diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,49951 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 8321,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-08,
+      "loss": 3.9242,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-08,
+      "loss": 3.8617,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 3.8225,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-08,
+      "loss": 3.7774,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 3.7767,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 3.8118,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4e-07,
+      "loss": 3.8031,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6e-07,
+      "loss": 3.8051,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-07,
+      "loss": 3.8376,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 3.8887,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2e-07,
+      "loss": 3.8142,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 3.7968,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6e-07,
+      "loss": 3.8923,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8e-07,
+      "loss": 3.7336,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 3.8475,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2e-07,
+      "loss": 3.9081,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 3.7773,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-07,
+      "loss": 3.7057,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8e-07,
+      "loss": 3.7931,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 3.805,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 3.7986,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4e-07,
+      "loss": 3.7782,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6000000000000004e-07,
+      "loss": 3.8399,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 3.7737,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 3.7203,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.2e-07,
+      "loss": 3.8388,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.4e-07,
+      "loss": 3.7202,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.6e-07,
+      "loss": 3.7141,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 3.7432,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 3.8619,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.200000000000001e-07,
+      "loss": 3.732,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.4e-07,
+      "loss": 3.7971,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.6e-07,
+      "loss": 3.6812,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 3.7114,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 3.7772,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.2e-07,
+      "loss": 3.7101,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.4e-07,
+      "loss": 3.8006,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.6e-07,
+      "loss": 3.8097,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.8e-07,
+      "loss": 3.8652,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 3.8237,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.200000000000001e-07,
+      "loss": 3.7929,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.400000000000001e-07,
+      "loss": 3.7505,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.6e-07,
+      "loss": 3.7066,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.8e-07,
+      "loss": 3.8587,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 3.7844,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 3.8553,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.400000000000001e-07,
+      "loss": 3.6816,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 3.9029,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 3.7718,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 3.7909,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.02e-06,
+      "loss": 3.8278,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.04e-06,
+      "loss": 3.7345,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.06e-06,
+      "loss": 3.7413,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.08e-06,
+      "loss": 3.7419,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1e-06,
+      "loss": 3.8845,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.12e-06,
+      "loss": 3.9534,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.14e-06,
+      "loss": 3.8334,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1600000000000001e-06,
+      "loss": 3.7625,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 3.8259,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 3.6946,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2200000000000002e-06,
+      "loss": 3.7113,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2400000000000002e-06,
+      "loss": 3.7766,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.26e-06,
+      "loss": 3.8434,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.28e-06,
+      "loss": 3.6297,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3e-06,
+      "loss": 3.7805,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.32e-06,
+      "loss": 3.7409,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.34e-06,
+      "loss": 3.724,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 3.7666,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 3.7239,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 3.8322,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.42e-06,
+      "loss": 3.6828,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.44e-06,
+      "loss": 3.7107,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.46e-06,
+      "loss": 3.7274,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.48e-06,
+      "loss": 3.7572,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5e-06,
+      "loss": 3.6947,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.52e-06,
+      "loss": 3.8282,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.54e-06,
+      "loss": 3.8193,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.56e-06,
+      "loss": 3.8075,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 3.7468,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 3.796,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6200000000000002e-06,
+      "loss": 3.7487,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6400000000000002e-06,
+      "loss": 3.6528,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6600000000000002e-06,
+      "loss": 3.5879,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 3.7779,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 3.801,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.72e-06,
+      "loss": 3.6974,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.74e-06,
+      "loss": 3.6281,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.76e-06,
+      "loss": 3.6381,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 3.7681,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 3.7352,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8200000000000002e-06,
+      "loss": 3.8206,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8400000000000002e-06,
+      "loss": 3.7518,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8600000000000002e-06,
+      "loss": 3.675,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 3.7537,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 3.8119,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 3.6506,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.94e-06,
+      "loss": 3.7337,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9600000000000003e-06,
+      "loss": 3.7753,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.98e-06,
+      "loss": 3.8469,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 3.7752,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.02e-06,
+      "loss": 3.7204,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.04e-06,
+      "loss": 3.6247,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.06e-06,
+      "loss": 3.8008,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.08e-06,
+      "loss": 3.81,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 3.7768,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.12e-06,
+      "loss": 3.739,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1400000000000003e-06,
+      "loss": 3.6578,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.16e-06,
+      "loss": 3.6617,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 3.6966,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2e-06,
+      "loss": 3.5697,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2200000000000003e-06,
+      "loss": 3.6827,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.24e-06,
+      "loss": 3.8,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2600000000000004e-06,
+      "loss": 3.7381,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.28e-06,
+      "loss": 3.7312,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 3.6106,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3200000000000002e-06,
+      "loss": 3.6731,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3400000000000005e-06,
+      "loss": 3.8287,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3600000000000003e-06,
+      "loss": 3.8317,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.38e-06,
+      "loss": 3.6784,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 3.8157,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.42e-06,
+      "loss": 3.7599,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4400000000000004e-06,
+      "loss": 3.6423,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.46e-06,
+      "loss": 3.6835,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 3.7004,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5e-06,
+      "loss": 3.8636,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.52e-06,
+      "loss": 3.6997,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5400000000000002e-06,
+      "loss": 3.7938,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.56e-06,
+      "loss": 3.7425,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 3.7353,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6e-06,
+      "loss": 3.5756,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6200000000000003e-06,
+      "loss": 3.6779,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.64e-06,
+      "loss": 3.7299,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6600000000000004e-06,
+      "loss": 3.7309,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.68e-06,
+      "loss": 3.6596,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 3.6988,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 3.7561,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7400000000000004e-06,
+      "loss": 3.6732,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7600000000000003e-06,
+      "loss": 3.6268,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 3.6154,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 3.7671,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.82e-06,
+      "loss": 3.7499,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.84e-06,
+      "loss": 3.7063,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.86e-06,
+      "loss": 3.7124,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.88e-06,
+      "loss": 3.6024,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9e-06,
+      "loss": 3.658,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.92e-06,
+      "loss": 3.6726,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 3.7112,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.96e-06,
+      "loss": 3.7108,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 3.7885,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3e-06,
+      "loss": 3.62,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0200000000000003e-06,
+      "loss": 3.6166,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.04e-06,
+      "loss": 3.7115,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0600000000000003e-06,
+      "loss": 3.6244,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.08e-06,
+      "loss": 3.7023,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 3.6993,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.12e-06,
+      "loss": 3.8328,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1400000000000004e-06,
+      "loss": 3.8162,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1600000000000002e-06,
+      "loss": 3.7715,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 3.69,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 3.7097,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2200000000000005e-06,
+      "loss": 3.709,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2400000000000003e-06,
+      "loss": 3.824,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2600000000000006e-06,
+      "loss": 3.7282,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 3.6195,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 3.7223,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 3.7112,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3400000000000006e-06,
+      "loss": 3.7253,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 3.7034,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 3.6619,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 3.7521,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4200000000000007e-06,
+      "loss": 3.6692,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.44e-06,
+      "loss": 3.6776,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.46e-06,
+      "loss": 3.726,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.48e-06,
+      "loss": 3.709,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5e-06,
+      "loss": 3.8562,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.52e-06,
+      "loss": 3.6923,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.54e-06,
+      "loss": 3.6839,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5600000000000002e-06,
+      "loss": 3.7267,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.58e-06,
+      "loss": 3.6352,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 3.6883,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.62e-06,
+      "loss": 3.7499,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6400000000000003e-06,
+      "loss": 3.6614,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.66e-06,
+      "loss": 3.5608,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 3.7707,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7e-06,
+      "loss": 3.7697,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7200000000000004e-06,
+      "loss": 3.6502,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.74e-06,
+      "loss": 3.7087,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 3.6289,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 3.5543,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 3.7225,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.820000000000001e-06,
+      "loss": 3.7115,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 3.6999,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.86e-06,
+      "loss": 3.5778,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.88e-06,
+      "loss": 3.6737,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 3.6318,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 3.7335,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.94e-06,
+      "loss": 3.6034,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.96e-06,
+      "loss": 3.7202,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 3.8494,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 3.6975,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 3.7161,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.04e-06,
+      "loss": 3.8157,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.060000000000001e-06,
+      "loss": 3.7058,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.08e-06,
+      "loss": 3.683,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1e-06,
+      "loss": 3.7003,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.12e-06,
+      "loss": 3.6889,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.14e-06,
+      "loss": 3.8366,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.16e-06,
+      "loss": 3.7216,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.18e-06,
+      "loss": 3.7108,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 3.5531,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.22e-06,
+      "loss": 3.7393,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.24e-06,
+      "loss": 3.6408,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.26e-06,
+      "loss": 3.6381,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.2800000000000005e-06,
+      "loss": 3.744,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3e-06,
+      "loss": 3.5259,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.32e-06,
+      "loss": 3.7271,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.34e-06,
+      "loss": 3.6933,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.360000000000001e-06,
+      "loss": 3.815,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.38e-06,
+      "loss": 3.8238,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4e-06,
+      "loss": 3.5751,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.42e-06,
+      "loss": 3.6139,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.440000000000001e-06,
+      "loss": 3.7663,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4600000000000005e-06,
+      "loss": 3.6863,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.48e-06,
+      "loss": 3.6873,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.5e-06,
+      "loss": 3.6822,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.520000000000001e-06,
+      "loss": 3.6825,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.540000000000001e-06,
+      "loss": 3.777,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.56e-06,
+      "loss": 3.7112,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.58e-06,
+      "loss": 3.6939,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 3.5372,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.620000000000001e-06,
+      "loss": 3.7515,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 3.5924,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.66e-06,
+      "loss": 3.7867,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.680000000000001e-06,
+      "loss": 3.6556,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7e-06,
+      "loss": 3.6705,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.7200000000000005e-06,
+      "loss": 3.7444,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.74e-06,
+      "loss": 3.7247,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.76e-06,
+      "loss": 3.7849,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.78e-06,
+      "loss": 3.7544,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 3.6284,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.8200000000000004e-06,
+      "loss": 3.7147,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.84e-06,
+      "loss": 3.6847,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.86e-06,
+      "loss": 3.6671,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 3.8216,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 3.6799,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.92e-06,
+      "loss": 3.574,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.94e-06,
+      "loss": 3.642,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 3.6624,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 3.696,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5e-06,
+      "loss": 3.7046,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.02e-06,
+      "loss": 3.711,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.04e-06,
+      "loss": 3.5865,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.060000000000001e-06,
+      "loss": 3.6864,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.0800000000000005e-06,
+      "loss": 3.6452,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.1e-06,
+      "loss": 3.664,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.12e-06,
+      "loss": 3.7518,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.140000000000001e-06,
+      "loss": 3.6661,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.1600000000000006e-06,
+      "loss": 3.6521,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.18e-06,
+      "loss": 3.6577,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.2e-06,
+      "loss": 3.6319,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.220000000000001e-06,
+      "loss": 3.7182,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.240000000000001e-06,
+      "loss": 3.6795,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.2600000000000005e-06,
+      "loss": 3.7021,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.28e-06,
+      "loss": 3.716,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.300000000000001e-06,
+      "loss": 3.6465,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.320000000000001e-06,
+      "loss": 3.7094,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.3400000000000005e-06,
+      "loss": 3.6403,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.36e-06,
+      "loss": 3.6917,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 3.703,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 3.7314,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.420000000000001e-06,
+      "loss": 3.7071,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 3.6287,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.460000000000001e-06,
+      "loss": 3.7593,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.480000000000001e-06,
+      "loss": 3.7588,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 3.76,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.5200000000000005e-06,
+      "loss": 3.7399,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.540000000000001e-06,
+      "loss": 3.5348,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.560000000000001e-06,
+      "loss": 3.6004,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 3.6404,
+      "step": 279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 3.6494,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.620000000000001e-06,
+      "loss": 3.6538,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.64e-06,
+      "loss": 3.5969,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.66e-06,
+      "loss": 3.6161,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.68e-06,
+      "loss": 3.7744,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.7e-06,
+      "loss": 3.7664,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.72e-06,
+      "loss": 3.7166,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.74e-06,
+      "loss": 3.6913,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.76e-06,
+      "loss": 3.6691,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.78e-06,
+      "loss": 3.7664,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.8e-06,
+      "loss": 3.6351,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.82e-06,
+      "loss": 3.761,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.84e-06,
+      "loss": 3.7172,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.86e-06,
+      "loss": 3.8605,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 3.7306,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9e-06,
+      "loss": 3.7133,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.92e-06,
+      "loss": 3.7017,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.94e-06,
+      "loss": 3.7016,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9600000000000005e-06,
+      "loss": 3.6207,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.98e-06,
+      "loss": 3.7597,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6e-06,
+      "loss": 3.7478,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.02e-06,
+      "loss": 3.6985,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.040000000000001e-06,
+      "loss": 3.6472,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 3.7135,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.08e-06,
+      "loss": 3.6145,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.1e-06,
+      "loss": 3.7021,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.120000000000001e-06,
+      "loss": 3.5754,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.1400000000000005e-06,
+      "loss": 3.7552,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.16e-06,
+      "loss": 3.7747,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.18e-06,
