07,
+      "loss": 2.1492,
+      "step": 8775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5882509748777809e-07,
+      "loss": 2.1554,
+      "step": 8776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5838898674857273e-07,
+      "loss": 2.1519,
+      "step": 8777
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5795346594498162e-07,
+      "loss": 2.0997,
+      "step": 8778
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5751853513007454e-07,
+      "loss": 2.1562,
+      "step": 8779
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5708419435684463e-07,
+      "loss": 2.2123,
+      "step": 8780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5665044367821513e-07,
+      "loss": 2.1847,
+      "step": 8781
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5621728314703822e-07,
+      "loss": 2.166,
+      "step": 8782
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5578471281609274e-07,
+      "loss": 2.1494,
+      "step": 8783
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.553527327380855e-07,
+      "loss": 2.1732,
+      "step": 8784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5492134296565264e-07,
+      "loss": 2.1127,
+      "step": 8785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5449054355135718e-07,
+      "loss": 2.2048,
+      "step": 8786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5406033454769154e-07,
+      "loss": 2.1828,
+      "step": 8787
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5363071600707435e-07,
+      "loss": 2.1785,
+      "step": 8788
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.532016879818532e-07,
+      "loss": 2.1467,
+      "step": 8789
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5277325052430569e-07,
+      "loss": 2.1917,
+      "step": 8790
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5234540368663343e-07,
+      "loss": 2.1649,
+      "step": 8791
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5191814752097024e-07,
+      "loss": 2.1627,
+      "step": 8792
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5149148207937447e-07,
+      "loss": 2.1895,
+      "step": 8793
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5106540741383402e-07,
+      "loss": 2.1897,
+      "step": 8794
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5063992357626623e-07,
+      "loss": 2.1368,
+      "step": 8795
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.502150306185135e-07,
+      "loss": 2.0624,
+      "step": 8796
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.497907285923489e-07,
+      "loss": 2.193,
+      "step": 8797
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4936701754947104e-07,
+      "loss": 2.1554,
+      "step": 8798
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4894389754150862e-07,
+      "loss": 2.1692,
+      "step": 8799
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4852136862001766e-07,
+      "loss": 2.244,
+      "step": 8800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4809943083648194e-07,
+      "loss": 2.1174,
+      "step": 8801
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4767808424231312e-07,
+      "loss": 2.2021,
+      "step": 8802
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4725732888885126e-07,
+      "loss": 2.1669,
+      "step": 8803
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4683716482736364e-07,
+      "loss": 2.191,
+      "step": 8804
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4641759210904605e-07,
+      "loss": 2.1543,
+      "step": 8805
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.459986107850231e-07,
+      "loss": 2.1628,
+      "step": 8806
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4558022090634504e-07,
+      "loss": 2.1674,
+      "step": 8807
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4516242252399227e-07,
+      "loss": 2.1141,
+      "step": 8808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4474521568887178e-07,
+      "loss": 2.1956,
+      "step": 8809
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4432860045182019e-07,
+      "loss": 2.2208,
+      "step": 8810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4391257686359906e-07,
+      "loss": 2.1434,
+      "step": 8811
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4349714497490009e-07,
+      "loss": 2.1598,
+      "step": 8812
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4308230483634334e-07,
+      "loss": 2.156,
+      "step": 8813
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4266805649847392e-07,
+      "loss": 2.1614,
+      "step": 8814
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4225440001176983e-07,
+      "loss": 2.1335,
+      "step": 8815
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4184133542663014e-07,
+      "loss": 2.1272,
+      "step": 8816
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4142886279338852e-07,
+      "loss": 2.1488,
+      "step": 8817
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4101698216230254e-07,
+      "loss": 2.177,
+      "step": 8818
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4060569358355703e-07,
+      "loss": 2.082,
+      "step": 8819
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4019499710726913e-07,
+      "loss": 2.1313,
+      "step": 8820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3978489278347883e-07,
+      "loss": 2.1974,
+      "step": 8821
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3937538066215672e-07,
+      "loss": 2.1916,
+      "step": 8822
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3896646079320064e-07,
+      "loss": 2.1951,
+      "step": 8823
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.385581332264363e-07,
+      "loss": 2.1516,
+      "step": 8824
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3815039801161723e-07,
+      "loss": 2.133,
+      "step": 8825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3774325519842423e-07,
+      "loss": 2.1028,
+      "step": 8826
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.373367048364671e-07,
+      "loss": 2.2219,
+      "step": 8827
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3693074697528231e-07,
+      "loss": 2.1553,
+      "step": 8828
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3652538166433527e-07,
+      "loss": 2.1288,
+      "step": 8829
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3612060895301759e-07,
+      "loss": 2.1281,
+      "step": 8830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3571642889064984e-07,
+      "loss": 2.16,
+      "step": 8831
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3531284152647983e-07,
+      "loss": 2.1877,
+      "step": 8832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3490984690968488e-07,
+      "loss": 2.1504,
+      "step": 8833
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3450744508936687e-07,
+      "loss": 2.1605,
+      "step": 8834
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.341056361145593e-07,
+      "loss": 2.1805,
+      "step": 8835
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3370442003421913e-07,
+      "loss": 2.1448,
+      "step": 8836
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.333037968972345e-07,
+      "loss": 2.1482,
+      "step": 8837
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3290376675242022e-07,
+      "loss": 2.1681,
+      "step": 8838
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.325043296485179e-07,
+      "loss": 2.1707,
+      "step": 8839
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3210548563419857e-07,
+      "loss": 2.1753,
+      "step": 8840
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3170723475806003e-07,
+      "loss": 2.2538,
+      "step": 8841
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.313095770686279e-07,
+      "loss": 2.2157,
+      "step": 8842
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3091251261435568e-07,
+      "loss": 2.1263,
+      "step": 8843
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3051604144362407e-07,
+      "loss": 2.1491,
+      "step": 8844
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3012016360474223e-07,
+      "loss": 2.1622,
+      "step": 8845
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.29724879145946e-07,
+      "loss": 2.1225,
+      "step": 8846
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2933018811540078e-07,
+      "loss": 2.2494,
+      "step": 8847
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.289360905611975e-07,
+      "loss": 2.1576,
+      "step": 8848
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.285425865313561e-07,
+      "loss": 2.2066,
+      "step": 8849
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2814967607382433e-07,
+      "loss": 2.1491,
+      "step": 8850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2775735923647614e-07,
+      "loss": 2.0975,
+      "step": 8851
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2736563606711384e-07,
+      "loss": 2.1553,
+      "step": 8852
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2697450661347033e-07,
+      "loss": 2.2163,
+      "step": 8853
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2658397092320028e-07,
+      "loss": 2.0757,
+      "step": 8854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.261940290438912e-07,
+      "loss": 2.1173,
+      "step": 8855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.258046810230562e-07,
+      "loss": 2.2346,
+      "step": 8856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2541592690813508e-07,
+      "loss": 2.1606,
+      "step": 8857
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2502776674649776e-07,
+      "loss": 2.1722,
+      "step": 8858
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2464020058543912e-07,
+      "loss": 2.1216,
+      "step": 8859
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2425322847218368e-07,
+      "loss": 2.122,
+      "step": 8860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2386685045388313e-07,
+      "loss": 2.2075,
+      "step": 8861
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2348106657761537e-07,
+      "loss": 2.1437,
+      "step": 8862
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2309587689038783e-07,
+      "loss": 2.1147,
+      "step": 8863
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2271128143913458e-07,
+      "loss": 2.1501,
+      "step": 8864
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2232728027071704e-07,
+      "loss": 2.1781,
+      "step": 8865
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2194387343192504e-07,