+      "loss": 3.6341,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 3.6245,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.220000000000001e-06,
+      "loss": 3.7573,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.24e-06,
+      "loss": 3.6961,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.26e-06,
+      "loss": 3.7073,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.280000000000001e-06,
+      "loss": 3.6539,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 3.692,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.3200000000000005e-06,
+      "loss": 3.7108,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.34e-06,
+      "loss": 3.6453,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.360000000000001e-06,
+      "loss": 3.802,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 3.5307,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 3.7518,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.42e-06,
+      "loss": 3.6997,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.440000000000001e-06,
+      "loss": 3.6028,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.460000000000001e-06,
+      "loss": 3.7022,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.480000000000001e-06,
+      "loss": 3.5998,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 3.6564,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.520000000000001e-06,
+      "loss": 3.6648,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.540000000000001e-06,
+      "loss": 3.6462,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 3.7019,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 3.6225,
+      "step": 329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 3.6028,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.620000000000001e-06,
+      "loss": 3.7265,
+      "step": 331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.640000000000001e-06,
+      "loss": 3.6655,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 3.6633,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.680000000000001e-06,
+      "loss": 3.6321,
+      "step": 334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 3.6429,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 3.6445,
+      "step": 336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.740000000000001e-06,
+      "loss": 3.5363,
+      "step": 337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.760000000000001e-06,
+      "loss": 3.6301,
+      "step": 338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 3.6154,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 3.7919,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.820000000000001e-06,
+      "loss": 3.6914,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.8400000000000014e-06,
+      "loss": 3.7509,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.860000000000001e-06,
+      "loss": 3.5597,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.88e-06,
+      "loss": 3.7706,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.9e-06,
+      "loss": 3.6204,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.92e-06,
+      "loss": 3.6895,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.9400000000000005e-06,
+      "loss": 3.6541,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.96e-06,
+      "loss": 3.6123,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.98e-06,
+      "loss": 3.6628,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7e-06,
+      "loss": 3.6493,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 3.6007,
+      "step": 351
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.04e-06,
+      "loss": 3.7148,
+      "step": 352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.06e-06,
+      "loss": 3.6571,
+      "step": 353
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.08e-06,
+      "loss": 3.5943,
+      "step": 354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.100000000000001e-06,
+      "loss": 3.7256,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.1200000000000004e-06,
+      "loss": 3.7247,
+      "step": 356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.14e-06,
+      "loss": 3.6752,
+      "step": 357
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.16e-06,
+      "loss": 3.6238,
+      "step": 358
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 3.6614,
+      "step": 359
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 3.7225,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.22e-06,
+      "loss": 3.675,
+      "step": 361
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.24e-06,
+      "loss": 3.7627,
+      "step": 362
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.260000000000001e-06,
+      "loss": 3.6232,
+      "step": 363
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.280000000000001e-06,
+      "loss": 3.6208,
+      "step": 364
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.3e-06,
+      "loss": 3.6779,
+      "step": 365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.32e-06,
+      "loss": 3.6886,
+      "step": 366
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.340000000000001e-06,
+      "loss": 3.6142,
+      "step": 367
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.360000000000001e-06,
+      "loss": 3.6737,
+      "step": 368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 3.5672,
+      "step": 369
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.4e-06,
+      "loss": 3.8012,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.420000000000001e-06,
+      "loss": 3.7008,
+      "step": 371
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.440000000000001e-06,
+      "loss": 3.6269,
+      "step": 372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.4600000000000006e-06,
+      "loss": 3.5628,
+      "step": 373
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.48e-06,
+      "loss": 3.644,
+      "step": 374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 3.61,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.520000000000001e-06,
+      "loss": 3.6147,
+      "step": 376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.540000000000001e-06,
+      "loss": 3.6291,
+      "step": 377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 3.6051,
+      "step": 378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.58e-06,
+      "loss": 3.756,
+      "step": 379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 3.5986,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.620000000000001e-06,
+      "loss": 3.6298,
+      "step": 381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.640000000000001e-06,
+      "loss": 3.6858,
+      "step": 382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.660000000000001e-06,
+      "loss": 3.662,
+      "step": 383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 3.6027,
+      "step": 384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.7e-06,
+      "loss": 3.7757,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.72e-06,
+      "loss": 3.6737,
+      "step": 386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.74e-06,
+      "loss": 3.6904,
+      "step": 387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.76e-06,
+      "loss": 3.7668,
+      "step": 388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.78e-06,
+      "loss": 3.6595,
+      "step": 389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 3.5439,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.820000000000001e-06,
+      "loss": 3.6709,
+      "step": 391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.840000000000001e-06,
+      "loss": 3.7002,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 3.8076,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.88e-06,
+      "loss": 3.5483,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.9e-06,
+      "loss": 3.7504,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.92e-06,
+      "loss": 3.6307,
+      "step": 396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.94e-06,
+      "loss": 3.5808,
+      "step": 397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.960000000000002e-06,
+      "loss": 3.7508,
+      "step": 398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 3.6698,
+      "step": 399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 3.6577,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.020000000000001e-06,
+      "loss": 3.7578,
+      "step": 401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 3.7194,
+      "step": 402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.06e-06,
+      "loss": 3.7681,
+      "step": 403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.08e-06,
+      "loss": 3.6419,
+      "step": 404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.1e-06,
+      "loss": 3.6597,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.120000000000002e-06,
+      "loss": 3.5929,
+      "step": 406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.14e-06,
+      "loss": 3.5638,
+      "step": 407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.16e-06,
+      "loss": 3.6658,
+      "step": 408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.18e-06,
+      "loss": 3.6207,
+      "step": 409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.2e-06,
+      "loss": 3.6345,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 3.5852,
+      "step": 411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.24e-06,
+      "loss": 3.7024,
+      "step": 412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.26e-06,
+      "loss": 3.7079,
+      "step": 413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.28e-06,
+      "loss": 3.6686,
+      "step": 414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.3e-06,
+      "loss": 3.7111,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.32e-06,
+      "loss": 3.6908,
+      "step": 416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.34e-06,
+      "loss": 3.6306,
+      "step": 417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.36e-06,
+      "loss": 3.6615,
+      "step": 418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 3.6186,
+      "step": 419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 3.5919,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.42e-06,
+      "loss": 3.5195,
+      "step": 421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.44e-06,
+      "loss": 3.597,
+      "step": 422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.46e-06,
+      "loss": 3.6815,
+      "step": 423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.48e-06,
+      "loss": 3.7456,
+      "step": 424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.5e-06,
+      "loss": 3.6235,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.52e-06,
+      "loss": 3.6166,
+      "step": 426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.540000000000001e-06,
+      "loss": 3.6313,
+      "step": 427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.560000000000001e-06,
+      "loss": 3.6517,
+      "step": 428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 3.691,
+      "step": 429
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.6e-06,
+      "loss": 3.6145,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.62e-06,
+      "loss": 3.6483,
+      "step": 431
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.64e-06,
+      "loss": 3.6634,
+      "step": 432
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.66e-06,
+      "loss": 3.6325,
+      "step": 433
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.68e-06,
+      "loss": 3.5314,
+      "step": 434
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.700000000000001e-06,
+      "loss": 3.6029,
+      "step": 435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.720000000000001e-06,
+      "loss": 3.6764,
+      "step": 436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.740000000000001e-06,
+      "loss": 3.7139,
+      "step": 437
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.76e-06,
+      "loss": 3.6814,
+      "step": 438
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.78e-06,
+      "loss": 3.7331,
+      "step": 439
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.8e-06,
+      "loss": 3.6442,
+      "step": 440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.82e-06,
+      "loss": 3.58,
+      "step": 441
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.84e-06,
+      "loss": 3.6461,
+      "step": 442
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.860000000000002e-06,
+      "loss": 3.583,
+      "step": 443
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.880000000000001e-06,
+      "loss": 3.6785,
+      "step": 444
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 3.7894,
+      "step": 445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.920000000000001e-06,
+      "loss": 3.6404,
+      "step": 446
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.94e-06,
+      "loss": 3.7158,
+      "step": 447
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.96e-06,
+      "loss": 3.6746,
+      "step": 448
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.98e-06,
+      "loss": 3.5432,
+      "step": 449
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9e-06,
+      "loss": 3.7641,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.020000000000002e-06,
+      "loss": 3.6467,
+      "step": 451
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.040000000000002e-06,
+      "loss": 3.6479,
+      "step": 452
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.060000000000001e-06,
+      "loss": 3.6096,
+      "step": 453
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.080000000000001e-06,
+      "loss": 3.6277,
+      "step": 454
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 3.6894,
+      "step": 455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.12e-06,
+      "loss": 3.689,
+      "step": 456
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.14e-06,
+      "loss": 3.6434,
+      "step": 457
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.16e-06,
+      "loss": 3.6957,
+      "step": 458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 3.5962,
+      "step": 459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 3.5787,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.220000000000002e-06,
+      "loss": 3.6679,
+      "step": 461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.240000000000001e-06,
+      "loss": 3.714,
+      "step": 462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.260000000000001e-06,
+      "loss": 3.6259,
+      "step": 463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 3.5997,
+      "step": 464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.3e-06,
+      "loss": 3.7637,
+      "step": 465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.32e-06,
+      "loss": 3.7078,
+      "step": 466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.340000000000002e-06,
+      "loss": 3.6296,
+      "step": 467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.360000000000002e-06,
+      "loss": 3.7694,
+      "step": 468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.38e-06,
+      "loss": 3.598,
+      "step": 469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.4e-06,
+      "loss": 3.5684,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.42e-06,
+      "loss": 3.5274,
+      "step": 471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.440000000000001e-06,
+      "loss": 3.6267,
+      "step": 472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.460000000000001e-06,
+      "loss": 3.6123,
+      "step": 473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.48e-06,
+      "loss": 3.6258,
+      "step": 474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.5e-06,
+      "loss": 3.63,
+      "step": 475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.52e-06,
+      "loss": 3.682,
+      "step": 476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.54e-06,
+      "loss": 3.6886,
+      "step": 477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.56e-06,
+      "loss": 3.5351,
+      "step": 478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.58e-06,
+      "loss": 3.6671,
+      "step": 479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 3.695,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.620000000000001e-06,
+      "loss": 3.6086,
+      "step": 481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.640000000000001e-06,
+      "loss": 3.6258,
+      "step": 482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.66e-06,
+      "loss": 3.6744,
+      "step": 483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.68e-06,
+      "loss": 3.5658,
+      "step": 484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.7e-06,
+      "loss": 3.6056,
+      "step": 485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.72e-06,
+      "loss": 3.6685,
+      "step": 486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.74e-06,
+      "loss": 3.6743,
+      "step": 487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 3.6686,
+      "step": 488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 3.6672,
+      "step": 489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 3.6602,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.820000000000001e-06,
+      "loss": 3.6479,
+      "step": 491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.84e-06,
+      "loss": 3.6234,
+      "step": 492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.86e-06,
+      "loss": 3.6914,
+      "step": 493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.88e-06,
+      "loss": 3.6637,
+      "step": 494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9e-06,
+      "loss": 3.5455,
+      "step": 495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 3.6607,
+      "step": 496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.940000000000001e-06,
+      "loss": 3.6905,
+      "step": 497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 3.7037,
+      "step": 498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 3.7422,
+      "step": 499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1e-05,
+      "loss": 3.6056,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999999596619244e-06,
+      "loss": 3.5949,
+      "step": 501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99999838647704e-06,
+      "loss": 3.655,
+      "step": 502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99999636957358e-06,
+      "loss": 3.5984,
+      "step": 503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999993545909194e-06,
+      "loss": 3.6571,
+      "step": 504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999989915484335e-06,
+      "loss": 3.6714,
+      "step": 505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99998547829959e-06,
+      "loss": 3.5893,
+      "step": 506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999980234355674e-06,
+      "loss": 3.6745,
+      "step": 507
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999974183653434e-06,
+      "loss": 3.658,
+      "step": 508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999967326193847e-06,
+      "loss": 3.5851,
+      "step": 509
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999959661978016e-06,
+      "loss": 3.6142,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999951191007182e-06,
+      "loss": 3.5839,
+      "step": 511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999941913282709e-06,
+      "loss": 3.5707,
+      "step": 512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999931828806095e-06,
+      "loss": 3.7773,
+      "step": 513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999920937578967e-06,
+      "loss": 3.6575,
+      "step": 514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999909239603081e-06,
+      "loss": 3.7726,
+      "step": 515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999896734880328e-06,
+      "loss": 3.5407,
+      "step": 516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999883423412721e-06,
+      "loss": 3.6638,
+      "step": 517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999869305202413e-06,
+      "loss": 3.676,
+      "step": 518
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999854380251679e-06,
+      "loss": 3.566,
+      "step": 519
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999838648562925e-06,
+      "loss": 3.6932,
+      "step": 520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999822110138694e-06,
+      "loss": 3.5625,
+      "step": 521
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999804764981651e-06,
+      "loss": 3.5221,
+      "step": 522
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999786613094597e-06,
+      "loss": 3.555,
+      "step": 523
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99976765448046e-06,
+      "loss": 3.5745,
+      "step": 524
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999747889142298e-06,
+      "loss": 3.6221,
+      "step": 525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999727317083301e-06,
+      "loss": 3.5763,
+      "step": 526
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999705938306789e-06,
+      "loss": 3.7183,
+      "step": 527
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99968375281621e-06,
+      "loss": 3.7005,
+      "step": 528
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999660760615147e-06,
+      "loss": 3.6939,
+      "step": 529
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999636961707305e-06,
+      "loss": 3.6118,
+      "step": 530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999612356096528e-06,
+      "loss": 3.6223,
+      "step": 531
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999586943786783e-06,
+      "loss": 3.7223,
+      "step": 532
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999560724782173e-06,
+      "loss": 3.6247,
+      "step": 533
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999533699086927e-06,
+      "loss": 3.7595,
+      "step": 534
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999505866705406e-06,
+      "loss": 3.4945,
+      "step": 535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999477227642102e-06,
+      "loss": 3.7185,
+      "step": 536
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999447781901634e-06,
+      "loss": 3.6087,
+      "step": 537
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999417529488755e-06,
+      "loss": 3.6554,
+      "step": 538
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999386470408344e-06,
+      "loss": 3.6007,
+      "step": 539
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.999354604665415e-06,
+      "loss": 3.7308,
+      "step": 540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999321932265107e-06,
+      "loss": 3.5735,
+      "step": 541
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999288453212695e-06,
+      "loss": 3.6251,
+      "step": 542
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999254167513576e-06,
+      "loss": 3.5402,
+      "step": 543
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999219075173289e-06,
+      "loss": 3.6788,
+      "step": 544
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99918317619749e-06,
+      "loss": 3.7838,
+      "step": 545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999146470591974e-06,
+      "loss": 3.7031,
+      "step": 546
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999108958362664e-06,
+      "loss": 3.7007,
+      "step": 547
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999070639515613e-06,
+      "loss": 3.6615,
+      "step": 548
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.999031514057001e-06,
+      "loss": 3.6618,
+      "step": 549
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998991581993144e-06,
+      "loss": 3.704,
+      "step": 550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998950843330483e-06,
+      "loss": 3.6194,
+      "step": 551
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998909298075593e-06,
+      "loss": 3.6171,
+      "step": 552
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998866946235176e-06,
+      "loss": 3.6128,
+      "step": 553
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998823787816067e-06,
+      "loss": 3.6685,
+      "step": 554
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99877982282523e-06,
+      "loss": 3.6085,
+      "step": 555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998735051269754e-06,
+      "loss": 3.6886,
+      "step": 556
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99868947315687e-06,
+      "loss": 3.6056,
+      "step": 557
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998643088493927e-06,
+      "loss": 3.5715,
+      "step": 558
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998595897288412e-06,
+      "loss": 3.657,
+      "step": 559
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998547899547939e-06,
+      "loss": 3.7192,
+      "step": 560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998499095280253e-06,
+      "loss": 3.5931,
+      "step": 561
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998449484493225e-06,
+      "loss": 3.617,
+      "step": 562
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998399067194864e-06,
+      "loss": 3.5786,
+      "step": 563
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998347843393305e-06,
+      "loss": 3.6231,
+      "step": 564
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99829581309681e-06,
+      "loss": 3.7829,
+      "step": 565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998242976313777e-06,
+      "loss": 3.6449,
+      "step": 566
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99818933305273e-06,
+      "loss": 3.7234,
+      "step": 567
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998134883322324e-06,
+      "loss": 3.5461,
+      "step": 568
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.998079627131345e-06,
+      "loss": 3.7164,
+      "step": 569
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99802356448871e-06,
+      "loss": 3.7565,
+      "step": 570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997966695403462e-06,
+      "loss": 3.7649,
+      "step": 571
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997909019884782e-06,
+      "loss": 3.6295,
+      "step": 572
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99785053794197e-06,
+      "loss": 3.6652,
+      "step": 573
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997791249584465e-06,
+      "loss": 3.5949,
+      "step": 574
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997731154821835e-06,
+      "loss": 3.6417,
+      "step": 575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997670253663774e-06,
+      "loss": 3.628,
+      "step": 576
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99760854612011e-06,
+      "loss": 3.6648,
+      "step": 577
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997546032200797e-06,
+      "loss": 3.6291,
+      "step": 578
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997482711915926e-06,
+      "loss": 3.6709,
+      "step": 579
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99741858527571e-06,
+      "loss": 3.6064,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9973536522905e-06,
+      "loss": 3.7636,
+      "step": 581
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997287912970769e-06,
+      "loss": 3.6099,
+      "step": 582
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997221367327126e-06,
+      "loss": 3.5913,
+      "step": 583
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997154015370308e-06,
+      "loss": 3.6595,
+      "step": 584
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997085857111181e-06,
+      "loss": 3.5621,
+      "step": 585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.997016892560748e-06,
+      "loss": 3.6683,
+      "step": 586
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99694712173013e-06,
+      "loss": 3.6334,
+      "step": 587
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996876544630588e-06,
+      "loss": 3.5948,
+      "step": 588
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996805161273509e-06,
+      "loss": 3.5673,
+      "step": 589
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99673297167041e-06,
+      "loss": 3.6598,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996659975832941e-06,
+      "loss": 3.674,
+      "step": 591
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996586173772877e-06,
+      "loss": 3.6888,
+      "step": 592
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99651156550213e-06,
+      "loss": 3.6469,
+      "step": 593
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996436151032737e-06,
+      "loss": 3.7105,
+      "step": 594
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996359930376865e-06,
+      "loss": 3.6695,
+      "step": 595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996282903546813e-06,
+      "loss": 3.6471,
+      "step": 596
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.996205070555008e-06,
+      "loss": 3.7825,
+      "step": 597
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99612643141401e-06,
+      "loss": 3.6905,
+      "step": 598
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99604698613651e-06,
+      "loss": 3.6774,
+      "step": 599
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995966734735322e-06,
+      "loss": 3.657,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9958856772234e-06,
+      "loss": 3.6387,
+      "step": 601
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995803813613816e-06,
+      "loss": 3.6629,
+      "step": 602
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995721143919785e-06,
+      "loss": 3.6362,
+      "step": 603
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995637668154645e-06,
+      "loss": 3.6771,
+      "step": 604
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99555338633186e-06,
+      "loss": 3.5768,
+      "step": 605
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995468298465036e-06,
+      "loss": 3.6711,
+      "step": 606
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995382404567898e-06,
+      "loss": 3.6739,
+      "step": 607
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995295704654304e-06,
+      "loss": 3.647,
+      "step": 608
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995208198738248e-06,
+      "loss": 3.6679,
+      "step": 609
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995119886833845e-06,
+      "loss": 3.6761,
+      "step": 610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.995030768955348e-06,
+      "loss": 3.6322,
+      "step": 611
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99494084511713e-06,
+      "loss": 3.6384,
+      "step": 612
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994850115333708e-06,
+      "loss": 3.7048,
+      "step": 613
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994758579619717e-06,
+      "loss": 3.6216,
+      "step": 614
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994666237989929e-06,
+      "loss": 3.6448,
+      "step": 615
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994573090459241e-06,
+      "loss": 3.6673,
+      "step": 616
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994479137042683e-06,
+      "loss": 3.5211,
+      "step": 617
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994384377755417e-06,
+      "loss": 3.6579,
+      "step": 618
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.99428881261273e-06,
+      "loss": 3.6371,
+      "step": 619
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994192441630044e-06,
+      "loss": 3.5891,
+      "step": 620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.994095264822906e-06,
+      "loss": 3.5527,
+      "step": 621
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993997282206997e-06,
+      "loss": 3.7353,
+      "step": 622
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993898493798127e-06,
+      "loss": 3.5739,
+      "step": 623
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.993798899612235e-06,
+      "loss": 3.7003,
+      "step": 624
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.993698499665391e-06,
+      "loss": 3.5968,
+      "step": 625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.993597293973796e-06,
+      "loss": 3.682,
+      "step": 626
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.993495282553777e-06,
+      "loss": 3.5719,
+      "step": 627
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.993392465421799e-06,
+      "loss": 3.6166,
+      "step": 628
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.993288842594446e-06,
+      "loss": 3.6971,
+      "step": 629
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.993184414088439e-06,
+      "loss": 3.653,
+      "step": 630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.99307917992063e-06,
+      "loss": 3.7134,
+      "step": 631
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992973140107998e-06,
+      "loss": 3.5412,
+      "step": 632
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992866294667652e-06,
+      "loss": 3.604,
+      "step": 633
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992758643616833e-06,
+      "loss": 3.5486,
+      "step": 634
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.99265018697291e-06,
+      "loss": 3.7385,
+      "step": 635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992540924753381e-06,
+      "loss": 3.6485,
+      "step": 636
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992430856975879e-06,
+      "loss": 3.7019,
+      "step": 637
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.99231998365816e-06,
+      "loss": 3.6636,
+      "step": 638
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992208304818117e-06,
+      "loss": 3.6036,
+      "step": 639
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.992095820473768e-06,
+      "loss": 3.7151,
+      "step": 640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991982530643264e-06,
+      "loss": 3.637,
+      "step": 641
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991868435344883e-06,
+      "loss": 3.5996,
+      "step": 642
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991753534597035e-06,
+      "loss": 3.598,
+      "step": 643
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991637828418259e-06,
+      "loss": 3.5974,
+      "step": 644
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991521316827226e-06,
+      "loss": 3.4904,
+      "step": 645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991403999842733e-06,
+      "loss": 3.4922,
+      "step": 646
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991285877483712e-06,
+      "loss": 3.7228,
+      "step": 647
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.991166949769222e-06,
+      "loss": 3.7068,
+      "step": 648
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.99104721671845e-06,
+      "loss": 3.6375,
+      "step": 649
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990926678350716e-06,
+      "loss": 3.6475,
+      "step": 650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.99080533468547e-06,
+      "loss": 3.58,
+      "step": 651
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990683185742292e-06,
+      "loss": 3.5296,
+      "step": 652
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990560231540889e-06,
+      "loss": 3.7257,
+      "step": 653
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990436472101101e-06,
+      "loss": 3.582,
+      "step": 654
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990311907442897e-06,
+      "loss": 3.4841,
+      "step": 655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990186537586375e-06,
+      "loss": 3.6985,
+      "step": 656
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.990060362551763e-06,
+      "loss": 3.6636,
+      "step": 657
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989933382359423e-06,
+      "loss": 3.5904,
+      "step": 658
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98980559702984e-06,
+      "loss": 3.6231,
+      "step": 659
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989677006583634e-06,
+      "loss": 3.6122,
+      "step": 660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989547611041554e-06,
+      "loss": 3.706,
+      "step": 661
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989417410424476e-06,
+      "loss": 3.7402,
+      "step": 662
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98928640475341e-06,
+      "loss": 3.6496,
+      "step": 663
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989154594049493e-06,
+      "loss": 3.5951,
+      "step": 664
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.989021978333996e-06,
+      "loss": 3.5752,
+      "step": 665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988888557628312e-06,
+      "loss": 3.5337,
+      "step": 666
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988754331953971e-06,
+      "loss": 3.5507,
+      "step": 667
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988619301332633e-06,
+      "loss": 3.6553,
+      "step": 668
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988483465786084e-06,
+      "loss": 3.6019,
+      "step": 669
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988346825336237e-06,
+      "loss": 3.5178,
+      "step": 670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988209380005145e-06,
+      "loss": 3.6081,
+      "step": 671
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.988071129814983e-06,
+      "loss": 3.6116,
+      "step": 672
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987932074788058e-06,
+      "loss": 3.695,
+      "step": 673
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987792214946807e-06,
+      "loss": 3.7126,
+      "step": 674
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987651550313798e-06,
+      "loss": 3.6056,
+      "step": 675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987510080911722e-06,
+      "loss": 3.5503,
+      "step": 676
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987367806763414e-06,
+      "loss": 3.6255,
+      "step": 677
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987224727891824e-06,
+      "loss": 3.5864,
+      "step": 678
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.987080844320041e-06,
+      "loss": 3.5838,
+      "step": 679
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986936156071278e-06,
+      "loss": 3.6649,
+      "step": 680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986790663168884e-06,
+      "loss": 3.6646,
+      "step": 681
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986644365636333e-06,
+      "loss": 3.7121,
+      "step": 682
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98649726349723e-06,
+      "loss": 3.6272,
+      "step": 683
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986349356775312e-06,
+      "loss": 3.603,
+      "step": 684
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986200645494444e-06,
+      "loss": 3.4967,
+      "step": 685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.986051129678618e-06,
+      "loss": 3.6435,
+      "step": 686
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985900809351962e-06,
+      "loss": 3.6732,
+      "step": 687
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985749684538728e-06,
+      "loss": 3.6507,
+      "step": 688
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985597755263304e-06,
+      "loss": 3.6384,
+      "step": 689
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9854450215502e-06,
+      "loss": 3.6715,
+      "step": 690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98529148342406e-06,
+      "loss": 3.587,
+      "step": 691
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.985137140909662e-06,
+      "loss": 3.5979,
+      "step": 692
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984981994031907e-06,
+      "loss": 3.5056,
+      "step": 693
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984826042815827e-06,
+      "loss": 3.6517,
+      "step": 694
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984669287286587e-06,
+      "loss": 3.6058,
+      "step": 695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984511727469479e-06,
+      "loss": 3.6204,
+      "step": 696
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984353363389927e-06,
+      "loss": 3.55,
+      "step": 697
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98419419507348e-06,
+      "loss": 3.6238,
+      "step": 698
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.984034222545824e-06,
+      "loss": 3.667,
+      "step": 699
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983873445832768e-06,
+      "loss": 3.615,
+      "step": 700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983711864960256e-06,
+      "loss": 3.7156,
+      "step": 701
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98354947995436e-06,
+      "loss": 3.6143,
+      "step": 702
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983386290841277e-06,
+      "loss": 3.654,
+      "step": 703
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983222297647342e-06,
+      "loss": 3.713,
+      "step": 704
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.983057500399013e-06,
+      "loss": 3.6809,
+      "step": 705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.982891899122883e-06,
+      "loss": 3.599,
+      "step": 706
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.98272549384567e-06,
+      "loss": 3.6289,
+      "step": 707
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.982558284594226e-06,
+      "loss": 3.6388,
+      "step": 708
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.982390271395528e-06,
+      "loss": 3.4682,
+      "step": 709
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.982221454276687e-06,
+      "loss": 3.6962,
+      "step": 710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.982051833264942e-06,
+      "loss": 3.6618,
+      "step": 711
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981881408387662e-06,
+      "loss": 3.6614,
+      "step": 712
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981710179672345e-06,
+      "loss": 3.6988,
+      "step": 713
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981538147146616e-06,
+      "loss": 3.7793,
+      "step": 714
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981365310838239e-06,
+      "loss": 3.6062,
+      "step": 715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981191670775097e-06,
+      "loss": 3.5544,
+      "step": 716
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.981017226985209e-06,
+      "loss": 3.6097,
+      "step": 717
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.980841979496721e-06,
+      "loss": 3.6198,
+      "step": 718
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.980665928337912e-06,
+      "loss": 3.6576,
+      "step": 719
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.980489073537183e-06,
+      "loss": 3.5918,
+      "step": 720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.980311415123077e-06,
+      "loss": 3.6698,
+      "step": 721
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.980132953124253e-06,
+      "loss": 3.6302,
+      "step": 722
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97995368756951e-06,
+      "loss": 3.684,
+      "step": 723
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.979773618487773e-06,
+      "loss": 3.5162,
+      "step": 724
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.979592745908096e-06,
+      "loss": 3.6525,
+      "step": 725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.979411069859662e-06,
+      "loss": 3.664,
+      "step": 726
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.979228590371785e-06,
+      "loss": 3.6995,
+      "step": 727
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.979045307473909e-06,
+      "loss": 3.5782,
+      "step": 728
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.978861221195609e-06,
+      "loss": 3.6217,
+      "step": 729
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.978676331566584e-06,
+      "loss": 3.6362,
+      "step": 730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.978490638616671e-06,
+      "loss": 3.6856,
+      "step": 731
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.978304142375828e-06,
+      "loss": 3.7141,
+      "step": 732
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.978116842874147e-06,
+      "loss": 3.6126,
+      "step": 733
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.977928740141851e-06,
+      "loss": 3.7301,
+      "step": 734
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97773983420929e-06,
+      "loss": 3.5918,
+      "step": 735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.977550125106943e-06,
+      "loss": 3.7103,
+      "step": 736
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.977359612865424e-06,
+      "loss": 3.7172,
+      "step": 737
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97716829751547e-06,
+      "loss": 3.5675,
+      "step": 738
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.976976179087948e-06,
+      "loss": 3.624,
+      "step": 739
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.976783257613859e-06,
+      "loss": 3.6096,
+      "step": 740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.976589533124333e-06,
+      "loss": 3.5896,
+      "step": 741
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.976395005650624e-06,
+      "loss": 3.5707,
+      "step": 742
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.976199675224123e-06,
+      "loss": 3.6516,
+      "step": 743
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.976003541876345e-06,
+      "loss": 3.5588,
+      "step": 744
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.975806605638939e-06,
+      "loss": 3.7175,
+      "step": 745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.975608866543677e-06,
+      "loss": 3.7018,
+      "step": 746
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.975410324622468e-06,
+      "loss": 3.6419,
+      "step": 747
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.975210979907344e-06,
+      "loss": 3.5875,
+      "step": 748
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.975010832430475e-06,
+      "loss": 3.5238,
+      "step": 749
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97480988222415e-06,
+      "loss": 3.6708,
+      "step": 750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.974608129320797e-06,
+      "loss": 3.5387,
+      "step": 751
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.974405573752965e-06,
+      "loss": 3.5525,
+      "step": 752
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97420221555334e-06,
+      "loss": 3.626,
+      "step": 753
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.973998054754733e-06,
+      "loss": 3.6791,
+      "step": 754
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.973793091390086e-06,
+      "loss": 3.6603,
+      "step": 755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.973587325492471e-06,
+      "loss": 3.5681,
+      "step": 756
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.973380757095088e-06,
+      "loss": 3.7506,
+      "step": 757
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.973173386231268e-06,
+      "loss": 3.6682,
+      "step": 758
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97296521293447e-06,
+      "loss": 3.6245,
+      "step": 759
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.972756237238284e-06,
+      "loss": 3.566,
+      "step": 760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.972546459176426e-06,
+      "loss": 3.5336,
+      "step": 761
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97233587878275e-06,
+      "loss": 3.6167,
+      "step": 762
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.972124496091227e-06,
+      "loss": 3.6481,
+      "step": 763
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.971912311135967e-06,
+      "loss": 3.61,
+      "step": 764
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.971699323951209e-06,
+      "loss": 3.6855,
+      "step": 765
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.971485534571314e-06,
+      "loss": 3.5437,
+      "step": 766
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.971270943030783e-06,
+      "loss": 3.6126,
+      "step": 767
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.971055549364237e-06,
+      "loss": 3.7255,
+      "step": 768
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.97083935360643e-06,
+      "loss": 3.5684,
+      "step": 769
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.970622355792248e-06,
+      "loss": 3.5738,
+      "step": 770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.970404555956702e-06,
+      "loss": 3.5064,
+      "step": 771
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.970185954134936e-06,
+      "loss": 3.6145,
+      "step": 772
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96996655036222e-06,
+      "loss": 3.5721,
+      "step": 773
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.969746344673959e-06,
+      "loss": 3.6386,
+      "step": 774
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.969525337105679e-06,
+      "loss": 3.5821,