+      "loss": 2.1412,
+      "step": 8866
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2156106096947563e-07,
+      "loss": 2.138,
+      "step": 8867
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.211788429300126e-07,
+      "loss": 2.1918,
+      "step": 8868
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.207972193601087e-07,
+      "loss": 2.1292,
+      "step": 8869
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2041619030626283e-07,
+      "loss": 2.1753,
+      "step": 8870
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.20035755814904e-07,
+      "loss": 2.1332,
+      "step": 8871
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1965591593238513e-07,
+      "loss": 2.1684,
+      "step": 8872
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1927667070498916e-07,
+      "loss": 2.1775,
+      "step": 8873
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1889802017892638e-07,
+      "loss": 2.1783,
+      "step": 8874
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.185199644003332e-07,
+      "loss": 2.1766,
+      "step": 8875
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1814250341527611e-07,
+      "loss": 2.168,
+      "step": 8876
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.177656372697461e-07,
+      "loss": 2.2107,
+      "step": 8877
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1738936600966366e-07,
+      "loss": 2.1488,
+      "step": 8878
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1701368968087711e-07,
+      "loss": 2.082,
+      "step": 8879
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.166386083291604e-07,
+      "loss": 2.1712,
+      "step": 8880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1626412200021697e-07,
+      "loss": 2.1807,
+      "step": 8881
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1589023073967586e-07,
+      "loss": 2.2051,
+      "step": 8882
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.155169345930951e-07,
+      "loss": 2.2104,
+      "step": 8883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1514423360595939e-07,
+      "loss": 2.1749,
+      "step": 8884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1477212782368185e-07,
+      "loss": 2.2028,
+      "step": 8885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1440061729160235e-07,
+      "loss": 2.1924,
+      "step": 8886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1402970205498742e-07,
+      "loss": 2.1462,
+      "step": 8887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.136593821590326e-07,
+      "loss": 2.1392,
+      "step": 8888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1328965764886069e-07,
+      "loss": 2.202,
+      "step": 8889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1292052856952063e-07,
+      "loss": 2.1866,
+      "step": 8890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1255199496599034e-07,
+      "loss": 2.1641,
+      "step": 8891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1218405688317447e-07,
+      "loss": 2.1671,
+      "step": 8892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.118167143659038e-07,
+      "loss": 2.1694,
+      "step": 8893
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1144996745894033e-07,
+      "loss": 2.1884,
+      "step": 8894
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1108381620696885e-07,
+      "loss": 2.1138,
+      "step": 8895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.107182606546059e-07,
+      "loss": 2.1204,
+      "step": 8896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1035330084639084e-07,
+      "loss": 2.1728,
+      "step": 8897
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0998893682679479e-07,
+      "loss": 2.1878,
+      "step": 8898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0962516864021388e-07,
+      "loss": 2.2001,
+      "step": 8899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0926199633097156e-07,
+      "loss": 2.1534,
+      "step": 8900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0889941994332077e-07,
+      "loss": 2.1463,
+      "step": 8901
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0853743952143836e-07,
+      "loss": 2.1977,
+      "step": 8902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0817605510943241e-07,
+      "loss": 2.2023,
+      "step": 8903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0781526675133492e-07,
+      "loss": 2.2052,
+      "step": 8904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0745507449110792e-07,
+      "loss": 2.158,
+      "step": 8905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0709547837263967e-07,
+      "loss": 2.1797,
+      "step": 8906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.067364784397451e-07,
+      "loss": 2.122,
+      "step": 8907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0637807473616812e-07,
+      "loss": 2.1699,
+      "step": 8908
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0602026730557879e-07,
+      "loss": 2.2495,
+      "step": 8909
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0566305619157502e-07,
+      "loss": 2.131,
+      "step": 8910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0530644143768143e-07,
+      "loss": 2.0753,
+      "step": 8911
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0495042308735104e-07,
+      "loss": 2.1276,
+      "step": 8912
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0459500118396304e-07,
+      "loss": 2.1801,
+      "step": 8913
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0424017577082556e-07,
+      "loss": 2.1436,
+      "step": 8914
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.038859468911707e-07,
+      "loss": 2.1888,
+      "step": 8915
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0353231458816338e-07,
+      "loss": 2.1148,
+      "step": 8916
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0317927890489021e-07,
+      "loss": 2.1542,
+      "step": 8917
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0282683988436792e-07,
+      "loss": 2.1646,
+      "step": 8918
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.024749975695416e-07,
+      "loss": 2.1878,
+      "step": 8919
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0212375200327973e-07,
+      "loss": 2.1576,
+      "step": 8920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0177310322838251e-07,
+      "loss": 2.1373,
+      "step": 8921
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0142305128757468e-07,
+      "loss": 2.0989,
+      "step": 8922
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0107359622350877e-07,
+      "loss": 2.116,
+      "step": 8923
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.007247380787657e-07,
+      "loss": 2.202,
+      "step": 8924
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0037647689585207e-07,
+      "loss": 2.1345,
+      "step": 8925
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0002881271720222e-07,
+      "loss": 2.1717,
+      "step": 8926
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.968174558517895e-08,
+      "loss": 2.1991,
+      "step": 8927
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.933527554207012e-08,
+      "loss": 2.1717,
+      "step": 8928
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.898940263009304e-08,
+      "loss": 2.2104,
+      "step": 8929
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.864412689139124e-08,
+      "loss": 2.181,
+      "step": 8930
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.82994483680344e-08,
+      "loss": 2.1691,
+      "step": 8931
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.795536710202169e-08,
+      "loss": 2.2136,
+      "step": 8932
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.761188313527792e-08,
+      "loss": 2.1786,
+      "step": 8933
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.726899650965626e-08,
+      "loss": 2.2405,
+      "step": 8934
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.692670726693498e-08,
+      "loss": 2.1981,
+      "step": 8935
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.658501544882182e-08,
+      "loss": 2.1822,
+      "step": 8936
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.62439210969518e-08,
+      "loss": 2.1701,
+      "step": 8937
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.590342425288446e-08,
+      "loss": 2.1498,
+      "step": 8938
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.556352495810994e-08,
+      "loss": 2.1121,
+      "step": 8939
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.522422325404234e-08,
+      "loss": 2.1597,
+      "step": 8940
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.488551918202527e-08,
+      "loss": 2.1954,
+      "step": 8941
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.454741278333013e-08,
+      "loss": 2.21,
+      "step": 8942
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.420990409915176e-08,
+      "loss": 2.2073,
+      "step": 8943
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.387299317061615e-08,
+      "loss": 2.1401,
+      "step": 8944
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.353668003877437e-08,
+      "loss": 2.1835,
+      "step": 8945
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.320096474460527e-08,
+      "loss": 2.2007,
+      "step": 8946
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.28658473290145e-08,
+      "loss": 2.0936,
+      "step": 8947
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.253132783283548e-08,
+      "loss": 2.1705,
+      "step": 8948
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.219740629682838e-08,
+      "loss": 2.1002,
+      "step": 8949
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.186408276168012e-08,
+      "loss": 2.2386,
+      "step": 8950
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.153135726800599e-08,
+      "loss": 2.1763,
+      "step": 8951
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.119922985634633e-08,
+      "loss": 2.1518,
+      "step": 8952
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.086770056717099e-08,
+      "loss": 2.1611,
+      "step": 8953
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.053676944087542e-08,
+      "loss": 2.1729,
+      "step": 8954
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.020643651778183e-08,
+      "loss": 2.1621,
+      "step": 8955
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.987670183814134e-08,
+      "loss": 2.2044,
+      "step": 8956
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.954756544213128e-08,
+      "loss": 2.1268,
+      "step": 8957
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.921902736985399e-08,
+      "loss": 2.1898,
+      "step": 8958
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.889108766134358e-08,
+      "loss": 2.1374,
+      "step": 8959
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.856374635655696e-08,
+      "loss": 2.1012,
+      "step": 8960
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.823700349537945e-08,
+      "loss": 2.1462,
+      "step": 8961
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.791085911762476e-08,
+      "loss": 2.168,
+      "step": 8962
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.758531326303054e-08,
+      "loss": 2.1654,
+      "step": 8963
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.726036597126619e-08,
+      "loss": 2.1846,
+      "step": 8964
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.693601728192392e-08,
+      "loss": 2.2159,
+      "step": 8965
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.661226723452542e-08,
+      "loss": 2.1963,
+      "step": 8966
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.628911586851752e-08,
+      "loss": 2.1383,
+      "step": 8967
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.596656322327645e-08,
+      "loss": 2.1856,
+      "step": 8968
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.564460933810414e-08,
+      "loss": 2.2007,
+      "step": 8969
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.53232542522292e-08,
+      "loss": 2.2133,
+      "step": 8970
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.500249800480754e-08,
+      "loss": 2.1696,
+      "step": 8971
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.468234063492287e-08,
+      "loss": 2.1935,
+      "step": 8972
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.436278218158511e-08,
+      "loss": 2.0994,
+      "step": 8973
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.404382268373145e-08,
+      "loss": 2.1279,
+      "step": 8974
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.372546218022747e-08,
+      "loss": 2.1263,
+      "step": 8975
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.340770070986215e-08,
+      "loss": 2.1467,
+      "step": 8976
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.30905383113556e-08,
+      "loss": 2.145,
+      "step": 8977
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.277397502335194e-08,
+      "loss": 2.1383,
+      "step": 8978
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.245801088442362e-08,
+      "loss": 2.1219,
+      "step": 8979
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.214264593307097e-08,
+      "loss": 2.1577,
+      "step": 8980
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.182788020771826e-08,
+      "loss": 2.2223,
+      "step": 8981
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.151371374672146e-08,
+      "loss": 2.1947,
+      "step": 8982
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.120014658835828e-08,
+      "loss": 2.111,
+      "step": 8983
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.088717877083706e-08,
+      "loss": 2.145,
+      "step": 8984
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.057481033229176e-08,
+      "loss": 2.1734,
+      "step": 8985
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.02630413107841e-08,
+      "loss": 2.1523,
+      "step": 8986
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.995187174430152e-08,
+      "loss": 2.1328,
+      "step": 8987
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.964130167075923e-08,
+      "loss": 2.1667,
+      "step": 8988
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.933133112799918e-08,
+      "loss": 2.1499,
+      "step": 8989
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.90219601537906e-08,
+      "loss": 2.1544,
+      "step": 8990
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.871318878582889e-08,
+      "loss": 2.1444,
+      "step": 8991
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.840501706173786e-08,
+      "loss": 2.2301,
+      "step": 8992
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.809744501906635e-08,
+      "loss": 2.1004,
+      "step": 8993
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.779047269529105e-08,
+      "loss": 2.1743,
+      "step": 8994
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.748410012781705e-08,
+      "loss": 2.212,
+      "step": 8995
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.717832735397335e-08,
+      "loss": 2.0805,
+      "step": 8996
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.687315441101795e-08,
+      "loss": 2.1899,
+      "step": 8997
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.656858133613498e-08,
+      "loss": 2.1305,
+      "step": 8998
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.626460816643588e-08,
+      "loss": 2.1409,
+      "step": 8999
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.59612349389599e-08,
+      "loss": 2.1663,
+      "step": 9000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.565846169067026e-08,
+      "loss": 2.2241,
+      "step": 9001
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.535628845846077e-08,
+      "loss": 2.1646,
+      "step": 9002
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.50547152791492e-08,
+      "loss": 2.1631,
+      "step": 9003
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.475374218948118e-08,
+      "loss": 2.2475,
+      "step": 9004
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.445336922613067e-08,
+      "loss": 2.1757,
+      "step": 9005
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.415359642569564e-08,
+      "loss": 2.1645,
+      "step": 9006
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.385442382470354e-08,
+      "loss": 2.1876,
+      "step": 9007
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.355585145960743e-08,
+      "loss": 2.1712,
+      "step": 9008
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.325787936678708e-08,
+      "loss": 2.1989,
+      "step": 9009
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.296050758254958e-08,
+      "loss": 2.1856,
+      "step": 9010
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.266373614312927e-08,
+      "loss": 2.1172,
+      "step": 9011
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.236756508468612e-08,
+      "loss": 2.192,
+      "step": 9012
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.207199444330847e-08,
+      "loss": 2.1706,
+      "step": 9013
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.177702425500977e-08,
+      "loss": 2.1085,
+      "step": 9014
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.148265455573233e-08,
+      "loss": 2.2579,
+      "step": 9015
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.118888538134361e-08,
+      "loss": 2.2094,
+      "step": 9016
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.089571676763773e-08,
+      "loss": 2.192,
+      "step": 9017
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.060314875033836e-08,
+      "loss": 2.1008,
+      "step": 9018
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.031118136509196e-08,
+      "loss": 2.1891,
+      "step": 9019
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.001981464747565e-08,
+      "loss": 2.1652,
+      "step": 9020
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.972904863298991e-08,
+      "loss": 2.1435,
+      "step": 9021
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.943888335706472e-08,
+      "loss": 2.1743,
+      "step": 9022
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.914931885505626e-08,
+      "loss": 2.1445,
+      "step": 9023
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.88603551622452e-08,
+      "loss": 2.1724,
+      "step": 9024
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.857199231384282e-08,
+      "loss": 2.139,
+      "step": 9025
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.828423034498488e-08,
+      "loss": 2.1927,
+      "step": 9026
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.799706929073335e-08,
+      "loss": 2.1532,
+      "step": 9027
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.771050918607913e-08,
+      "loss": 2.1616,
+      "step": 9028
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.742455006593762e-08,
+      "loss": 2.2195,
+      "step": 9029
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.713919196515317e-08,
+      "loss": 2.1352,
+      "step": 9030
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.685443491849464e-08,
+      "loss": 2.1783,
+      "step": 9031
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.657027896065982e-08,
+      "loss": 2.143,
+      "step": 9032
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.628672412627158e-08,
+      "loss": 2.2007,
+      "step": 9033
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.60037704498806e-08,
+      "loss": 2.1572,
+      "step": 9034
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.572141796596376e-08,
+      "loss": 2.1127,
+      "step": 9035
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.543966670892465e-08,
+      "loss": 2.2377,
+      "step": 9036
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.515851671309414e-08,
+      "loss": 2.1726,
+      "step": 9037
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.487796801272983e-08,
+      "loss": 2.1707,
+      "step": 9038
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.459802064201437e-08,
+      "loss": 2.173,
+      "step": 9039
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.431867463506047e-08,
+      "loss": 2.1848,
+      "step": 9040
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.403993002590425e-08,
+      "loss": 2.1889,
+      "step": 9041
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.376178684850965e-08,
+      "loss": 2.165,
+      "step": 9042
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.348424513676898e-08,
+      "loss": 2.1309,
+      "step": 9043
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.3207304924498e-08,
+      "loss": 2.1435,
+      "step": 9044
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.293096624544304e-08,
+      "loss": 2.1372,
+      "step": 9045
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.265522913327326e-08,
+      "loss": 2.1537,
+      "step": 9046
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.238009362158793e-08,
+      "loss": 2.2683,
+      "step": 9047
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.210555974391075e-08,
+      "loss": 2.1537,
+      "step": 9048