+      "step": 775
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.969303527693043e-06,
+      "loss": 3.6102,
+      "step": 776
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96908091647184e-06,
+      "loss": 3.5283,
+      "step": 777
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.968857503477991e-06,
+      "loss": 3.608,
+      "step": 778
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96863328874754e-06,
+      "loss": 3.6187,
+      "step": 779
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.968408272316665e-06,
+      "loss": 3.4588,
+      "step": 780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.968182454221677e-06,
+      "loss": 3.7069,
+      "step": 781
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.967955834499008e-06,
+      "loss": 3.5737,
+      "step": 782
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.967728413185226e-06,
+      "loss": 3.7103,
+      "step": 783
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.967500190317025e-06,
+      "loss": 3.5306,
+      "step": 784
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96727116593123e-06,
+      "loss": 3.5047,
+      "step": 785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.967041340064794e-06,
+      "loss": 3.6077,
+      "step": 786
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.9668107127548e-06,
+      "loss": 3.5629,
+      "step": 787
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.96657928403846e-06,
+      "loss": 3.6149,
+      "step": 788
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.966347053953115e-06,
+      "loss": 3.646,
+      "step": 789
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.966114022536238e-06,
+      "loss": 3.5514,
+      "step": 790
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.965880189825427e-06,
+      "loss": 3.7216,
+      "step": 791
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.965645555858414e-06,
+      "loss": 3.6735,
+      "step": 792
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.965410120673054e-06,
+      "loss": 3.6193,
+      "step": 793
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.965173884307337e-06,
+      "loss": 3.5928,
+      "step": 794
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.964936846799382e-06,
+      "loss": 3.567,
+      "step": 795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.964699008187434e-06,
+      "loss": 3.6617,
+      "step": 796
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.964460368509868e-06,
+      "loss": 3.6684,
+      "step": 797
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.964220927805189e-06,
+      "loss": 3.6024,
+      "step": 798
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.963980686112034e-06,
+      "loss": 3.5846,
+      "step": 799
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.963739643469164e-06,
+      "loss": 3.5781,
+      "step": 800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.963497799915471e-06,
+      "loss": 3.6643,
+      "step": 801
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.963255155489979e-06,
+      "loss": 3.6978,
+      "step": 802
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.963011710231838e-06,
+      "loss": 3.6352,
+      "step": 803
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.96276746418033e-06,
+      "loss": 3.6928,
+      "step": 804
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.962522417374865e-06,
+      "loss": 3.5775,
+      "step": 805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.962276569854978e-06,
+      "loss": 3.608,
+      "step": 806
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.962029921660343e-06,
+      "loss": 3.6341,
+      "step": 807
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.961782472830753e-06,
+      "loss": 3.618,
+      "step": 808
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.961534223406133e-06,
+      "loss": 3.6309,
+      "step": 809
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.961285173426543e-06,
+      "loss": 3.6005,
+      "step": 810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.961035322932164e-06,
+      "loss": 3.6066,
+      "step": 811
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.960784671963313e-06,
+      "loss": 3.6948,
+      "step": 812
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.960533220560433e-06,
+      "loss": 3.5911,
+      "step": 813
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.960280968764093e-06,
+      "loss": 3.5471,
+      "step": 814
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.960027916614998e-06,
+      "loss": 3.5611,
+      "step": 815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.959774064153977e-06,
+      "loss": 3.65,
+      "step": 816
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.959519411421989e-06,
+      "loss": 3.5664,
+      "step": 817
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.959263958460125e-06,
+      "loss": 3.6151,
+      "step": 818
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.9590077053096e-06,
+      "loss": 3.5969,
+      "step": 819
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.958750652011763e-06,
+      "loss": 3.6774,
+      "step": 820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95849279860809e-06,
+      "loss": 3.7344,
+      "step": 821
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.958234145140187e-06,
+      "loss": 3.7209,
+      "step": 822
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.957974691649785e-06,
+      "loss": 3.6151,
+      "step": 823
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95771443817875e-06,
+      "loss": 3.6222,
+      "step": 824
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.957453384769076e-06,
+      "loss": 3.6068,
+      "step": 825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95719153146288e-06,
+      "loss": 3.6786,
+      "step": 826
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.956928878302418e-06,
+      "loss": 3.6255,
+      "step": 827
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.956665425330066e-06,
+      "loss": 3.7268,
+      "step": 828
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.956401172588333e-06,
+      "loss": 3.6159,
+      "step": 829
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.956136120119858e-06,
+      "loss": 3.8129,
+      "step": 830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.955870267967408e-06,
+      "loss": 3.6471,
+      "step": 831
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.955603616173879e-06,
+      "loss": 3.6918,
+      "step": 832
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.955336164782292e-06,
+      "loss": 3.6214,
+      "step": 833
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.955067913835807e-06,
+      "loss": 3.62,
+      "step": 834
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.954798863377704e-06,
+      "loss": 3.6778,
+      "step": 835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.954529013451393e-06,
+      "loss": 3.6493,
+      "step": 836
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.954258364100418e-06,
+      "loss": 3.5887,
+      "step": 837
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.953986915368448e-06,
+      "loss": 3.4474,
+      "step": 838
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.95371466729928e-06,
+      "loss": 3.5984,
+      "step": 839
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.953441619936844e-06,
+      "loss": 3.5578,
+      "step": 840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.953167773325197e-06,
+      "loss": 3.6485,
+      "step": 841
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.952893127508523e-06,
+      "loss": 3.6205,
+      "step": 842
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.952617682531138e-06,
+      "loss": 3.7129,
+      "step": 843
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.952341438437488e-06,
+      "loss": 3.5975,
+      "step": 844
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.952064395272139e-06,
+      "loss": 3.7722,
+      "step": 845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.951786553079799e-06,
+      "loss": 3.6677,
+      "step": 846
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.951507911905295e-06,
+      "loss": 3.6327,
+      "step": 847
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.951228471793588e-06,
+      "loss": 3.5549,
+      "step": 848
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.950948232789766e-06,
+      "loss": 3.6557,
+      "step": 849
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.950667194939045e-06,
+      "loss": 3.7141,
+      "step": 850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.950385358286772e-06,
+      "loss": 3.5041,
+      "step": 851
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.950102722878422e-06,
+      "loss": 3.649,
+      "step": 852
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.949819288759601e-06,
+      "loss": 3.6562,
+      "step": 853
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.949535055976038e-06,
+      "loss": 3.6673,
+      "step": 854
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.949250024573598e-06,
+      "loss": 3.5884,
+      "step": 855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.948964194598268e-06,
+      "loss": 3.6155,
+      "step": 856
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94867756609617e-06,
+      "loss": 3.6318,
+      "step": 857
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.948390139113551e-06,
+      "loss": 3.7503,
+      "step": 858
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.948101913696788e-06,
+      "loss": 3.5492,
+      "step": 859
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.947812889892387e-06,
+      "loss": 3.6471,
+      "step": 860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.947523067746983e-06,
+      "loss": 3.5344,
+      "step": 861
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94723244730734e-06,
+      "loss": 3.617,
+      "step": 862
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.946941028620349e-06,
+      "loss": 3.5731,
+      "step": 863
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.94664881173303e-06,
+      "loss": 3.7601,
+      "step": 864
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.946355796692538e-06,
+      "loss": 3.5882,
+      "step": 865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.946061983546145e-06,
+      "loss": 3.6013,
+      "step": 866
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.945767372341262e-06,
+      "loss": 3.596,
+      "step": 867
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.945471963125425e-06,
+      "loss": 3.5863,
+      "step": 868
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.9451757559463e-06,
+      "loss": 3.5865,
+      "step": 869
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.944878750851679e-06,
+      "loss": 3.6849,
+      "step": 870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.944580947889484e-06,
+      "loss": 3.5706,
+      "step": 871
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.944282347107767e-06,
+      "loss": 3.6757,
+      "step": 872
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.943982948554709e-06,
+      "loss": 3.7066,
+      "step": 873
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.943682752278616e-06,
+      "loss": 3.657,
+      "step": 874
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.943381758327928e-06,
+      "loss": 3.6756,
+      "step": 875
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.94307996675121e-06,
+      "loss": 3.6406,
+      "step": 876
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.942777377597157e-06,
+      "loss": 3.6285,
+      "step": 877
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.942473990914593e-06,
+      "loss": 3.5964,
+      "step": 878
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.942169806752468e-06,
+      "loss": 3.6531,
+      "step": 879
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.941864825159865e-06,
+      "loss": 3.669,
+      "step": 880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.941559046185992e-06,
+      "loss": 3.6621,
+      "step": 881
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.941252469880189e-06,
+      "loss": 3.5973,
+      "step": 882
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.940945096291922e-06,
+      "loss": 3.5612,
+      "step": 883
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.940636925470783e-06,
+      "loss": 3.5219,
+      "step": 884
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.940327957466502e-06,
+      "loss": 3.6684,
+      "step": 885
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.940018192328928e-06,
+      "loss": 3.6793,
+      "step": 886
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.939707630108045e-06,
+      "loss": 3.5493,
+      "step": 887
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.939396270853958e-06,
+      "loss": 3.6072,
+      "step": 888
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.939084114616911e-06,
+      "loss": 3.5877,
+      "step": 889
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.938771161447268e-06,
+      "loss": 3.7272,
+      "step": 890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.938457411395524e-06,
+      "loss": 3.5513,
+      "step": 891
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.938142864512306e-06,
+      "loss": 3.5645,
+      "step": 892
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.937827520848366e-06,
+      "loss": 3.6498,
+      "step": 893
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.937511380454585e-06,
+      "loss": 3.5047,
+      "step": 894
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.937194443381972e-06,
+      "loss": 3.5523,
+      "step": 895
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.936876709681668e-06,
+      "loss": 3.6301,
+      "step": 896
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.936558179404938e-06,
+      "loss": 3.5344,
+      "step": 897
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.936238852603179e-06,
+      "loss": 3.6719,
+      "step": 898
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.935918729327912e-06,
+      "loss": 3.6247,
+      "step": 899
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.935597809630795e-06,
+      "loss": 3.6561,
+      "step": 900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.935276093563604e-06,
+      "loss": 3.5549,
+      "step": 901
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.934953581178252e-06,
+      "loss": 3.6614,
+      "step": 902
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.934630272526776e-06,
+      "loss": 3.4169,
+      "step": 903
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.934306167661341e-06,
+      "loss": 3.7025,
+      "step": 904
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.933981266634244e-06,
+      "loss": 3.6834,
+      "step": 905
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.933655569497908e-06,
+      "loss": 3.6698,
+      "step": 906
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.933329076304885e-06,
+      "loss": 3.6624,
+      "step": 907
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.933001787107856e-06,
+      "loss": 3.6202,
+      "step": 908
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.932673701959629e-06,
+      "loss": 3.6506,
+      "step": 909
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.932344820913141e-06,
+      "loss": 3.55,
+      "step": 910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.932015144021458e-06,
+      "loss": 3.6954,
+      "step": 911
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.931684671337777e-06,
+      "loss": 3.6114,
+      "step": 912
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.931353402915415e-06,
+      "loss": 3.5896,
+      "step": 913
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.931021338807828e-06,
+      "loss": 3.7012,
+      "step": 914
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.930688479068593e-06,
+      "loss": 3.6433,
+      "step": 915
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.930354823751416e-06,
+      "loss": 3.7045,
+      "step": 916
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.930020372910137e-06,
+      "loss": 3.5615,
+      "step": 917
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.929685126598718e-06,
+      "loss": 3.6605,
+      "step": 918
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.929349084871252e-06,
+      "loss": 3.6228,
+      "step": 919
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.929012247781959e-06,
+      "loss": 3.6468,
+      "step": 920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.928674615385192e-06,
+      "loss": 3.7079,
+      "step": 921
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.928336187735425e-06,
+      "loss": 3.5576,
+      "step": 922
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.927996964887266e-06,
+      "loss": 3.6353,
+      "step": 923
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.927656946895448e-06,
+      "loss": 3.5936,
+      "step": 924
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.927316133814835e-06,
+      "loss": 3.5716,
+      "step": 925
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.926974525700418e-06,
+      "loss": 3.6768,
+      "step": 926
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.926632122607316e-06,
+      "loss": 3.6934,
+      "step": 927
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.926288924590776e-06,
+      "loss": 3.5576,
+      "step": 928
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.925944931706174e-06,
+      "loss": 3.7098,
+      "step": 929
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.925600144009013e-06,
+      "loss": 3.6249,
+      "step": 930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.92525456155493e-06,
+      "loss": 3.6312,
+      "step": 931
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.924908184399678e-06,
+      "loss": 3.6543,
+      "step": 932
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.924561012599152e-06,
+      "loss": 3.6079,
+      "step": 933
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.924213046209365e-06,
+      "loss": 3.6141,
+      "step": 934
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.923864285286466e-06,
+      "loss": 3.6542,
+      "step": 935
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.923514729886724e-06,
+      "loss": 3.5399,
+      "step": 936
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.923164380066546e-06,
+      "loss": 3.6414,
+      "step": 937
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.922813235882457e-06,
+      "loss": 3.4174,
+      "step": 938
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.922461297391115e-06,
+      "loss": 3.6486,
+      "step": 939
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.922108564649308e-06,
+      "loss": 3.6996,
+      "step": 940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.921755037713951e-06,
+      "loss": 3.5693,
+      "step": 941
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.921400716642087e-06,
+      "loss": 3.6419,
+      "step": 942
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.921045601490881e-06,
+      "loss": 3.6524,
+      "step": 943
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.920689692317637e-06,
+      "loss": 3.6226,
+      "step": 944
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.920332989179782e-06,
+      "loss": 3.5603,
+      "step": 945
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.919975492134867e-06,
+      "loss": 3.7235,
+      "step": 946
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.919617201240577e-06,
+      "loss": 3.6169,
+      "step": 947
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.919258116554724e-06,
+      "loss": 3.6389,
+      "step": 948
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.918898238135245e-06,
+      "loss": 3.6403,
+      "step": 949
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.918537566040209e-06,
+      "loss": 3.5654,
+      "step": 950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.91817610032781e-06,
+      "loss": 3.508,
+      "step": 951
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.917813841056373e-06,
+      "loss": 3.5951,
+      "step": 952
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.917450788284349e-06,
+      "loss": 3.6272,
+      "step": 953
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.917086942070315e-06,
+      "loss": 3.5657,
+      "step": 954
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.916722302472982e-06,
+      "loss": 3.5581,
+      "step": 955
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.916356869551183e-06,
+      "loss": 3.551,
+      "step": 956
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.91599064336388e-06,
+      "loss": 3.5605,
+      "step": 957
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.91562362397017e-06,
+      "loss": 3.5456,
+      "step": 958
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.915255811429267e-06,
+      "loss": 3.5741,
+      "step": 959
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.91488720580052e-06,
+      "loss": 3.4771,
+      "step": 960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.914517807143407e-06,
+      "loss": 3.751,
+      "step": 961
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.914147615517527e-06,
+      "loss": 3.635,
+      "step": 962
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.913776630982614e-06,
+      "loss": 3.6199,
+      "step": 963
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.913404853598525e-06,
+      "loss": 3.58,
+      "step": 964
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.91303228342525e-06,
+      "loss": 3.6407,
+      "step": 965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.912658920522902e-06,
+      "loss": 3.6083,
+      "step": 966
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.912284764951726e-06,
+      "loss": 3.595,
+      "step": 967
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.911909816772093e-06,
+      "loss": 3.5805,
+      "step": 968
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.911534076044497e-06,
+      "loss": 3.7272,
+      "step": 969
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.911157542829568e-06,
+      "loss": 3.5881,
+      "step": 970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.910780217188062e-06,
+      "loss": 3.5329,
+      "step": 971
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.91040209918086e-06,
+      "loss": 3.5791,
+      "step": 972
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.910023188868972e-06,
+      "loss": 3.5248,
+      "step": 973
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.909643486313533e-06,
+      "loss": 3.5654,
+      "step": 974
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.909262991575816e-06,
+      "loss": 3.6959,
+      "step": 975
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.908881704717208e-06,
+      "loss": 3.6105,
+      "step": 976
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.908499625799235e-06,
+      "loss": 3.5816,
+      "step": 977
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.908116754883543e-06,
+      "loss": 3.5959,
+      "step": 978
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.907733092031912e-06,
+      "loss": 3.6466,
+      "step": 979
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.907348637306245e-06,
+      "loss": 3.6454,
+      "step": 980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.906963390768573e-06,
+      "loss": 3.6994,
+      "step": 981
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.906577352481062e-06,
+      "loss": 3.4917,
+      "step": 982
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.906190522505993e-06,
+      "loss": 3.5691,
+      "step": 983
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.905802900905789e-06,
+      "loss": 3.5579,
+      "step": 984
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.905414487742987e-06,
+      "loss": 3.627,
+      "step": 985
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.905025283080265e-06,
+      "loss": 3.6899,
+      "step": 986
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.904635286980418e-06,
+      "loss": 3.5653,
+      "step": 987
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.904244499506372e-06,
+      "loss": 3.5133,
+      "step": 988
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.903852920721182e-06,
+      "loss": 3.523,
+      "step": 989
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.903460550688033e-06,