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.183162753369221e-08,
+      "loss": 2.1966,
+      "step": 9049
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.15582970243117e-08,
+      "loss": 2.1172,
+      "step": 9050
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.128556824907205e-08,
+      "loss": 2.1396,
+      "step": 9051
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.101344124120557e-08,
+      "loss": 2.1475,
+      "step": 9052
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.074191603386958e-08,
+      "loss": 2.136,
+      "step": 9053
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.047099266014877e-08,
+      "loss": 2.1636,
+      "step": 9054
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.020067115305451e-08,
+      "loss": 2.1375,
+      "step": 9055
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.993095154552431e-08,
+      "loss": 2.1489,
+      "step": 9056
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.966183387042246e-08,
+      "loss": 2.1753,
+      "step": 9057
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.939331816054161e-08,
+      "loss": 2.1929,
+      "step": 9058
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.9125404448597825e-08,
+      "loss": 2.1087,
+      "step": 9059
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.8858092767236084e-08,
+      "loss": 2.2,
+      "step": 9060
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.8591383149028126e-08,
+      "loss": 2.2112,
+      "step": 9061
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.8325275626470166e-08,
+      "loss": 2.1353,
+      "step": 9062
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.80597702319885e-08,
+      "loss": 2.2107,
+      "step": 9063
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.7794866997933355e-08,
+      "loss": 2.1585,
+      "step": 9064
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.753056595658224e-08,
+      "loss": 2.1819,
+      "step": 9065
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.726686714013996e-08,
+      "loss": 2.1605,
+      "step": 9066
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.700377058073636e-08,
+      "loss": 2.1091,
+      "step": 9067
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.674127631043025e-08,
+      "loss": 2.1861,
+      "step": 9068
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.647938436120437e-08,
+      "loss": 2.1445,
+      "step": 9069
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.621809476497098e-08,
+      "loss": 2.2238,
+      "step": 9070
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.595740755356627e-08,
+      "loss": 2.1492,
+      "step": 9071
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.569732275875428e-08,
+      "loss": 2.2108,
+      "step": 9072
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.543784041222633e-08,
+      "loss": 2.1169,
+      "step": 9073
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.517896054559879e-08,
+      "loss": 2.1413,
+      "step": 9074
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.492068319041588e-08,
+      "loss": 2.1646,
+      "step": 9075
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.466300837814797e-08,
+      "loss": 2.1648,
+      "step": 9076
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.440593614019107e-08,
+      "loss": 2.1441,
+      "step": 9077
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.414946650786957e-08,
+      "loss": 2.1934,
+      "step": 9078
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.389359951243345e-08,
+      "loss": 2.1561,
+      "step": 9079
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.363833518505834e-08,
+      "loss": 2.2144,
+      "step": 9080
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.338367355684881e-08,
+      "loss": 2.1714,
+      "step": 9081
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.312961465883393e-08,
+      "loss": 2.0961,
+      "step": 9082
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.2876158521969476e-08,
+      "loss": 2.1359,
+      "step": 9083
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.262330517713965e-08,
+      "loss": 2.2185,
+      "step": 9084
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.237105465515258e-08,
+      "loss": 2.2217,
+      "step": 9085
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.2119406986745336e-08,
+      "loss": 2.1929,
+      "step": 9086
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.186836220257951e-08,
+      "loss": 2.0745,
+      "step": 9087
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.161792033324398e-08,
+      "loss": 2.123,
+      "step": 9088
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.136808140925542e-08,
+      "loss": 2.1478,
+      "step": 9089
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.111884546105506e-08,
+      "loss": 2.2194,
+      "step": 9090
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.0870212519012477e-08,
+      "loss": 2.1768,
+      "step": 9091
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.062218261342122e-08,
+      "loss": 2.216,
+      "step": 9092
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.0374755774504346e-08,
+      "loss": 2.1306,
+      "step": 9093
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.012793203240995e-08,
+      "loss": 2.1544,
+      "step": 9094
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.988171141721232e-08,
+      "loss": 2.1235,
+      "step": 9095
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9636093958913e-08,
+      "loss": 2.1874,
+      "step": 9096
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.939107968743917e-08,
+      "loss": 2.1901,
+      "step": 9097
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.914666863264528e-08,
+      "loss": 2.2106,
+      "step": 9098
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.890286082431306e-08,
+      "loss": 2.1646,
+      "step": 9099
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.865965629214819e-08,
+      "loss": 2.1572,
+      "step": 9100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.841705506578587e-08,
+      "loss": 2.1664,
+      "step": 9101
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8175057174785766e-08,
+      "loss": 2.1429,
+      "step": 9102
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.793366264863375e-08,
+      "loss": 2.1542,
+      "step": 9103
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.769287151674407e-08,
+      "loss": 2.2093,
+      "step": 9104
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7452683808456026e-08,
+      "loss": 2.1993,
+      "step": 9105
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7213099553035655e-08,
+      "loss": 2.1757,
+      "step": 9106
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.697411877967573e-08,
+      "loss": 2.1692,
+      "step": 9107
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6735741517495715e-08,
+      "loss": 2.2082,
+      "step": 9108
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.649796779554016e-08,
+      "loss": 2.117,
+      "step": 9109
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.626079764278202e-08,
+      "loss": 2.1798,
+      "step": 9110
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6024231088119266e-08,
+      "loss": 2.1454,
+      "step": 9111
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.578826816037718e-08,
+      "loss": 2.1174,
+      "step": 9112
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5552908888306654e-08,
+      "loss": 2.1673,
+      "step": 9113
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.531815330058586e-08,
+      "loss": 2.1549,
+      "step": 9114
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.508400142581859e-08,
+      "loss": 2.1383,
+      "step": 9115
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.485045329253646e-08,
+      "loss": 2.155,
+      "step": 9116
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4617508929195585e-08,
+      "loss": 2.1807,
+      "step": 9117
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.438516836417994e-08,
+      "loss": 2.1744,
+      "step": 9118
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.415343162580022e-08,
+      "loss": 2.1803,
+      "step": 9119
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.392229874229159e-08,
+      "loss": 2.0817,
+      "step": 9120
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.36917697418171e-08,
+      "loss": 2.1802,
+      "step": 9121
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.346184465246761e-08,
+      "loss": 2.1471,
+      "step": 9122
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3232523502256264e-08,
+      "loss": 2.1315,
+      "step": 9123
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3003806319127376e-08,
+      "loss": 2.1783,
+      "step": 9124
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2775693130948094e-08,
+      "loss": 2.2076,
+      "step": 9125
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2548183965513415e-08,
+      "loss": 2.1733,
+      "step": 9126
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.23212788505456e-08,
+      "loss": 2.1471,
+      "step": 9127
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.209497781369143e-08,
+      "loss": 2.1879,
+      "step": 9128
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1869280882525506e-08,
+      "loss": 2.052,
+      "step": 9129
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.164418808454806e-08,
+      "loss": 2.1899,
+      "step": 9130
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1419699447186045e-08,
+      "loss": 2.134,
+      "step": 9131
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1195814997792014e-08,
+      "loss": 2.0993,
+      "step": 9132
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.097253476364693e-08,
+      "loss": 2.1557,
+      "step": 9133
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0749858771956253e-08,
+      "loss": 2.165,
+      "step": 9134
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.052778704985216e-08,
+      "loss": 2.1687,
+      "step": 9135
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.030631962439302e-08,
+      "loss": 2.204,
+      "step": 9136
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.008545652256502e-08,
+      "loss": 2.2441,
+      "step": 9137
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.986519777127884e-08,
+      "loss": 2.1727,
+      "step": 9138
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.964554339737303e-08,
+      "loss": 2.1895,
+      "step": 9139
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9426493427611177e-08,
+      "loss": 2.1177,
+      "step": 9140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9208047888683597e-08,
+      "loss": 2.2144,
+      "step": 9141
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.899020680720844e-08,
+      "loss": 2.1561,
+      "step": 9142
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.877297020972781e-08,
+      "loss": 2.1626,
+      "step": 9143
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.855633812271165e-08,
+      "loss": 2.1662,
+      "step": 9144
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.83403105725566e-08,
+      "loss": 2.1457,
+      "step": 9145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.812488758558386e-08,
+      "loss": 2.099,
+      "step": 9146
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.791006918804296e-08,
+      "loss": 2.1417,
+      "step": 9147
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.769585540610799e-08,
+      "loss": 2.1665,
+      "step": 9148
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.748224626588137e-08,
+      "loss": 2.1654,
+      "step": 9149
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.726924179339009e-08,
+      "loss": 2.2192,
+      "step": 9150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7056842014587815e-08,
+      "loss": 2.1619,
+      "step": 9151
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.684504695535496e-08,
+      "loss": 2.2047,
+      "step": 9152
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.663385664149866e-08,
+      "loss": 2.2033,
+      "step": 9153
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.642327109875166e-08,
+      "loss": 2.1692,
+      "step": 9154
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.621329035277232e-08,
+      "loss": 2.221,
+      "step": 9155
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.600391442914741e-08,
+      "loss": 2.1354,
+      "step": 9156
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.579514335338763e-08,
+      "loss": 2.2204,
+      "step": 9157
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.558697715093207e-08,
+      "loss": 2.233,
+      "step": 9158
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5379415847143775e-08,
+      "loss": 2.1278,
+      "step": 9159
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.517245946731529e-08,
+      "loss": 2.1257,
+      "step": 9160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4966108036662006e-08,
+      "loss": 2.2249,
+      "step": 9161
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.47603615803288e-08,
+      "loss": 2.1608,
+      "step": 9162
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4555220123383416e-08,
+      "loss": 2.1579,
+      "step": 9163
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.435068369082306e-08,
+      "loss": 2.1462,
+      "step": 9164
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.414675230757003e-08,
+      "loss": 2.1468,
+      "step": 9165
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.394342599847111e-08,
+      "loss": 2.1303,
+      "step": 9166
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.374070478830316e-08,
+      "loss": 2.1666,
+      "step": 9167
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3538588701765296e-08,
+      "loss": 2.2,
+      "step": 9168
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3337077763485605e-08,
+      "loss": 2.1611,
+      "step": 9169
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.313617199801777e-08,
+      "loss": 2.1213,
+      "step": 9170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2935871429841116e-08,
+      "loss": 2.1491,
+      "step": 9171
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2736176083362216e-08,
+      "loss": 2.2281,
+      "step": 9172
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.253708598291272e-08,
+      "loss": 2.1101,
+      "step": 9173
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2338601152751e-08,
+      "loss": 2.1091,
+      "step": 9174
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.214072161706272e-08,
+      "loss": 2.2582,
+      "step": 9175
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.194344739995803e-08,
+      "loss": 2.2107,
+      "step": 9176
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1746778525474916e-08,
+      "loss": 2.2175,
+      "step": 9177
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1550715017575895e-08,
+      "loss": 2.1945,
+      "step": 9178
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.135525690015184e-08,
+      "loss": 2.1294,
+      "step": 9179
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1160404197018155e-08,
+      "loss": 2.1951,
+      "step": 9180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0966156931916955e-08,
+      "loss": 2.17,
+      "step": 9181
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.077251512851709e-08,
+      "loss": 2.1434,
+      "step": 9182
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.057947881041301e-08,
+      "loss": 2.1908,
+      "step": 9183
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.038704800112535e-08,
+      "loss": 2.1824,
+      "step": 9184
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.019522272410202e-08,
+      "loss": 2.1406,
+      "step": 9185
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0004003002714886e-08,
+      "loss": 2.1553,
+      "step": 9186
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.981338886026475e-08,
+      "loss": 2.2293,
+      "step": 9187
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9623380319976912e-08,
+      "loss": 2.1897,
+      "step": 9188
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9433977405003976e-08,
+      "loss": 2.1116,
+      "step": 9189
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9245180138423033e-08,
+      "loss": 2.147,
+      "step": 9190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9056988543239018e-08,
+      "loss": 2.1718,
+      "step": 9191
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8869402642382473e-08,
+      "loss": 2.1552,
+      "step": 9192
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8682422458710667e-08,
+      "loss": 2.1654,
+      "step": 9193
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8496048015005385e-08,
+      "loss": 2.1427,
+      "step": 9194
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8310279333976786e-08,
+      "loss": 2.1526,
+      "step": 9195
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8125116438260104e-08,
+      "loss": 2.2252,
+      "step": 9196
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.794055935041673e-08,
+      "loss": 2.1321,
+      "step": 9197
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7756608092933678e-08,
+      "loss": 2.1555,
+      "step": 9198
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7573262688226355e-08,
+      "loss": 2.2136,
+      "step": 9199
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7390523158633552e-08,
+      "loss": 2.1952,
+      "step": 9200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7208389526421907e-08,
+      "loss": 2.2111,
+      "step": 9201
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7026861813783668e-08,
+      "loss": 2.1732,
+      "step": 9202
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.684594004283836e-08,
+      "loss": 2.1348,
+      "step": 9203
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6665624235629463e-08,
+      "loss": 2.1124,
+      "step": 9204
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.64859144141294e-08,
+      "loss": 2.218,
+      "step": 9205
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6306810600233435e-08,
+      "loss": 2.1676,
+      "step": 9206
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6128312815766332e-08,
+      "loss": 2.1592,
+      "step": 9207
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5950421082476805e-08,
+      "loss": 2.1598,
+      "step": 9208
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5773135422040296e-08,
+      "loss": 2.1588,
+      "step": 9209
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5596455856058966e-08,
+      "loss": 2.1409,
+      "step": 9210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5420382406060595e-08,
+      "loss": 2.2284,
+      "step": 9211
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5244915093499134e-08,
+      "loss": 2.1115,
+      "step": 9212
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5070053939754702e-08,
+      "loss": 2.167,
+      "step": 9213
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.489579896613359e-08,
+      "loss": 2.2158,
+      "step": 9214
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.472215019386881e-08,
+      "loss": 2.0378,
+      "step": 9215
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4549107644117888e-08,
+      "loss": 2.1339,
+      "step": 9216
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4376671337966174e-08,
+      "loss": 2.1739,
+      "step": 9217
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4204841296424086e-08,
+      "loss": 2.2424,
+      "step": 9218
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.403361754042932e-08,
+      "loss": 2.1942,
+      "step": 9219
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.386300009084408e-08,
+      "loss": 2.1532,
+      "step": 9220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3692988968458398e-08,
+      "loss": 2.2014,
+      "step": 9221
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3523584193986816e-08,
+      "loss": 2.1302,
+      "step": 9222
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3354785788072265e-08,
+      "loss": 2.1946,
+      "step": 9223
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3186593771280518e-08,
+      "loss": 2.1608,
+      "step": 9224
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.301900816410574e-08,
+      "loss": 2.2127,
+      "step": 9225
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.285202898696881e-08,
+      "loss": 2.1376,
+      "step": 9226
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.268565626021457e-08,
+      "loss": 2.182,
+      "step": 9227
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.251989000411514e-08,
+      "loss": 2.1765,
+      "step": 9228
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2354730238868804e-08,
+      "loss": 2.1528,
+      "step": 9229
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.219017698460002e-08,
+      "loss": 2.2028,
+      "step": 9230