+      "loss": 3.5449,
+      "step": 990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.903067389470231e-06,
+      "loss": 3.5794,
+      "step": 991
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.902673437131219e-06,
+      "loss": 3.5493,
+      "step": 992
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.902278693734554e-06,
+      "loss": 3.5293,
+      "step": 993
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.901883159343936e-06,
+      "loss": 3.5686,
+      "step": 994
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.901486834023182e-06,
+      "loss": 3.6523,
+      "step": 995
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.901089717836239e-06,
+      "loss": 3.5573,
+      "step": 996
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.900691810847187e-06,
+      "loss": 3.5271,
+      "step": 997
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.900293113120223e-06,
+      "loss": 3.6023,
+      "step": 998
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.899893624719685e-06,
+      "loss": 3.6175,
+      "step": 999
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.899493345710024e-06,
+      "loss": 3.6414,
+      "step": 1000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.899092276155832e-06,
+      "loss": 3.5838,
+      "step": 1001
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.898690416121818e-06,
+      "loss": 3.6743,
+      "step": 1002
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.898287765672826e-06,
+      "loss": 3.5708,
+      "step": 1003
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.897884324873824e-06,
+      "loss": 3.6002,
+      "step": 1004
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.897480093789908e-06,
+      "loss": 3.6264,
+      "step": 1005
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.8970750724863e-06,
+      "loss": 3.5659,
+      "step": 1006
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.896669261028353e-06,
+      "loss": 3.5559,
+      "step": 1007
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.896262659481546e-06,
+      "loss": 3.5802,
+      "step": 1008
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.895855267911483e-06,
+      "loss": 3.638,
+      "step": 1009
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.8954470863839e-06,
+      "loss": 3.4966,
+      "step": 1010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.895038114964655e-06,
+      "loss": 3.7066,
+      "step": 1011
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.894628353719739e-06,
+      "loss": 3.6816,
+      "step": 1012
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.894217802715268e-06,
+      "loss": 3.5392,
+      "step": 1013
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.893806462017484e-06,
+      "loss": 3.7145,
+      "step": 1014
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.893394331692758e-06,
+      "loss": 3.6605,
+      "step": 1015
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.892981411807588e-06,
+      "loss": 3.7961,
+      "step": 1016
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.892567702428601e-06,
+      "loss": 3.6276,
+      "step": 1017
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.892153203622548e-06,
+      "loss": 3.6437,
+      "step": 1018
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.891737915456311e-06,
+      "loss": 3.7372,
+      "step": 1019
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.891321837996897e-06,
+      "loss": 3.7182,
+      "step": 1020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.890904971311442e-06,
+      "loss": 3.5315,
+      "step": 1021
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.890487315467205e-06,
+      "loss": 3.6564,
+      "step": 1022
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.890068870531579e-06,
+      "loss": 3.6539,
+      "step": 1023
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.889649636572082e-06,
+      "loss": 3.7104,
+      "step": 1024
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.889229613656356e-06,
+      "loss": 3.5846,
+      "step": 1025
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.888808801852172e-06,
+      "loss": 3.6891,
+      "step": 1026
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.888387201227431e-06,
+      "loss": 3.5349,
+      "step": 1027
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.887964811850159e-06,
+      "loss": 3.6009,
+      "step": 1028
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.887541633788508e-06,
+      "loss": 3.5667,
+      "step": 1029
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.887117667110761e-06,
+      "loss": 3.6578,
+      "step": 1030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.886692911885323e-06,
+      "loss": 3.5958,
+      "step": 1031
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.886267368180733e-06,
+      "loss": 3.7553,
+      "step": 1032
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.88584103606565e-06,
+      "loss": 3.5302,
+      "step": 1033
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.885413915608866e-06,
+      "loss": 3.6414,
+      "step": 1034
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.884986006879296e-06,
+      "loss": 3.6074,
+      "step": 1035
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.884557309945985e-06,
+      "loss": 3.7021,
+      "step": 1036
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.884127824878106e-06,
+      "loss": 3.6365,
+      "step": 1037
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.883697551744953e-06,
+      "loss": 3.5834,
+      "step": 1038
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.883266490615957e-06,
+      "loss": 3.5641,
+      "step": 1039
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.882834641560665e-06,
+      "loss": 3.5752,
+      "step": 1040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.882402004648762e-06,
+      "loss": 3.6483,
+      "step": 1041
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.881968579950053e-06,
+      "loss": 3.6788,
+      "step": 1042
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.88153436753447e-06,
+      "loss": 3.6587,
+      "step": 1043
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.881099367472076e-06,
+      "loss": 3.5708,
+      "step": 1044
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.88066357983306e-06,
+      "loss": 3.6623,
+      "step": 1045
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.880227004687737e-06,
+      "loss": 3.566,
+      "step": 1046
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.87978964210655e-06,
+      "loss": 3.6642,
+      "step": 1047
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.879351492160066e-06,
+      "loss": 3.6095,
+      "step": 1048
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.878912554918982e-06,
+      "loss": 3.5443,
+      "step": 1049
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.878472830454124e-06,
+      "loss": 3.6692,
+      "step": 1050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.87803231883644e-06,
+      "loss": 3.5974,
+      "step": 1051
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.87759102013701e-06,
+      "loss": 3.6007,
+      "step": 1052
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.877148934427037e-06,
+      "loss": 3.568,
+      "step": 1053
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.876706061777851e-06,
+      "loss": 3.5471,
+      "step": 1054
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.876262402260914e-06,
+      "loss": 3.6201,
+      "step": 1055
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.875817955947809e-06,
+      "loss": 3.68,
+      "step": 1056
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.875372722910247e-06,
+      "loss": 3.625,
+      "step": 1057
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.874926703220074e-06,
+      "loss": 3.6344,
+      "step": 1058
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.874479896949249e-06,
+      "loss": 3.7392,
+      "step": 1059
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.874032304169868e-06,
+      "loss": 3.5968,
+      "step": 1060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.873583924954152e-06,
+      "loss": 3.7178,
+      "step": 1061
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.873134759374448e-06,
+      "loss": 3.6268,
+      "step": 1062
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.872684807503226e-06,
+      "loss": 3.6304,
+      "step": 1063
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.872234069413093e-06,
+      "loss": 3.5302,
+      "step": 1064
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.871782545176773e-06,
+      "loss": 3.6319,
+      "step": 1065
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.87133023486712e-06,
+      "loss": 3.6158,
+      "step": 1066
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.870877138557115e-06,
+      "loss": 3.6291,
+      "step": 1067
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.87042325631987e-06,
+      "loss": 3.5808,
+      "step": 1068
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.869968588228616e-06,
+      "loss": 3.5891,
+      "step": 1069
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.869513134356717e-06,
+      "loss": 3.5249,
+      "step": 1070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.86905689477766e-06,
+      "loss": 3.6506,
+      "step": 1071
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.868599869565062e-06,
+      "loss": 3.5731,
+      "step": 1072
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.868142058792663e-06,
+      "loss": 3.7423,
+      "step": 1073
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.867683462534332e-06,
+      "loss": 3.6132,
+      "step": 1074
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.867224080864067e-06,
+      "loss": 3.5197,
+      "step": 1075
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.866763913855989e-06,
+      "loss": 3.6255,
+      "step": 1076
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.866302961584344e-06,
+      "loss": 3.6512,
+      "step": 1077
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.865841224123512e-06,
+      "loss": 3.6842,
+      "step": 1078
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.865378701547993e-06,
+      "loss": 3.6582,
+      "step": 1079
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.864915393932419e-06,
+      "loss": 3.5757,
+      "step": 1080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.864451301351544e-06,
+      "loss": 3.5595,
+      "step": 1081
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.863986423880248e-06,
+      "loss": 3.53,
+      "step": 1082
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.863520761593541e-06,
+      "loss": 3.5611,
+      "step": 1083
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.863054314566562e-06,
+      "loss": 3.6098,
+      "step": 1084
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.862587082874571e-06,
+      "loss": 3.4904,
+      "step": 1085
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.862119066592956e-06,
+      "loss": 3.6326,
+      "step": 1086
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.861650265797234e-06,
+      "loss": 3.545,
+      "step": 1087
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.861180680563046e-06,
+      "loss": 3.6722,
+      "step": 1088
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.860710310966162e-06,
+      "loss": 3.5747,
+      "step": 1089
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.860239157082476e-06,
+      "loss": 3.71,
+      "step": 1090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.85976721898801e-06,
+      "loss": 3.492,
+      "step": 1091
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.859294496758914e-06,
+      "loss": 3.5696,
+      "step": 1092
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.85882099047146e-06,
+      "loss": 3.5063,
+      "step": 1093
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.85834670020205e-06,
+      "loss": 3.6146,
+      "step": 1094
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.857871626027215e-06,
+      "loss": 3.7234,
+      "step": 1095
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.857395768023604e-06,
+      "loss": 3.5609,
+      "step": 1096
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.856919126268003e-06,
+      "loss": 3.5417,
+      "step": 1097
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.856441700837315e-06,
+      "loss": 3.6292,
+      "step": 1098
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.855963491808575e-06,
+      "loss": 3.5749,
+      "step": 1099
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.855484499258946e-06,
+      "loss": 3.5564,
+      "step": 1100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.855004723265712e-06,
+      "loss": 3.6027,
+      "step": 1101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.854524163906285e-06,
+      "loss": 3.6629,
+      "step": 1102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.854042821258205e-06,
+      "loss": 3.6072,
+      "step": 1103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.853560695399139e-06,
+      "loss": 3.6571,
+      "step": 1104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.85307778640688e-06,
+      "loss": 3.5141,
+      "step": 1105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.852594094359343e-06,
+      "loss": 3.5599,
+      "step": 1106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.852109619334575e-06,
+      "loss": 3.5546,
+      "step": 1107
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.851624361410745e-06,
+      "loss": 3.6487,
+      "step": 1108
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.851138320666156e-06,
+      "loss": 3.5045,
+      "step": 1109
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.850651497179226e-06,
+      "loss": 3.6402,
+      "step": 1110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.850163891028507e-06,
+      "loss": 3.6764,
+      "step": 1111
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.849675502292677e-06,
+      "loss": 3.5786,
+      "step": 1112
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.849186331050536e-06,
+      "loss": 3.6903,
+      "step": 1113
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.848696377381015e-06,
+      "loss": 3.7302,
+      "step": 1114
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.848205641363168e-06,
+      "loss": 3.6925,
+      "step": 1115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.847714123076175e-06,
+      "loss": 3.7257,
+      "step": 1116
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.847221822599347e-06,
+      "loss": 3.6444,
+      "step": 1117
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.846728740012117e-06,
+      "loss": 3.6791,
+      "step": 1118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.846234875394043e-06,
+      "loss": 3.6094,
+      "step": 1119
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.845740228824813e-06,
+      "loss": 3.6228,
+      "step": 1120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.845244800384238e-06,
+      "loss": 3.5478,
+      "step": 1121
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.844748590152258e-06,
+      "loss": 3.6911,
+      "step": 1122
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.844251598208936e-06,
+      "loss": 3.6109,
+      "step": 1123
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.843753824634464e-06,
+      "loss": 3.5574,
+      "step": 1124
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.843255269509158e-06,
+      "loss": 3.5466,
+      "step": 1125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.842755932913462e-06,
+      "loss": 3.7293,
+      "step": 1126
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.842255814927945e-06,
+      "loss": 3.6398,
+      "step": 1127
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.8417549156333e-06,
+      "loss": 3.6678,
+      "step": 1128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.841253235110352e-06,
+      "loss": 3.7132,
+      "step": 1129
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.840750773440047e-06,
+      "loss": 3.5197,
+      "step": 1130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.840247530703458e-06,
+      "loss": 3.5483,
+      "step": 1131
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.839743506981783e-06,
+      "loss": 3.5889,
+      "step": 1132
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.839238702356348e-06,
+      "loss": 3.6329,
+      "step": 1133
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.838733116908605e-06,
+      "loss": 3.5221,
+      "step": 1134
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.838226750720132e-06,
+      "loss": 3.7086,
+      "step": 1135
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.837719603872634e-06,
+      "loss": 3.5773,
+      "step": 1136
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.837211676447935e-06,
+      "loss": 3.5518,
+      "step": 1137
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.836702968527993e-06,
+      "loss": 3.672,
+      "step": 1138
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.83619348019489e-06,
+      "loss": 3.6157,
+      "step": 1139
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.835683211530834e-06,
+      "loss": 3.5659,
+      "step": 1140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.835172162618154e-06,
+      "loss": 3.7237,
+      "step": 1141
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.834660333539314e-06,
+      "loss": 3.6326,
+      "step": 1142
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.834147724376894e-06,
+      "loss": 3.6575,
+      "step": 1143
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.833634335213609e-06,
+      "loss": 3.566,
+      "step": 1144
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.833120166132293e-06,
+      "loss": 3.7155,
+      "step": 1145
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.832605217215909e-06,
+      "loss": 3.6484,
+      "step": 1146
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.832089488547545e-06,
+      "loss": 3.6084,
+      "step": 1147
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.831572980210414e-06,
+      "loss": 3.644,
+      "step": 1148
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.831055692287858e-06,
+      "loss": 3.458,
+      "step": 1149
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.83053762486334e-06,
+      "loss": 3.5695,
+      "step": 1150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.830018778020457e-06,
+      "loss": 3.5199,
+      "step": 1151
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.829499151842918e-06,
+      "loss": 3.5167,
+      "step": 1152
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.828978746414571e-06,
+      "loss": 3.7173,
+      "step": 1153
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.828457561819383e-06,
+      "loss": 3.6946,
+      "step": 1154
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.827935598141449e-06,
+      "loss": 3.6854,
+      "step": 1155
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.827412855464989e-06,
+      "loss": 3.6256,
+      "step": 1156
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.826889333874347e-06,
+      "loss": 3.5858,
+      "step": 1157
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.826365033453997e-06,
+      "loss": 3.6007,
+      "step": 1158
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.825839954288537e-06,
+      "loss": 3.5408,
+      "step": 1159
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.825314096462686e-06,
+      "loss": 3.6109,
+      "step": 1160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.824787460061294e-06,
+      "loss": 3.6526,
+      "step": 1161
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.824260045169335e-06,
+      "loss": 3.6623,
+      "step": 1162
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.823731851871911e-06,
+      "loss": 3.6915,
+      "step": 1163
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.823202880254243e-06,
+      "loss": 3.5194,
+      "step": 1164
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.822673130401684e-06,
+      "loss": 3.6137,
+      "step": 1165
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.822142602399712e-06,
+      "loss": 3.5164,
+      "step": 1166
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.821611296333924e-06,
+      "loss": 3.59,
+      "step": 1167
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.821079212290052e-06,
+      "loss": 3.6681,
+      "step": 1168
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.820546350353948e-06,
+      "loss": 3.5521,
+      "step": 1169
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.820012710611588e-06,
+      "loss": 3.5103,
+      "step": 1170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.81947829314908e-06,
+      "loss": 3.5807,
+      "step": 1171
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.818943098052652e-06,
+      "loss": 3.6513,
+      "step": 1172
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.818407125408655e-06,
+      "loss": 3.6425,
+      "step": 1173
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.817870375303576e-06,
+      "loss": 3.6021,
+      "step": 1174
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.817332847824018e-06,
+      "loss": 3.5944,
+      "step": 1175
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.81679454305671e-06,
+      "loss": 3.5931,
+      "step": 1176
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.81625546108851e-06,
+      "loss": 3.7043,
+      "step": 1177
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.815715602006403e-06,
+      "loss": 3.6366,
+      "step": 1178
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.815174965897495e-06,
+      "loss": 3.6443,
+      "step": 1179
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.814633552849017e-06,
+      "loss": 3.5109,
+      "step": 1180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.814091362948327e-06,
+      "loss": 3.5929,
+      "step": 1181
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.813548396282912e-06,
+      "loss": 3.5869,
+      "step": 1182
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.81300465294038e-06,
+      "loss": 3.6381,
+      "step": 1183
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.812460133008464e-06,
+      "loss": 3.5915,
+      "step": 1184
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.811914836575021e-06,
+      "loss": 3.496,
+      "step": 1185
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.811368763728041e-06,
+      "loss": 3.6162,
+      "step": 1186
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.81082191455563e-06,
+      "loss": 3.6058,
+      "step": 1187
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.810274289146027e-06,
+      "loss": 3.6341,
+      "step": 1188
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.809725887587589e-06,
+      "loss": 3.5941,
+      "step": 1189
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.809176709968803e-06,
+      "loss": 3.5188,
+      "step": 1190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.808626756378283e-06,
+      "loss": 3.5879,
+      "step": 1191
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.808076026904763e-06,
+      "loss": 3.5436,
+      "step": 1192
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.807524521637103e-06,
+      "loss": 3.6195,
+      "step": 1193
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.806972240664293e-06,
+      "loss": 3.6082,
+      "step": 1194
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.80641918407544e-06,
+      "loss": 3.5362,
+      "step": 1195
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.805865351959784e-06,
+      "loss": 3.5931,
+      "step": 1196
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.80531074440669e-06,
+      "loss": 3.5687,
+      "step": 1197
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.804755361505639e-06,
+      "loss": 3.622,
+      "step": 1198
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.804199203346247e-06,
+      "loss": 3.6007,
+      "step": 1199
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.803642270018251e-06,
+      "loss": 3.6187,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.803084561611513e-06,
+      "loss": 3.4323,
+      "step": 1201
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.802526078216021e-06,
+      "loss": 3.6458,
+      "step": 1202
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.801966819921888e-06,