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.202623026135886e-08,
+      "loss": 2.1744,
+      "step": 9231
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1862890089121567e-08,
+      "loss": 2.1926,
+      "step": 9232
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1700156487790557e-08,
+      "loss": 2.132,
+      "step": 9233
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1538029477195522e-08,
+      "loss": 2.1559,
+      "step": 9234
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1376509077089546e-08,
+      "loss": 2.1567,
+      "step": 9235
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1215595307154667e-08,
+      "loss": 2.1879,
+      "step": 9236
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.105528818699687e-08,
+      "loss": 2.166,
+      "step": 9237
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0895587736149414e-08,
+      "loss": 2.2101,
+      "step": 9238
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0736493974071736e-08,
+      "loss": 2.1394,
+      "step": 9239
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.057800692014833e-08,
+      "loss": 2.1457,
+      "step": 9240
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0420126593690416e-08,
+      "loss": 2.1294,
+      "step": 9241
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.026285301393538e-08,
+      "loss": 2.1338,
+      "step": 9242
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.010618620004734e-08,
+      "loss": 2.1443,
+      "step": 9243
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.995012617111436e-08,
+      "loss": 2.1948,
+      "step": 9244
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9794672946152337e-08,
+      "loss": 2.1911,
+      "step": 9245
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.963982654410279e-08,
+      "loss": 2.1116,
+      "step": 9246
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9485586983833404e-08,
+      "loss": 2.1589,
+      "step": 9247
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9331954284137476e-08,
+      "loss": 2.179,
+      "step": 9248
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9178928463735593e-08,
+      "loss": 2.1654,
+      "step": 9249
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9026509541272276e-08,
+      "loss": 2.0799,
+      "step": 9250
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8874697535319897e-08,
+      "loss": 2.1678,
+      "step": 9251
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8723492464376992e-08,
+      "loss": 2.1811,
+      "step": 9252
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8572894346866043e-08,
+      "loss": 2.1435,
+      "step": 9253
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.842290320113793e-08,
+      "loss": 2.1709,
+      "step": 9254
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8273519045468035e-08,
+      "loss": 2.1766,
+      "step": 9255
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8124741898058462e-08,
+      "loss": 2.0566,
+      "step": 9256
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7976571777038044e-08,
+      "loss": 2.1169,
+      "step": 9257
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7829008700460116e-08,
+      "loss": 2.1492,
+      "step": 9258
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.768205268630474e-08,
+      "loss": 2.0853,
+      "step": 9259
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.753570375247815e-08,
+      "loss": 2.1676,
+      "step": 9260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.73899619168133e-08,
+      "loss": 2.2094,
+      "step": 9261
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7244827197067103e-08,
+      "loss": 1.9992,
+      "step": 9262
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7100299610924297e-08,
+      "loss": 2.1342,
+      "step": 9263
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6956379175995796e-08,
+      "loss": 2.1316,
+      "step": 9264
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.681306590981702e-08,
+      "loss": 2.0565,
+      "step": 9265
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6670359829850657e-08,
+      "loss": 2.1358,
+      "step": 9266
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6528260953484476e-08,
+      "loss": 2.1891,
+      "step": 9267
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6386769298034067e-08,
+      "loss": 2.1335,
+      "step": 9268
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6245884880738415e-08,
+      "loss": 2.1939,
+      "step": 9269
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.610560771876435e-08,
+      "loss": 2.1626,
+      "step": 9270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5965937829204302e-08,
+      "loss": 2.1777,
+      "step": 9271
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.582687522907633e-08,
+      "loss": 2.1468,
+      "step": 9272
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5688419935325216e-08,
+      "loss": 2.2188,
+      "step": 9273
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5550571964820793e-08,
+      "loss": 2.1181,
+      "step": 9274
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.541333133436018e-08,
+      "loss": 2.0651,
+      "step": 9275
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5276698060665007e-08,
+      "loss": 2.1715,
+      "step": 9276
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5140672160384174e-08,
+      "loss": 2.1326,
+      "step": 9277
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.500525365009109e-08,
+      "loss": 2.1352,
+      "step": 9278
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4870442546287555e-08,
+      "loss": 2.1624,
+      "step": 9279
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4736238865398766e-08,
+      "loss": 2.1803,
+      "step": 9280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4602642623777752e-08,
+      "loss": 2.1856,
+      "step": 9281
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4469653837701491e-08,
+      "loss": 2.1827,
+      "step": 9282
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4337272523375911e-08,
+      "loss": 2.1478,
+      "step": 9283
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4205498696930332e-08,
+      "loss": 2.1663,
+      "step": 9284
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4074332374421351e-08,
+      "loss": 2.2357,
+      "step": 9285
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3943773571831188e-08,
+      "loss": 2.1744,
+      "step": 9286
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3813822305067115e-08,
+      "loss": 2.1335,
+      "step": 9287
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3684478589964801e-08,
+      "loss": 2.1888,
+      "step": 9288
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.355574244228386e-08,
+      "loss": 2.1903,
+      "step": 9289
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3427613877709523e-08,
+      "loss": 2.1075,
+      "step": 9290
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3300092911854856e-08,
+      "loss": 2.0829,
+      "step": 9291
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3173179560257432e-08,
+      "loss": 2.1645,
+      "step": 9292
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3046873838381546e-08,
+      "loss": 2.0987,
+      "step": 9293
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.292117576161711e-08,
+      "loss": 2.2373,
+      "step": 9294
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2796085345280207e-08,
+      "loss": 2.2754,
+      "step": 9295
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2671602604612531e-08,
+      "loss": 2.1138,
+      "step": 9296
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2547727554781398e-08,
+      "loss": 2.2128,
+      "step": 9297
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2424460210881394e-08,
+      "loss": 2.2266,
+      "step": 9298
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2301800587932179e-08,
+      "loss": 2.1476,
+      "step": 9299
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2179748700879013e-08,
+      "loss": 2.1615,
+      "step": 9300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2058304564593893e-08,
+      "loss": 2.1411,
+      "step": 9301
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1937468193873869e-08,
+      "loss": 2.1795,
+      "step": 9302
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1817239603443276e-08,
+      "loss": 2.1402,
+      "step": 9303
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1697618807951504e-08,
+      "loss": 2.1979,
+      "step": 9304
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1578605821973566e-08,
+      "loss": 2.2127,
+      "step": 9305
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.146020066001119e-08,
+      "loss": 2.1045,
+      "step": 9306
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.134240333649117e-08,
+      "loss": 2.2121,
+      "step": 9307
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1225213865767026e-08,
+      "loss": 2.1331,
+      "step": 9308
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1108632262118446e-08,
+      "loss": 2.1642,
+      "step": 9309
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0992658539750179e-08,
+      "loss": 2.1971,
+      "step": 9310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0877292712792586e-08,
+      "loss": 2.1606,
+      "step": 9311
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.076253479530387e-08,
+      "loss": 2.217,
+      "step": 9312
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0648384801266176e-08,
+      "loss": 2.2349,
+      "step": 9313
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0534842744588381e-08,
+      "loss": 2.1788,
+      "step": 9314
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0421908639104971e-08,
+      "loss": 2.1706,
+      "step": 9315
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.030958249857772e-08,
+      "loss": 2.1758,
+      "step": 9316
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0197864336691788e-08,
+      "loss": 2.1596,
+      "step": 9317
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.008675416706073e-08,
+      "loss": 2.1409,
+      "step": 9318
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.976252003223164e-09,
+      "loss": 2.2212,
+      "step": 9319
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.866357858642206e-09,
+      "loss": 2.1333,
+      "step": 9320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.757071746708812e-09,
+      "loss": 2.1814,
+      "step": 9321
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.64839368074011e-09,
+      "loss": 2.1968,
+      "step": 9322
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.540323673976615e-09,
+      "loss": 2.1858,
+      "step": 9323
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.432861739586685e-09,
+      "loss": 2.1422,
+      "step": 9324
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.3260078906654e-09,
+      "loss": 2.1542,
+      "step": 9325
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.219762140231237e-09,
+      "loss": 2.1315,
+      "step": 9326
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.114124501230504e-09,
+      "loss": 2.1525,
+      "step": 9327
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.009094986534572e-09,
+      "loss": 2.2182,
+      "step": 9328
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.904673608940983e-09,
+      "loss": 2.2154,
+      "step": 9329
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.800860381173448e-09,
+      "loss": 2.1512,
+      "step": 9330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.697655315881293e-09,
+      "loss": 2.1729,
+      "step": 9331
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.595058425640012e-09,
+      "loss": 2.0918,
+      "step": 9332
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.49306972294961e-09,
+      "loss": 2.1723,
+      "step": 9333
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.391689220238474e-09,
+      "loss": 2.0881,
+      "step": 9334
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.290916929858394e-09,
+      "loss": 2.1198,
+      "step": 9335
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.190752864088436e-09,
+      "loss": 2.1361,
+      "step": 9336
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.091197035133836e-09,
+      "loss": 2.1927,
+      "step": 9337
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.992249455124889e-09,
+      "loss": 2.1677,
+      "step": 9338
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.89391013611751e-09,
+      "loss": 2.1349,
+      "step": 9339
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.796179090094891e-09,
+      "loss": 2.1622,
+      "step": 9340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.699056328964726e-09,
+      "loss": 2.2245,
+      "step": 9341
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.602541864561442e-09,
+      "loss": 2.1422,
+      "step": 9342
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.506635708645072e-09,
+      "loss": 2.0996,
+      "step": 9343
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.411337872900715e-09,
+      "loss": 2.2012,
+      "step": 9344
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.3166483689413035e-09,
+      "loss": 2.2169,
+      "step": 9345
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.222567208303721e-09,
+      "loss": 2.2025,
+      "step": 9346
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.129094402451575e-09,
+      "loss": 2.1408,
+      "step": 9347
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.036229962774088e-09,
+      "loss": 2.1785,
+      "step": 9348
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.943973900586654e-09,
+      "loss": 2.1403,
+      "step": 9349
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.852326227130835e-09,
+      "loss": 2.1459,
+      "step": 9350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.761286953572699e-09,
+      "loss": 2.2144,
+      "step": 9351
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.670856091006151e-09,
+      "loss": 2.1537,
+      "step": 9352
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.581033650449043e-09,
+      "loss": 2.1518,
+      "step": 9353
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.491819642846509e-09,
+      "loss": 2.1724,
+      "step": 9354
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.403214079069298e-09,
+      "loss": 2.1639,
+      "step": 9355
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.315216969912663e-09,
+      "loss": 2.1443,
+      "step": 9356
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.227828326099139e-09,
+      "loss": 2.164,
+      "step": 9357
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.141048158277429e-09,
+      "loss": 2.1546,
+      "step": 9358
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.054876477021299e-09,
+      "loss": 2.1493,
+      "step": 9359
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.969313292830126e-09,
+      "loss": 2.1849,
+      "step": 9360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8843586161289045e-09,
+      "loss": 2.1679,
+      "step": 9361
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.800012457270466e-09,
+      "loss": 2.1795,
+      "step": 9362
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.716274826531587e-09,
+      "loss": 2.19,
+      "step": 9363
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.633145734114665e-09,
+      "loss": 2.174,
+      "step": 9364
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.5506251901504825e-09,
+      "loss": 2.1206,
+      "step": 9365
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.468713204692111e-09,
+      "loss": 2.1909,
+      "step": 9366
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.38740978772101e-09,
+      "loss": 2.191,
+      "step": 9367
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.306714949143699e-09,
+      "loss": 2.0707,
+      "step": 9368
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.226628698792868e-09,
+      "loss": 2.1272,
+      "step": 9369
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.147151046426824e-09,
+      "loss": 2.1009,
+      "step": 9370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.06828200172893e-09,
+      "loss": 2.1842,
+      "step": 9371
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.990021574309834e-09,
+      "loss": 2.1708,
+      "step": 9372
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9123697737052386e-09,
+      "loss": 2.211,
+      "step": 9373
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.835326609376468e-09,
+      "loss": 2.2191,
+      "step": 9374
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.758892090711009e-09,
+      "loss": 2.1477,
+      "step": 9375
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.683066227023081e-09,
+      "loss": 2.1154,
+      "step": 9376
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.607849027550293e-09,
+      "loss": 2.1571,
+      "step": 9377
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.533240501459202e-09,
+      "loss": 2.1498,
+      "step": 9378
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.459240657839203e-09,
+      "loss": 2.169,
+      "step": 9379
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.385849505708084e-09,
+      "loss": 2.1721,
+      "step": 9380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.313067054008135e-09,
+      "loss": 2.1219,
+      "step": 9381
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2408933116072635e-09,
+      "loss": 2.1603,
+      "step": 9382
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.169328287299545e-09,
+      "loss": 2.134,
+      "step": 9383
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.098371989805227e-09,
+      "loss": 2.2103,
+      "step": 9384
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.028024427770172e-09,
+      "loss": 2.1936,
+      "step": 9385
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9582856097658554e-09,
+      "loss": 2.1748,
+      "step": 9386
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.889155544289924e-09,
+      "loss": 2.1826,
+      "step": 9387
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.820634239765642e-09,
+      "loss": 2.1591,
+      "step": 9388
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.752721704541884e-09,
+      "loss": 2.1944,
+      "step": 9389
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.685417946894254e-09,
+      "loss": 2.198,
+      "step": 9390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.618722975022304e-09,
+      "loss": 2.1512,
+      "step": 9391
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5526367970539765e-09,
+      "loss": 2.149,
+      "step": 9392
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.487159421040609e-09,
+      "loss": 2.1772,
+      "step": 9393
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4222908549608193e-09,
+      "loss": 2.1553,
+      "step": 9394
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3580311067188396e-09,
+      "loss": 2.1888,
+      "step": 9395
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.294380184143964e-09,
+      "loss": 2.1539,
+      "step": 9396
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2313380949927643e-09,
+      "loss": 2.2039,
+      "step": 9397
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1689048469457638e-09,
+      "loss": 2.1323,
+      "step": 9398
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1070804476113213e-09,
+      "loss": 2.1774,
+      "step": 9399
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0458649045211897e-09,
+      "loss": 2.1532,
+      "step": 9400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9852582251355124e-09,
+      "loss": 2.2157,
+      "step": 9401
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9252604168383826e-09,
+      "loss": 2.1784,
+      "step": 9402
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.865871486940619e-09,
+      "loss": 2.1228,
+      "step": 9403
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8070914426786555e-09,
+      "loss": 2.1897,
+      "step": 9404
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.748920291214541e-09,
+      "loss": 2.164,
+      "step": 9405
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6913580396359384e-09,
+      "loss": 2.1863,
+      "step": 9406
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6344046949566825e-09,
+      "loss": 2.1885,
+      "step": 9407
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5780602641167774e-09,
+      "loss": 2.1269,
+      "step": 9408
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.522324753981842e-09,
+      "loss": 2.1755,
+      "step": 9409
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4671981713420003e-09,
+      "loss": 2.2134,
+      "step": 9410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.412680522915767e-09,
+      "loss": 2.1404,
+      "step": 9411
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.358771815344496e-09,
+      "loss": 2.1604,
+      "step": 9412