+      "loss": 3.5522,
+      "step": 1203
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.801406786819352e-06,
+      "loss": 3.6361,
+      "step": 1204
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.800845978998772e-06,
+      "loss": 3.5654,
+      "step": 1205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.80028439655064e-06,
+      "loss": 3.6609,
+      "step": 1206
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.799722039565567e-06,
+      "loss": 3.7049,
+      "step": 1207
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.79915890813429e-06,
+      "loss": 3.5629,
+      "step": 1208
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.798595002347671e-06,
+      "loss": 3.5969,
+      "step": 1209
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.798030322296701e-06,
+      "loss": 3.5583,
+      "step": 1210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.797464868072489e-06,
+      "loss": 3.7678,
+      "step": 1211
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.796898639766271e-06,
+      "loss": 3.528,
+      "step": 1212
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.796331637469415e-06,
+      "loss": 3.5694,
+      "step": 1213
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.795763861273401e-06,
+      "loss": 3.7213,
+      "step": 1214
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.795195311269848e-06,
+      "loss": 3.6144,
+      "step": 1215
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.794625987550488e-06,
+      "loss": 3.5719,
+      "step": 1216
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.794055890207184e-06,
+      "loss": 3.591,
+      "step": 1217
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.793485019331924e-06,
+      "loss": 3.5138,
+      "step": 1218
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.792913375016817e-06,
+      "loss": 3.5878,
+      "step": 1219
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.7923409573541e-06,
+      "loss": 3.5828,
+      "step": 1220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.791767766436136e-06,
+      "loss": 3.6167,
+      "step": 1221
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.791193802355405e-06,
+      "loss": 3.6911,
+      "step": 1222
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.790619065204525e-06,
+      "loss": 3.6045,
+      "step": 1223
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.790043555076225e-06,
+      "loss": 3.5728,
+      "step": 1224
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.789467272063369e-06,
+      "loss": 3.6159,
+      "step": 1225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.78889021625894e-06,
+      "loss": 3.5917,
+      "step": 1226
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.788312387756046e-06,
+      "loss": 3.6322,
+      "step": 1227
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.787733786647922e-06,
+      "loss": 3.6081,
+      "step": 1228
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.787154413027926e-06,
+      "loss": 3.6784,
+      "step": 1229
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.786574266989543e-06,
+      "loss": 3.5946,
+      "step": 1230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.78599334862638e-06,
+      "loss": 3.6731,
+      "step": 1231
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.78541165803217e-06,
+      "loss": 3.5591,
+      "step": 1232
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.784829195300767e-06,
+      "loss": 3.6112,
+      "step": 1233
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.784245960526157e-06,
+      "loss": 3.6369,
+      "step": 1234
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.783661953802444e-06,
+      "loss": 3.5369,
+      "step": 1235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.78307717522386e-06,
+      "loss": 3.5729,
+      "step": 1236
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.782491624884759e-06,
+      "loss": 3.5858,
+      "step": 1237
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.781905302879621e-06,
+      "loss": 3.7804,
+      "step": 1238
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.781318209303052e-06,
+      "loss": 3.67,
+      "step": 1239
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.78073034424978e-06,
+      "loss": 3.6012,
+      "step": 1240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.780141707814658e-06,
+      "loss": 3.6532,
+      "step": 1241
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.779552300092664e-06,
+      "loss": 3.6443,
+      "step": 1242
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.7789621211789e-06,
+      "loss": 3.7162,
+      "step": 1243
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.778371171168594e-06,
+      "loss": 3.6666,
+      "step": 1244
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.777779450157097e-06,
+      "loss": 3.6643,
+      "step": 1245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.777186958239881e-06,
+      "loss": 3.6561,
+      "step": 1246
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.776593695512551e-06,
+      "loss": 3.6161,
+      "step": 1247
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.775999662070828e-06,
+      "loss": 3.7508,
+      "step": 1248
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.775404858010563e-06,
+      "loss": 3.5412,
+      "step": 1249
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.774809283427727e-06,
+      "loss": 3.5875,
+      "step": 1250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.774212938418417e-06,
+      "loss": 3.5246,
+      "step": 1251
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.773615823078857e-06,
+      "loss": 3.5074,
+      "step": 1252
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.773017937505391e-06,
+      "loss": 3.6009,
+      "step": 1253
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.77241928179449e-06,
+      "loss": 3.6139,
+      "step": 1254
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.771819856042749e-06,
+      "loss": 3.5004,
+      "step": 1255
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.771219660346885e-06,
+      "loss": 3.6785,
+      "step": 1256
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.770618694803744e-06,
+      "loss": 3.6094,
+      "step": 1257
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.77001695951029e-06,
+      "loss": 3.5282,
+      "step": 1258
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.769414454563614e-06,
+      "loss": 3.5869,
+      "step": 1259
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.768811180060935e-06,
+      "loss": 3.5768,
+      "step": 1260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.76820713609959e-06,
+      "loss": 3.4802,
+      "step": 1261
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.767602322777043e-06,
+      "loss": 3.6584,
+      "step": 1262
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.766996740190883e-06,
+      "loss": 3.5896,
+      "step": 1263
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.766390388438825e-06,
+      "loss": 3.491,
+      "step": 1264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.765783267618698e-06,
+      "loss": 3.5586,
+      "step": 1265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.76517537782847e-06,
+      "loss": 3.6197,
+      "step": 1266
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.764566719166219e-06,
+      "loss": 3.4932,
+      "step": 1267
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.763957291730157e-06,
+      "loss": 3.697,
+      "step": 1268
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.763347095618617e-06,
+      "loss": 3.6545,
+      "step": 1269
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.762736130930056e-06,
+      "loss": 3.5768,
+      "step": 1270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.762124397763048e-06,
+      "loss": 3.5285,
+      "step": 1271
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.761511896216306e-06,
+      "loss": 3.6466,
+      "step": 1272
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.760898626388655e-06,
+      "loss": 3.591,
+      "step": 1273
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.760284588379048e-06,
+      "loss": 3.5334,
+      "step": 1274
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.759669782286561e-06,
+      "loss": 3.6128,
+      "step": 1275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.759054208210392e-06,
+      "loss": 3.5732,
+      "step": 1276
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.758437866249871e-06,
+      "loss": 3.6027,
+      "step": 1277
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.75782075650444e-06,
+      "loss": 3.5994,
+      "step": 1278
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.757202879073675e-06,
+      "loss": 3.5872,
+      "step": 1279
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.756584234057272e-06,
+      "loss": 3.6669,
+      "step": 1280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.755964821555048e-06,
+      "loss": 3.5628,
+      "step": 1281
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.755344641666951e-06,
+      "loss": 3.5791,
+      "step": 1282
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.754723694493043e-06,
+      "loss": 3.6468,
+      "step": 1283
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.754101980133519e-06,
+      "loss": 3.6464,
+      "step": 1284
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.753479498688694e-06,
+      "loss": 3.5679,
+      "step": 1285
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.752856250259005e-06,
+      "loss": 3.6482,
+      "step": 1286
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.752232234945015e-06,
+      "loss": 3.5689,
+      "step": 1287
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.751607452847414e-06,
+      "loss": 3.6373,
+      "step": 1288
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.750981904067004e-06,
+      "loss": 3.6839,
+      "step": 1289
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.750355588704728e-06,
+      "loss": 3.6204,
+      "step": 1290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.749728506861638e-06,
+      "loss": 3.5309,
+      "step": 1291
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.749100658638914e-06,
+      "loss": 3.629,
+      "step": 1292
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.748472044137864e-06,
+      "loss": 3.5241,
+      "step": 1293
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.747842663459915e-06,
+      "loss": 3.6797,
+      "step": 1294
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.74721251670662e-06,
+      "loss": 3.6605,
+      "step": 1295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.746581603979654e-06,
+      "loss": 3.6095,
+      "step": 1296
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.745949925380815e-06,
+      "loss": 3.5882,
+      "step": 1297
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.74531748101203e-06,
+      "loss": 3.6682,
+      "step": 1298
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.744684270975338e-06,
+      "loss": 3.6736,
+      "step": 1299
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.744050295372915e-06,
+      "loss": 3.5693,
+      "step": 1300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.743415554307053e-06,
+      "loss": 3.4924,
+      "step": 1301
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.742780047880168e-06,
+      "loss": 3.6966,
+      "step": 1302
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.742143776194801e-06,
+      "loss": 3.5872,
+      "step": 1303
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.741506739353617e-06,
+      "loss": 3.5283,
+      "step": 1304
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.7408689374594e-06,
+      "loss": 3.6102,
+      "step": 1305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.740230370615066e-06,
+      "loss": 3.5548,
+      "step": 1306
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.739591038923643e-06,
+      "loss": 3.5742,
+      "step": 1307
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.738950942488295e-06,
+      "loss": 3.5681,
+      "step": 1308
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.7383100814123e-06,
+      "loss": 3.5351,
+      "step": 1309
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.737668455799058e-06,
+      "loss": 3.7092,
+      "step": 1310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.737026065752104e-06,
+      "loss": 3.4545,
+      "step": 1311
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.736382911375087e-06,
+      "loss": 3.5996,
+      "step": 1312
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.735738992771779e-06,
+      "loss": 3.5092,
+      "step": 1313
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.735094310046081e-06,
+      "loss": 3.5979,
+      "step": 1314
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.734448863302013e-06,
+      "loss": 3.6137,
+      "step": 1315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.733802652643716e-06,
+      "loss": 3.6207,
+      "step": 1316
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.733155678175463e-06,
+      "loss": 3.619,
+      "step": 1317
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.732507940001642e-06,
+      "loss": 3.6345,
+      "step": 1318
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.731859438226765e-06,
+      "loss": 3.5593,
+      "step": 1319
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.731210172955473e-06,
+      "loss": 3.6679,
+      "step": 1320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.730560144292525e-06,
+      "loss": 3.5926,
+      "step": 1321
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.729909352342803e-06,
+      "loss": 3.5757,
+      "step": 1322
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.729257797211317e-06,
+      "loss": 3.5355,
+      "step": 1323
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.728605479003195e-06,
+      "loss": 3.5483,
+      "step": 1324
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.72795239782369e-06,
+      "loss": 3.5416,
+      "step": 1325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.72729855377818e-06,
+      "loss": 3.6458,
+      "step": 1326
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.726643946972161e-06,
+      "loss": 3.601,
+      "step": 1327
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.725988577511256e-06,
+      "loss": 3.6044,
+      "step": 1328
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.725332445501213e-06,
+      "loss": 3.6331,
+      "step": 1329
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.724675551047899e-06,
+      "loss": 3.5171,
+      "step": 1330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.724017894257304e-06,
+      "loss": 3.5403,
+      "step": 1331
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.723359475235545e-06,
+      "loss": 3.7481,
+      "step": 1332
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.722700294088858e-06,
+      "loss": 3.6193,
+      "step": 1333
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.722040350923602e-06,
+      "loss": 3.5291,
+      "step": 1334
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.721379645846263e-06,
+      "loss": 3.6502,
+      "step": 1335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.720718178963446e-06,
+      "loss": 3.4632,
+      "step": 1336
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.720055950381879e-06,
+      "loss": 3.6426,
+      "step": 1337
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.719392960208418e-06,
+      "loss": 3.5875,
+      "step": 1338
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.718729208550033e-06,
+      "loss": 3.6754,
+      "step": 1339
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.718064695513826e-06,
+      "loss": 3.5413,
+      "step": 1340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.717399421207015e-06,
+      "loss": 3.6287,
+      "step": 1341
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.716733385736946e-06,
+      "loss": 3.5539,
+      "step": 1342
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.716066589211082e-06,
+      "loss": 3.6051,
+      "step": 1343
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.715399031737014e-06,
+      "loss": 3.5913,
+      "step": 1344
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.714730713422455e-06,
+      "loss": 3.6674,
+      "step": 1345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.714061634375238e-06,
+      "loss": 3.722,
+      "step": 1346
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.713391794703323e-06,
+      "loss": 3.5729,
+      "step": 1347
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.712721194514784e-06,
+      "loss": 3.6208,
+      "step": 1348
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.712049833917833e-06,
+      "loss": 3.5448,
+      "step": 1349
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.711377713020789e-06,
+      "loss": 3.6927,
+      "step": 1350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.7107048319321e-06,
+      "loss": 3.6828,
+      "step": 1351
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.710031190760343e-06,
+      "loss": 3.5956,
+      "step": 1352
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.709356789614204e-06,
+      "loss": 3.7497,
+      "step": 1353
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.708681628602504e-06,
+      "loss": 3.6681,
+      "step": 1354
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.70800570783418e-06,
+      "loss": 3.5517,
+      "step": 1355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.707329027418293e-06,
+      "loss": 3.5661,
+      "step": 1356
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.70665158746403e-06,
+      "loss": 3.6245,
+      "step": 1357
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.705973388080694e-06,
+      "loss": 3.656,
+      "step": 1358
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.705294429377713e-06,
+      "loss": 3.62,
+      "step": 1359
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.704614711464645e-06,
+      "loss": 3.5634,
+      "step": 1360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.703934234451158e-06,
+      "loss": 3.5751,
+      "step": 1361
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.70325299844705e-06,
+      "loss": 3.6216,
+      "step": 1362
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.702571003562241e-06,
+      "loss": 3.6475,
+      "step": 1363
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.701888249906773e-06,
+      "loss": 3.5378,
+      "step": 1364
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.701204737590806e-06,
+      "loss": 3.4801,
+      "step": 1365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.70052046672463e-06,
+      "loss": 3.5129,
+      "step": 1366
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.699835437418655e-06,
+      "loss": 3.6922,
+      "step": 1367
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.699149649783406e-06,
+      "loss": 3.6331,
+      "step": 1368
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.698463103929542e-06,
+      "loss": 3.5488,
+      "step": 1369
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.697775799967838e-06,
+      "loss": 3.585,
+      "step": 1370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.69708773800919e-06,
+      "loss": 3.5418,
+      "step": 1371
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.69639891816462e-06,
+      "loss": 3.5912,
+      "step": 1372
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.695709340545269e-06,
+      "loss": 3.5899,
+      "step": 1373
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.695019005262402e-06,
+      "loss": 3.7269,
+      "step": 1374
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.69432791242741e-06,
+      "loss": 3.6665,
+      "step": 1375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.693636062151798e-06,
+      "loss": 3.4974,
+      "step": 1376
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.6929434545472e-06,
+      "loss": 3.5276,
+      "step": 1377
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.692250089725369e-06,
+      "loss": 3.6294,
+      "step": 1378
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.69155596779818e-06,
+      "loss": 3.5755,
+      "step": 1379
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.690861088877633e-06,
+      "loss": 3.5312,
+      "step": 1380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.690165453075847e-06,
+      "loss": 3.5876,
+      "step": 1381
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.689469060505064e-06,
+      "loss": 3.5394,
+      "step": 1382
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.688771911277652e-06,
+      "loss": 3.5143,
+      "step": 1383
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.688074005506094e-06,
+      "loss": 3.666,
+      "step": 1384
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.687375343302999e-06,
+      "loss": 3.4818,
+      "step": 1385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.686675924781098e-06,
+      "loss": 3.5952,
+      "step": 1386
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.685975750053248e-06,
+      "loss": 3.4632,
+      "step": 1387
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.685274819232417e-06,
+      "loss": 3.6147,
+      "step": 1388
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.684573132431707e-06,
+      "loss": 3.5541,
+      "step": 1389
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.683870689764335e-06,
+      "loss": 3.6659,
+      "step": 1390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.68316749134364e-06,
+      "loss": 3.7476,
+      "step": 1391
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.682463537283086e-06,
+      "loss": 3.5598,
+      "step": 1392
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.681758827696259e-06,
+      "loss": 3.572,
+      "step": 1393
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.681053362696865e-06,
+      "loss": 3.5418,
+      "step": 1394
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.68034714239873e-06,
+      "loss": 3.6515,
+      "step": 1395
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.679640166915807e-06,
+      "loss": 3.6235,
+      "step": 1396
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.678932436362168e-06,
+      "loss": 3.6605,
+      "step": 1397
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.678223950852005e-06,
+      "loss": 3.5375,
+      "step": 1398
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.677514710499635e-06,
+      "loss": 3.5908,
+      "step": 1399
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.676804715419494e-06,
+      "loss": 3.5295,
+      "step": 1400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.676093965726146e-06,
+      "loss": 3.6247,
+      "step": 1401
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.675382461534266e-06,
+      "loss": 3.7669,
+      "step": 1402
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.674670202958661e-06,
+      "loss": 3.5613,
+      "step": 1403
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.673957190114255e-06,
+      "loss": 3.7169,
+      "step": 1404
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.673243423116091e-06,
+      "loss": 3.7191,
+      "step": 1405
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.672528902079341e-06,
+      "loss": 3.6464,
+      "step": 1406
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.671813627119294e-06,
+      "loss": 3.6151,
+      "step": 1407
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.67109759835136e-06,
+      "loss": 3.6152,
+      "step": 1408
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.670380815891071e-06,
+      "loss": 3.5276,
+      "step": 1409
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.669663279854084e-06,
+      "loss": 3.5291,
+      "step": 1410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.668944990356174e-06,
+      "loss": 3.6524,
+      "step": 1411
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.668225947513239e-06,
+      "loss": 3.5784,
+      "step": 1412
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.667506151441295e-06,
+      "loss": 3.5985,
+      "step": 1413
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.666785602256488e-06,
+      "loss": 3.6048,
+      "step": 1414
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.666064300075078e-06,
+      "loss": 3.6134,
+      "step": 1415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.665342245013447e-06,
+      "loss": 3.6272,
+      "step": 1416
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.664619437188101e-06,
+      "loss": 3.597,
+      "step": 1417
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.66389587671567e-06,
+      "loss": 3.6419,
+      "step": 1418
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.663171563712898e-06,
+      "loss": 3.5643,
+      "step": 1419
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.662446498296655e-06,
+      "loss": 3.4712,
+      "step": 1420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.661720680583937e-06,
+      "loss": 3.5608,
+      "step": 1421
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.660994110691851e-06,
+      "loss": 3.5995,
+      "step": 1422
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.660266788737631e-06,
+      "loss": 3.5874,
+      "step": 1423
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.659538714838635e-06,
+      "loss": 3.587,
+      "step": 1424
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.658809889112336e-06,
+      "loss": 3.551,
+      "step": 1425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.658080311676334e-06,
+      "loss": 3.6216,
+      "step": 1426
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.657349982648347e-06,
+      "loss": 3.6221,
+      "step": 1427
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.656618902146216e-06,
+      "loss": 3.665,
+      "step": 1428
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.655887070287903e-06,
+      "loss": 3.6058,
+      "step": 1429
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.655154487191489e-06,
+      "loss": 3.5914,
+      "step": 1430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.65442115297518e-06,
+      "loss": 3.5589,
+      "step": 1431
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.653687067757301e-06,
+      "loss": 3.5605,
+      "step": 1432
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.652952231656294e-06,
+      "loss": 3.6445,
+      "step": 1433
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.652216644790734e-06,
+      "loss": 3.5296,
+      "step": 1434
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.651480307279302e-06,
+      "loss": 3.6027,
+      "step": 1435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.650743219240814e-06,
+      "loss": 3.6266,
+      "step": 1436
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.650005380794196e-06,
+      "loss": 3.7359,
+      "step": 1437
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.649266792058504e-06,
+      "loss": 3.5876,
+      "step": 1438
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.648527453152908e-06,
+      "loss": 3.457,
+      "step": 1439
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.647787364196704e-06,
+      "loss": 3.5659,
+      "step": 1440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.647046525309305e-06,
+      "loss": 3.5941,
+      "step": 1441
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.64630493661025e-06,
+      "loss": 3.5959,
+      "step": 1442
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.645562598219193e-06,
+      "loss": 3.625,
+      "step": 1443
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.644819510255913e-06,
+      "loss": 3.6927,
+      "step": 1444
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.644075672840311e-06,
+      "loss": 3.5585,
+      "step": 1445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.643331086092404e-06,
+      "loss": 3.7151,
+      "step": 1446
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.642585750132334e-06,
+      "loss": 3.502,
+      "step": 1447
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.641839665080363e-06,
+      "loss": 3.634,
+      "step": 1448
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.641092831056874e-06,
+      "loss": 3.5936,
+      "step": 1449
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.640345248182368e-06,
+      "loss": 3.6961,
+      "step": 1450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.639596916577472e-06,
+      "loss": 3.6124,
+      "step": 1451
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.63884783636293e-06,
+      "loss": 3.6336,
+      "step": 1452
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.638098007659607e-06,
+      "loss": 3.743,
+      "step": 1453
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.63734743058849e-06,
+      "loss": 3.5867,
+      "step": 1454
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.636596105270685e-06,
+      "loss": 3.6204,
+      "step": 1455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.635844031827424e-06,
+      "loss": 3.5207,
+      "step": 1456
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.635091210380052e-06,
+      "loss": 3.491,
+      "step": 1457
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.63433764105004e-06,
+      "loss": 3.5781,
+      "step": 1458
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.63358332395898e-06,
+      "loss": 3.5794,
+      "step": 1459
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.632828259228577e-06,
+      "loss": 3.6138,
+      "step": 1460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.63207244698067e-06,
+      "loss": 3.499,
+      "step": 1461
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.631315887337206e-06,
+      "loss": 3.588,
+      "step": 1462
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.630558580420258e-06,
+      "loss": 3.7301,
+      "step": 1463
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.629800526352022e-06,
+      "loss": 3.6337,
+      "step": 1464
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.629041725254808e-06,
+      "loss": 3.5689,
+      "step": 1465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.628282177251054e-06,
+      "loss": 3.6708,
+      "step": 1466
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.62752188246331e-06,
+      "loss": 3.6258,
+      "step": 1467
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.626760841014258e-06,
+      "loss": 3.4478,
+      "step": 1468
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.62599905302669e-06,
+      "loss": 3.6213,
+      "step": 1469
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.625236518623519e-06,
+      "loss": 3.5625,
+      "step": 1470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.624473237927788e-06,
+      "loss": 3.6082,
+      "step": 1471
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.62370921106265e-06,
+      "loss": 3.5581,
+      "step": 1472
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.622944438151385e-06,
+      "loss": 3.7039,
+      "step": 1473
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.622178919317388e-06,
+      "loss": 3.6975,
+      "step": 1474
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.62141265468418e-06,
+      "loss": 3.6658,
+      "step": 1475
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.620645644375397e-06,
+      "loss": 3.5821,
+      "step": 1476
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.6198778885148e-06,
+      "loss": 3.597,
+      "step": 1477
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.619109387226266e-06,
+      "loss": 3.5714,
+      "step": 1478
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.618340140633797e-06,
+      "loss": 3.6059,
+      "step": 1479
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.617570148861513e-06,
+      "loss": 3.6572,
+      "step": 1480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.616799412033651e-06,
+      "loss": 3.6565,
+      "step": 1481
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.616027930274573e-06,
+      "loss": 3.6926,
+      "step": 1482
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.615255703708758e-06,
+      "loss": 3.6176,
+      "step": 1483
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.61448273246081e-06,
+      "loss": 3.612,
+      "step": 1484
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.613709016655445e-06,
+      "loss": 3.6246,
+      "step": 1485
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.612934556417505e-06,
+      "loss": 3.5353,
+      "step": 1486
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.612159351871954e-06,
+      "loss": 3.6682,
+      "step": 1487
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.611383403143872e-06,
+      "loss": 3.6992,
+      "step": 1488
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.610606710358457e-06,
+      "loss": 3.6379,
+      "step": 1489
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.609829273641034e-06,
+      "loss": 3.6327,
+      "step": 1490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.609051093117043e-06,
+      "loss": 3.5174,
+      "step": 1491
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.608272168912043e-06,
+      "loss": 3.4852,
+      "step": 1492
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.607492501151718e-06,
+      "loss": 3.5413,
+      "step": 1493
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.606712089961867e-06,
+      "loss": 3.6486,
+      "step": 1494
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.605930935468412e-06,
+      "loss": 3.6058,
+      "step": 1495
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.605149037797396e-06,
+      "loss": 3.5597,
+      "step": 1496
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.604366397074978e-06,
+      "loss": 3.7202,
+      "step": 1497
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.603583013427438e-06,
+      "loss": 3.517,
+      "step": 1498
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.60279888698118e-06,
+      "loss": 3.5893,
+      "step": 1499
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.602014017862723e-06,
+      "loss": 3.6086,
+      "step": 1500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.601228406198704e-06,
+      "loss": 3.6948,
+      "step": 1501
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.600442052115888e-06,
+      "loss": 3.4992,
+      "step": 1502
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.599654955741154e-06,
+      "loss": 3.5215,
+      "step": 1503
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.5988671172015e-06,
+      "loss": 3.6816,
+      "step": 1504
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.598078536624046e-06,
+      "loss": 3.7155,
+      "step": 1505
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.597289214136035e-06,
+      "loss": 3.6345,
+      "step": 1506
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.596499149864822e-06,
+      "loss": 3.6101,
+      "step": 1507
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.595708343937885e-06,
+      "loss": 3.5634,
+      "step": 1508
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.594916796482825e-06,
+      "loss": 3.6724,
+      "step": 1509
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.59412450762736e-06,
+      "loss": 3.6383,
+      "step": 1510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.593331477499325e-06,
+      "loss": 3.6222,
+      "step": 1511
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.59253770622668e-06,
+      "loss": 3.639,
+      "step": 1512
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.5917431939375e-06,
+      "loss": 3.5942,
+      "step": 1513
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.590947940759982e-06,
+      "loss": 3.491,
+      "step": 1514
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.590151946822444e-06,
+      "loss": 3.5182,
+      "step": 1515
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.589355212253318e-06,
+      "loss": 3.5175,
+      "step": 1516
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.58855773718116e-06,
+      "loss": 3.5772,
+      "step": 1517
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.587759521734646e-06,
+      "loss": 3.6096,
+      "step": 1518
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.58696056604257e-06,
+      "loss": 3.5148,
+      "step": 1519
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.58616087023384e-06,
+      "loss": 3.75,
+      "step": 1520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.585360434437497e-06,
+      "loss": 3.6883,
+      "step": 1521
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.584559258782688e-06,
+      "loss": 3.6379,
+      "step": 1522
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.583757343398685e-06,
+      "loss": 3.6167,
+      "step": 1523
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.582954688414881e-06,
+      "loss": 3.576,
+      "step": 1524
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.582151293960782e-06,
+      "loss": 3.5806,
+      "step": 1525
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.581347160166024e-06,
+      "loss": 3.5407,
+      "step": 1526
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.580542287160348e-06,
+      "loss": 3.4797,
+      "step": 1527
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.57973667507363e-06,
+      "loss": 3.5523,
+      "step": 1528
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.578930324035852e-06,
+      "loss": 3.5242,
+      "step": 1529
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.57812323417712e-06,
+      "loss": 3.5245,
+      "step": 1530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.577315405627665e-06,
+      "loss": 3.661,
+      "step": 1531
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.576506838517829e-06,
+      "loss": 3.4303,
+      "step": 1532
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.575697532978074e-06,
+      "loss": 3.7282,
+      "step": 1533
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.574887489138987e-06,
+      "loss": 3.5525,
+      "step": 1534
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.57407670713127e-06,
+      "loss": 3.5939,
+      "step": 1535
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.573265187085742e-06,
+      "loss": 3.5593,
+      "step": 1536
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.572452929133344e-06,
+      "loss": 3.5469,
+      "step": 1537
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.571639933405139e-06,
+      "loss": 3.6245,
+      "step": 1538
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.570826200032303e-06,
+      "loss": 3.5341,
+      "step": 1539
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.570011729146134e-06,
+      "loss": 3.575,
+      "step": 1540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.569196520878048e-06,
+      "loss": 3.7337,
+      "step": 1541
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.568380575359584e-06,
+      "loss": 3.6549,
+      "step": 1542
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.567563892722392e-06,
+      "loss": 3.5899,
+      "step": 1543
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.566746473098249e-06,
+      "loss": 3.5614,
+      "step": 1544
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.565928316619048e-06,
+      "loss": 3.6187,
+      "step": 1545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.565109423416799e-06,
+      "loss": 3.5066,
+      "step": 1546
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.56428979362363e-06,
+      "loss": 3.5147,
+      "step": 1547
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.563469427371794e-06,
+      "loss": 3.5859,
+      "step": 1548
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.562648324793657e-06,
+      "loss": 3.6135,
+      "step": 1549
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.561826486021706e-06,
+      "loss": 3.5007,
+      "step": 1550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.561003911188548e-06,
+      "loss": 3.6604,
+      "step": 1551
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.560180600426905e-06,
+      "loss": 3.665,
+      "step": 1552
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.559356553869623e-06,
+      "loss": 3.536,
+      "step": 1553
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.55853177164966e-06,
+      "loss": 3.5195,
+      "step": 1554
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.5577062539001e-06,
+      "loss": 3.6014,
+      "step": 1555
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.55688000075414e-06,
+      "loss": 3.6212,
+      "step": 1556
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.556053012345098e-06,
+      "loss": 3.6646,
+      "step": 1557
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.555225288806413e-06,
+      "loss": 3.6834,
+      "step": 1558
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.554396830271638e-06,
+      "loss": 3.5594,
+      "step": 1559
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.553567636874445e-06,
+      "loss": 3.7195,
+      "step": 1560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.552737708748632e-06,
+      "loss": 3.6383,
+      "step": 1561
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.551907046028102e-06,
+      "loss": 3.734,
+      "step": 1562
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.55107564884689e-06,
+      "loss": 3.5569,
+      "step": 1563
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.550243517339142e-06,
+      "loss": 3.5899,
+      "step": 1564
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.549410651639126e-06,
+      "loss": 3.624,
+      "step": 1565
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.548577051881223e-06,
+      "loss": 3.6068,
+      "step": 1566
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.547742718199939e-06,
+      "loss": 3.5381,
+      "step": 1567
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.546907650729896e-06,
+      "loss": 3.6393,
+      "step": 1568
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.546071849605834e-06,
+      "loss": 3.5736,
+      "step": 1569
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.545235314962609e-06,
+      "loss": 3.6065,
+      "step": 1570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.5443980469352e-06,
+      "loss": 3.5964,
+      "step": 1571
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.543560045658701e-06,
+      "loss": 3.6255,
+      "step": 1572
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.542721311268328e-06,
+      "loss": 3.5666,
+      "step": 1573
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.54188184389941e-06,
+      "loss": 3.6207,
+      "step": 1574
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.541041643687397e-06,
+      "loss": 3.6519,
+      "step": 1575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.540200710767859e-06,
+      "loss": 3.6317,
+      "step": 1576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.539359045276481e-06,
+      "loss": 3.6879,
+      "step": 1577
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.538516647349068e-06,
+      "loss": 3.5662,
+      "step": 1578
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.537673517121544e-06,
+      "loss": 3.655,
+      "step": 1579
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.536829654729949e-06,
+      "loss": 3.6676,
+      "step": 1580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.53598506031044e-06,
+      "loss": 3.684,
+      "step": 1581
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.535139733999299e-06,
+      "loss": 3.6872,
+      "step": 1582
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.534293675932916e-06,
+      "loss": 3.6167,
+      "step": 1583
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.53344688624781e-06,
+      "loss": 3.5937,
+      "step": 1584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.532599365080609e-06,
+      "loss": 3.5705,
+      "step": 1585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.531751112568061e-06,
+      "loss": 3.5129,
+      "step": 1586
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.530902128847038e-06,
+      "loss": 3.5829,
+      "step": 1587
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.530052414054522e-06,
+      "loss": 3.5983,
+      "step": 1588
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.529201968327618e-06,
+      "loss": 3.4988,
+      "step": 1589
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.528350791803546e-06,
+      "loss": 3.6219,
+      "step": 1590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.527498884619646e-06,
+      "loss": 3.5984,
+      "step": 1591
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.526646246913377e-06,
+      "loss": 3.5992,
+      "step": 1592
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.525792878822312e-06,
+      "loss": 3.6542,
+      "step": 1593
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.524938780484143e-06,
+      "loss": 3.6522,
+      "step": 1594
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.524083952036684e-06,
+      "loss": 3.6344,
+      "step": 1595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.523228393617861e-06,
+      "loss": 3.5888,
+      "step": 1596
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.52237210536572e-06,
+      "loss": 3.4658,
+      "step": 1597
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.521515087418426e-06,
+      "loss": 3.5697,
+      "step": 1598
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.520657339914262e-06,
+      "loss": 3.5583,
+      "step": 1599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.519798862991627e-06,
+      "loss": 3.5513,
+      "step": 1600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.518939656789037e-06,
+      "loss": 3.6221,
+      "step": 1601
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.518079721445126e-06,
+      "loss": 3.7185,
+      "step": 1602
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.51721905709865e-06,
+      "loss": 3.5286,
+      "step": 1603
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.516357663888476e-06,
+      "loss": 3.6041,
+      "step": 1604
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.515495541953593e-06,
+      "loss": 3.679,
+      "step": 1605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.514632691433108e-06,
+      "loss": 3.6784,
+      "step": 1606
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.51376911246624e-06,
+      "loss": 3.609,
+      "step": 1607
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.512904805192333e-06,
+      "loss": 3.7097,
+      "step": 1608
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.512039769750843e-06,
+      "loss": 3.5185,
+      "step": 1609
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.511174006281348e-06,
+      "loss": 3.5538,
+      "step": 1610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.510307514923536e-06,
+      "loss": 3.6803,
+      "step": 1611
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.509440295817223e-06,
+      "loss": 3.5258,
+      "step": 1612
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.508572349102335e-06,
+      "loss": 3.5669,
+      "step": 1613
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.507703674918914e-06,
+      "loss": 3.6365,
+      "step": 1614
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.506834273407126e-06,
+      "loss": 3.6376,
+      "step": 1615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.505964144707251e-06,
+      "loss": 3.6069,
+      "step": 1616
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.505093288959683e-06,
+      "loss": 3.5258,
+      "step": 1617
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.50422170630494e-06,
+      "loss": 3.689,
+      "step": 1618
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.503349396883654e-06,
+      "loss": 3.6379,
+      "step": 1619
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.502476360836573e-06,
+      "loss": 3.6651,
+      "step": 1620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.50160259830456e-06,
+      "loss": 3.5802,
+      "step": 1621
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.500728109428603e-06,
+      "loss": 3.5085,
+      "step": 1622
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.499852894349804e-06,
+      "loss": 3.6163,
+      "step": 1623
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.498976953209377e-06,
+      "loss": 3.484,
+      "step": 1624
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.49810028614866e-06,
+      "loss": 3.6133,
+      "step": 1625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.497222893309101e-06,
+      "loss": 3.6988,
+      "step": 1626
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.496344774832275e-06,
+      "loss": 3.6187,
+      "step": 1627
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.495465930859864e-06,
+      "loss": 3.5663,
+      "step": 1628
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.494586361533674e-06,