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3054720551973773e-09,
+      "loss": 2.1331,
+      "step": 9413
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2527812489692156e-09,
+      "loss": 2.189,
+      "step": 9414
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2006994030798758e-09,
+      "loss": 2.24,
+      "step": 9415
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.149226523874837e-09,
+      "loss": 2.1434,
+      "step": 9416
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.098362617626859e-09,
+      "loss": 2.1974,
+      "step": 9417
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0481076905332074e-09,
+      "loss": 2.1569,
+      "step": 9418
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9984617487173174e-09,
+      "loss": 2.1835,
+      "step": 9419
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9494247982282386e-09,
+      "loss": 2.1326,
+      "step": 9420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9009968450406368e-09,
+      "loss": 2.1443,
+      "step": 9421
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8531778950564572e-09,
+      "loss": 2.1562,
+      "step": 9422
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.80596795410104e-09,
+      "loss": 2.2371,
+      "step": 9423
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.759367027927561e-09,
+      "loss": 2.1422,
+      "step": 9424
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7133751222137007e-09,
+      "loss": 2.1437,
+      "step": 9425
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6679922425638651e-09,
+      "loss": 2.1803,
+      "step": 9426
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6232183945075197e-09,
+      "loss": 2.1789,
+      "step": 9427
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5790535835003006e-09,
+      "loss": 2.1752,
+      "step": 9428
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.535497814923459e-09,
+      "loss": 2.1902,
+      "step": 9429
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4925510940844157e-09,
+      "loss": 2.1652,
+      "step": 9430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4502134262156519e-09,
+      "loss": 2.1942,
+      "step": 9431
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4084848164763742e-09,
+      "loss": 2.211,
+      "step": 9432
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3673652699508487e-09,
+      "loss": 2.181,
+      "step": 9433
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3268547916495124e-09,
+      "loss": 2.1295,
+      "step": 9434
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.286953386508416e-09,
+      "loss": 2.2146,
+      "step": 9435
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.247661059389227e-09,
+      "loss": 2.1973,
+      "step": 9436
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2089778150797816e-09,
+      "loss": 2.1628,
+      "step": 9437
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.170903658293532e-09,
+      "loss": 2.2037,
+      "step": 9438
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1334385936695447e-09,
+      "loss": 2.1537,
+      "step": 9439
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.096582625772502e-09,
+      "loss": 2.1719,
+      "step": 9440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0603357590938112e-09,
+      "loss": 2.1547,
+      "step": 9441
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0246979980499395e-09,
+      "loss": 2.1538,
+      "step": 9442
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.896693469829689e-10,
+      "loss": 2.173,
+      "step": 9443
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.55249810161152e-10,
+      "loss": 2.2252,
+      "step": 9444
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.214393917789111e-10,
+      "loss": 2.1308,
+      "step": 9445
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.88238095955174e-10,
+      "loss": 2.2062,
+      "step": 9446
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.556459267355932e-10,
+      "loss": 2.1671,
+      "step": 9447
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.236628880914365e-10,
+      "loss": 2.1856,
+      "step": 9448
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.92288983920142e-10,
+      "loss": 2.0468,
+      "step": 9449
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.615242180436521e-10,
+      "loss": 2.1524,
+      "step": 9450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.313685942117454e-10,
+      "loss": 2.2536,
+      "step": 9451
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.018221160981498e-10,
+      "loss": 2.1596,
+      "step": 9452
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.728847873027633e-10,
+      "loss": 2.1294,
+      "step": 9453
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.445566113516544e-10,
+      "loss": 2.1448,
+      "step": 9454
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.168375916970615e-10,
+      "loss": 2.1368,
+      "step": 9455
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.897277317157279e-10,
+      "loss": 2.1641,
+      "step": 9456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.632270347116775e-10,
+      "loss": 2.1754,
+      "step": 9457
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.373355039128836e-10,
+      "loss": 2.2503,
+      "step": 9458
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.120531424751551e-10,
+      "loss": 2.1469,
+      "step": 9459
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.87379953478806e-10,
+      "loss": 2.1443,
+      "step": 9460
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6331593993032e-10,
+      "loss": 2.184,
+      "step": 9461
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.398611047612411e-10,
+      "loss": 2.2128,
+      "step": 9462
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1701545082928343e-10,
+      "loss": 2.1354,
+      "step": 9463
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9477898091944135e-10,
+      "loss": 2.1568,
+      "step": 9464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7315169774010397e-10,
+      "loss": 2.2267,
+      "step": 9465
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.521336039263856e-10,
+      "loss": 2.1915,
+      "step": 9466
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3172470204012597e-10,
+      "loss": 2.1551,
+      "step": 9467
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1192499456766947e-10,
+      "loss": 2.1309,
+      "step": 9468
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9273448392097557e-10,
+      "loss": 2.1387,
+      "step": 9469
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.741531724392843e-10,
+      "loss": 2.1857,
+      "step": 9470
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5618106238634033e-10,
+      "loss": 2.1357,
+      "step": 9471
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.388181559515035e-10,
+      "loss": 2.1159,
+      "step": 9472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2206445525085886e-10,
+      "loss": 2.1884,
+      "step": 9473
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0591996232610656e-10,
+      "loss": 2.143,
+      "step": 9474
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.903846791434516e-10,
+      "loss": 2.1716,
+      "step": 9475
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7545860759693446e-10,
+      "loss": 2.1494,
+      "step": 9476
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.611417495045453e-10,
+      "loss": 2.1017,
+      "step": 9477
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4743410661044454e-10,
+      "loss": 2.1363,
+      "step": 9478
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3433568058607293e-10,
+      "loss": 2.201,
+      "step": 9479
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2184647302626585e-10,
+      "loss": 2.1372,
+      "step": 9480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0996648545313904e-10,
+      "loss": 2.1169,
+      "step": 9481
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.869571931442334e-11,
+      "loss": 2.1452,
+      "step": 9482
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.803417598346465e-11,
+      "loss": 2.1478,
+      "step": 9483
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.798185675866876e-11,
+      "loss": 2.1793,
+      "step": 9484
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.853876286627703e-11,
+      "loss": 2.1784,
+      "step": 9485
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.970489545537028e-11,
+      "loss": 2.0957,
+      "step": 9486
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.148025560341996e-11,
+      "loss": 2.1343,
+      "step": 9487
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3864844311847235e-11,
+      "loss": 2.2606,
+      "step": 9488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.685866250879855e-11,
+      "loss": 2.1775,
+      "step": 9489
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0461711048035415e-11,
+      "loss": 2.227,
+      "step": 9490
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.467399070893439e-11,
+      "loss": 2.1655,
+      "step": 9491
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9495502197042214e-11,
+      "loss": 2.2134,
+      "step": 9492
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4926246142965562e-11,
+      "loss": 2.1862,
+      "step": 9493
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0966223103481278e-11,
+      "loss": 2.1471,
+      "step": 9494
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.615433561536379e-12,
+      "loss": 2.1352,
+      "step": 9495
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.873877924582715e-12,
+      "loss": 2.1815,
+      "step": 9496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.741556527352529e-12,
+      "loss": 2.1824,
+      "step": 9497
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2184696296380083e-12,
+      "loss": 2.161,
+      "step": 9498
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.046174168463978e-13,
+      "loss": 2.2396,
+      "step": 9499
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 2.1549,
+      "step": 9500
+    },
+    {
+      "epoch": 1.0,
+      "step": 9500,
+      "total_flos": 1.224911049719808e+18,
+      "train_loss": 2.288730341296447,
+      "train_runtime": 10450.4593,
+      "train_samples_per_second": 58.179,
+      "train_steps_per_second": 0.909
+    }
+  ],
+  "max_steps": 9500,
+  "num_train_epochs": 1,
+  "total_flos": 1.224911049719808e+18,
+  "trial_name": null,
+  "trial_params": null
+}