+      "loss": 3.5779,
+      "step": 1629
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.493706066995625e-06,
+      "loss": 3.6707,
+      "step": 1630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.492825047387753e-06,
+      "loss": 3.6756,
+      "step": 1631
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.491943302852216e-06,
+      "loss": 3.6095,
+      "step": 1632
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.491060833531283e-06,
+      "loss": 3.5726,
+      "step": 1633
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.490177639567341e-06,
+      "loss": 3.6683,
+      "step": 1634
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.489293721102899e-06,
+      "loss": 3.4427,
+      "step": 1635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.488409078280576e-06,
+      "loss": 3.584,
+      "step": 1636
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.487523711243113e-06,
+      "loss": 3.5412,
+      "step": 1637
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.486637620133367e-06,
+      "loss": 3.6271,
+      "step": 1638
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.48575080509431e-06,
+      "loss": 3.5218,
+      "step": 1639
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.484863266269029e-06,
+      "loss": 3.6148,
+      "step": 1640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.483975003800734e-06,
+      "loss": 3.567,
+      "step": 1641
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.483086017832746e-06,
+      "loss": 3.5649,
+      "step": 1642
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.482196308508506e-06,
+      "loss": 3.6591,
+      "step": 1643
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.481305875971571e-06,
+      "loss": 3.5816,
+      "step": 1644
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.480414720365612e-06,
+      "loss": 3.5432,
+      "step": 1645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.479522841834423e-06,
+      "loss": 3.5305,
+      "step": 1646
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.478630240521907e-06,
+      "loss": 3.5528,
+      "step": 1647
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.477736916572088e-06,
+      "loss": 3.6611,
+      "step": 1648
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.476842870129108e-06,
+      "loss": 3.6378,
+      "step": 1649
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.47594810133722e-06,
+      "loss": 3.5254,
+      "step": 1650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.475052610340802e-06,
+      "loss": 3.5857,
+      "step": 1651
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.474156397284337e-06,
+      "loss": 3.6029,
+      "step": 1652
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.473259462312438e-06,
+      "loss": 3.6417,
+      "step": 1653
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.472361805569823e-06,
+      "loss": 3.476,
+      "step": 1654
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.47146342720133e-06,
+      "loss": 3.5334,
+      "step": 1655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.470564327351919e-06,
+      "loss": 3.7366,
+      "step": 1656
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.469664506166659e-06,
+      "loss": 3.5939,
+      "step": 1657
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.468763963790737e-06,
+      "loss": 3.6105,
+      "step": 1658
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.46786270036946e-06,
+      "loss": 3.5048,
+      "step": 1659
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.466960716048248e-06,
+      "loss": 3.5863,
+      "step": 1660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.466058010972638e-06,
+      "loss": 3.7056,
+      "step": 1661
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.465154585288284e-06,
+      "loss": 3.6584,
+      "step": 1662
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.464250439140957e-06,
+      "loss": 3.4412,
+      "step": 1663
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.46334557267654e-06,
+      "loss": 3.5872,
+      "step": 1664
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.462439986041036e-06,
+      "loss": 3.5528,
+      "step": 1665
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.461533679380568e-06,
+      "loss": 3.6985,
+      "step": 1666
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.460626652841366e-06,
+      "loss": 3.5201,
+      "step": 1667
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.45971890656978e-06,
+      "loss": 3.5613,
+      "step": 1668
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.45881044071228e-06,
+      "loss": 3.7039,
+      "step": 1669
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.457901255415447e-06,
+      "loss": 3.6012,
+      "step": 1670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.456991350825983e-06,
+      "loss": 3.5686,
+      "step": 1671
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.456080727090701e-06,
+      "loss": 3.7264,
+      "step": 1672
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.455169384356532e-06,
+      "loss": 3.5514,
+      "step": 1673
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.454257322770523e-06,
+      "loss": 3.6217,
+      "step": 1674
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.453344542479839e-06,
+      "loss": 3.6017,
+      "step": 1675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.452431043631757e-06,
+      "loss": 3.4295,
+      "step": 1676
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.451516826373676e-06,
+      "loss": 3.7162,
+      "step": 1677
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.450601890853104e-06,
+      "loss": 3.5068,
+      "step": 1678
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.449686237217667e-06,
+      "loss": 3.5677,
+      "step": 1679
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.448769865615109e-06,
+      "loss": 3.5345,
+      "step": 1680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.44785277619329e-06,
+      "loss": 3.6221,
+      "step": 1681
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.446934969100184e-06,
+      "loss": 3.5892,
+      "step": 1682
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.44601644448388e-06,
+      "loss": 3.6002,
+      "step": 1683
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.445097202492583e-06,
+      "loss": 3.6673,
+      "step": 1684
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.444177243274619e-06,
+      "loss": 3.526,
+      "step": 1685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.44325656697842e-06,
+      "loss": 3.6335,
+      "step": 1686
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.442335173752546e-06,
+      "loss": 3.6366,
+      "step": 1687
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.44141306374566e-06,
+      "loss": 3.5917,
+      "step": 1688
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.44049023710655e-06,
+      "loss": 3.5067,
+      "step": 1689
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.439566693984115e-06,
+      "loss": 3.5692,
+      "step": 1690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.43864243452737e-06,
+      "loss": 3.6091,
+      "step": 1691
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.437717458885447e-06,
+      "loss": 3.5746,
+      "step": 1692
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.436791767207595e-06,
+      "loss": 3.5435,
+      "step": 1693
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.435865359643172e-06,
+      "loss": 3.6222,
+      "step": 1694
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.43493823634166e-06,
+      "loss": 3.5495,
+      "step": 1695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.43401039745265e-06,
+      "loss": 3.6268,
+      "step": 1696
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.433081843125855e-06,
+      "loss": 3.6167,
+      "step": 1697
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.432152573511094e-06,
+      "loss": 3.4809,
+      "step": 1698
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.431222588758311e-06,
+      "loss": 3.6183,
+      "step": 1699
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.430291889017557e-06,
+      "loss": 3.5813,
+      "step": 1700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.429360474439007e-06,
+      "loss": 3.5784,
+      "step": 1701
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.428428345172944e-06,
+      "loss": 3.6127,
+      "step": 1702
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.42749550136977e-06,
+      "loss": 3.6373,
+      "step": 1703
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.426561943180001e-06,
+      "loss": 3.5803,
+      "step": 1704
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.425627670754272e-06,
+      "loss": 3.5012,
+      "step": 1705
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.424692684243324e-06,
+      "loss": 3.6305,
+      "step": 1706
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.423756983798025e-06,
+      "loss": 3.5955,
+      "step": 1707
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.422820569569349e-06,
+      "loss": 3.569,
+      "step": 1708
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.421883441708389e-06,
+      "loss": 3.5496,
+      "step": 1709
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.420945600366353e-06,
+      "loss": 3.6607,
+      "step": 1710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.420007045694567e-06,
+      "loss": 3.5721,
+      "step": 1711
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.419067777844464e-06,
+      "loss": 3.6103,
+      "step": 1712
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.418127796967599e-06,
+      "loss": 3.4767,
+      "step": 1713
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.41718710321564e-06,
+      "loss": 3.4654,
+      "step": 1714
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.41624569674037e-06,
+      "loss": 3.6577,
+      "step": 1715
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.41530357769369e-06,
+      "loss": 3.5397,
+      "step": 1716
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.414360746227608e-06,
+      "loss": 3.624,
+      "step": 1717
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.413417202494256e-06,
+      "loss": 3.5591,
+      "step": 1718
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.412472946645875e-06,
+      "loss": 3.614,
+      "step": 1719
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.411527978834823e-06,
+      "loss": 3.5791,
+      "step": 1720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.410582299213574e-06,
+      "loss": 3.591,
+      "step": 1721
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.409635907934714e-06,
+      "loss": 3.6466,
+      "step": 1722
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.408688805150947e-06,
+      "loss": 3.3976,
+      "step": 1723
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.407740991015088e-06,
+      "loss": 3.65,
+      "step": 1724
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.40679246568007e-06,
+      "loss": 3.5904,
+      "step": 1725
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.40584322929894e-06,
+      "loss": 3.5122,
+      "step": 1726
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.40489328202486e-06,
+      "loss": 3.6082,
+      "step": 1727
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.403942624011107e-06,
+      "loss": 3.6581,
+      "step": 1728
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.402991255411069e-06,
+      "loss": 3.4988,
+      "step": 1729
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.402039176378254e-06,
+      "loss": 3.6236,
+      "step": 1730
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.40108638706628e-06,
+      "loss": 3.6086,
+      "step": 1731
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.400132887628886e-06,
+      "loss": 3.5758,
+      "step": 1732
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.399178678219916e-06,
+      "loss": 3.6165,
+      "step": 1733
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.398223758993336e-06,
+      "loss": 3.6021,
+      "step": 1734
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.397268130103224e-06,
+      "loss": 3.5625,
+      "step": 1735
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.396311791703775e-06,
+      "loss": 3.6084,
+      "step": 1736
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.395354743949294e-06,
+      "loss": 3.6131,
+      "step": 1737
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.394396986994205e-06,
+      "loss": 3.6075,
+      "step": 1738
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.393438520993042e-06,
+      "loss": 3.6954,
+      "step": 1739
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.392479346100455e-06,
+      "loss": 3.5727,
+      "step": 1740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.391519462471214e-06,
+      "loss": 3.5676,
+      "step": 1741
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.390558870260193e-06,
+      "loss": 3.6572,
+      "step": 1742
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.38959756962239e-06,
+      "loss": 3.6631,
+      "step": 1743
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.388635560712912e-06,
+      "loss": 3.6124,
+      "step": 1744
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.387672843686977e-06,
+      "loss": 3.5748,
+      "step": 1745
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.386709418699927e-06,
+      "loss": 3.5915,
+      "step": 1746
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.38574528590721e-06,
+      "loss": 3.674,
+      "step": 1747
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.384780445464394e-06,
+      "loss": 3.5615,
+      "step": 1748
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.383814897527154e-06,
+      "loss": 3.498,
+      "step": 1749
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.382848642251286e-06,
+      "loss": 3.5864,
+      "step": 1750
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.381881679792698e-06,
+      "loss": 3.6572,
+      "step": 1751
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.38091401030741e-06,
+      "loss": 3.6019,
+      "step": 1752
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.37994563395156e-06,
+      "loss": 3.666,
+      "step": 1753
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.378976550881393e-06,
+      "loss": 3.5295,
+      "step": 1754
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.378006761253277e-06,
+      "loss": 3.5454,
+      "step": 1755
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.377036265223692e-06,
+      "loss": 3.6831,
+      "step": 1756
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.376065062949223e-06,
+      "loss": 3.6937,
+      "step": 1757
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.375093154586581e-06,
+      "loss": 3.5195,
+      "step": 1758
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.374120540292581e-06,
+      "loss": 3.4879,
+      "step": 1759
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.373147220224162e-06,
+      "loss": 3.4515,
+      "step": 1760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.372173194538368e-06,
+      "loss": 3.6568,
+      "step": 1761
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.371198463392363e-06,
+      "loss": 3.6261,
+      "step": 1762
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.370223026943418e-06,
+      "loss": 3.6312,
+      "step": 1763
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.369246885348926e-06,
+      "loss": 3.5417,
+      "step": 1764
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.368270038766387e-06,
+      "loss": 3.6661,
+      "step": 1765
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.367292487353417e-06,
+      "loss": 3.5091,
+      "step": 1766
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.36631423126775e-06,
+      "loss": 3.5025,
+      "step": 1767
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.365335270667226e-06,
+      "loss": 3.525,
+      "step": 1768
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.364355605709805e-06,
+      "loss": 3.5817,
+      "step": 1769
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.363375236553555e-06,
+      "loss": 3.663,
+      "step": 1770
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.362394163356664e-06,
+      "loss": 3.5086,
+      "step": 1771
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.361412386277428e-06,
+      "loss": 3.6342,
+      "step": 1772
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.360429905474264e-06,
+      "loss": 3.7396,
+      "step": 1773
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.359446721105692e-06,
+      "loss": 3.5353,
+      "step": 1774
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.358462833330353e-06,
+      "loss": 3.5606,
+      "step": 1775
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.357478242306998e-06,
+      "loss": 3.6039,
+      "step": 1776
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.356492948194496e-06,
+      "loss": 3.6475,
+      "step": 1777
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.355506951151825e-06,
+      "loss": 3.5373,
+      "step": 1778
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.354520251338077e-06,
+      "loss": 3.6445,
+      "step": 1779
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.35353284891246e-06,
+      "loss": 3.6264,
+      "step": 1780
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.352544744034292e-06,
+      "loss": 3.5788,
+      "step": 1781
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.351555936863008e-06,
+      "loss": 3.6307,
+      "step": 1782
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.350566427558153e-06,
+      "loss": 3.6252,
+      "step": 1783
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.349576216279386e-06,
+      "loss": 3.6756,
+      "step": 1784
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.348585303186481e-06,
+      "loss": 3.5279,
+      "step": 1785
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.347593688439324e-06,
+      "loss": 3.5207,
+      "step": 1786
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.346601372197914e-06,
+      "loss": 3.5129,
+      "step": 1787
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.345608354622364e-06,
+      "loss": 3.625,
+      "step": 1788
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.344614635872897e-06,
+      "loss": 3.6626,
+      "step": 1789
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.343620216109856e-06,
+      "loss": 3.5788,
+      "step": 1790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.34262509549369e-06,
+      "loss": 3.5858,
+      "step": 1791
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.341629274184966e-06,
+      "loss": 3.4497,
+      "step": 1792
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.340632752344361e-06,
+      "loss": 3.5848,
+      "step": 1793
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.339635530132667e-06,
+      "loss": 3.6274,
+      "step": 1794
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.338637607710785e-06,
+      "loss": 3.6087,
+      "step": 1795
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.337638985239736e-06,
+      "loss": 3.5853,
+      "step": 1796
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.336639662880649e-06,
+      "loss": 3.4805,
+      "step": 1797
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.335639640794764e-06,
+      "loss": 3.7036,
+      "step": 1798
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.33463891914344e-06,
+      "loss": 3.5191,
+      "step": 1799
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.333637498088146e-06,
+      "loss": 3.6177,
+      "step": 1800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.332635377790462e-06,
+      "loss": 3.5458,
+      "step": 1801
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.331632558412083e-06,
+      "loss": 3.5561,
+      "step": 1802
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.330629040114816e-06,
+      "loss": 3.5449,
+      "step": 1803
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.32962482306058e-06,
+      "loss": 3.7045,
+      "step": 1804
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.328619907411408e-06,
+      "loss": 3.5574,
+      "step": 1805
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.32761429332945e-06,
+      "loss": 3.5615,
+      "step": 1806
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.326607980976958e-06,
+      "loss": 3.6046,
+      "step": 1807
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.325600970516306e-06,
+      "loss": 3.5971,
+      "step": 1808
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.324593262109978e-06,
+      "loss": 3.7391,
+      "step": 1809
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.323584855920567e-06,
+      "loss": 3.5982,
+      "step": 1810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.322575752110784e-06,
+      "loss": 3.5733,
+      "step": 1811
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.32156595084345e-06,
+      "loss": 3.6391,
+      "step": 1812
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.320555452281498e-06,
+      "loss": 3.6065,
+      "step": 1813
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.319544256587974e-06,
+      "loss": 3.5133,
+      "step": 1814
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.318532363926039e-06,
+      "loss": 3.5544,
+      "step": 1815
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.317519774458962e-06,
+      "loss": 3.606,
+      "step": 1816
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.316506488350128e-06,
+      "loss": 3.5955,
+      "step": 1817
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.315492505763031e-06,
+      "loss": 3.5909,
+      "step": 1818
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.314477826861282e-06,
+      "loss": 3.5714,
+      "step": 1819
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.3134624518086e-06,
+      "loss": 3.4856,
+      "step": 1820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.31244638076882e-06,
+      "loss": 3.5053,
+      "step": 1821
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.311429613905885e-06,
+      "loss": 3.5896,
+      "step": 1822
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.310412151383854e-06,
+      "loss": 3.4684,
+      "step": 1823
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.309393993366895e-06,
+      "loss": 3.7142,
+      "step": 1824
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.308375140019293e-06,
+      "loss": 3.5744,
+      "step": 1825
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.307355591505442e-06,
+      "loss": 3.5394,
+      "step": 1826
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.306335347989848e-06,
+      "loss": 3.6525,
+      "step": 1827
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.305314409637128e-06,
+      "loss": 3.6046,
+      "step": 1828
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.304292776612014e-06,
+      "loss": 3.5424,
+      "step": 1829
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.30327044907935e-06,
+      "loss": 3.567,
+      "step": 1830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.302247427204088e-06,
+      "loss": 3.5669,
+      "step": 1831
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.3012237111513e-06,
+      "loss": 3.5601,
+      "step": 1832
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.300199301086159e-06,
+      "loss": 3.6328,
+      "step": 1833
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.299174197173959e-06,
+      "loss": 3.564,
+      "step": 1834
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.298148399580103e-06,
+      "loss": 3.605,
+      "step": 1835
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.297121908470104e-06,
+      "loss": 3.4784,
+      "step": 1836
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.296094724009592e-06,
+      "loss": 3.564,
+      "step": 1837
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.295066846364302e-06,
+      "loss": 3.6637,
+      "step": 1838
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.294038275700087e-06,
+      "loss": 3.5713,
+      "step": 1839
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.293009012182907e-06,
+      "loss": 3.5882,
+      "step": 1840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.291979055978838e-06,
+      "loss": 3.6249,
+      "step": 1841
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.290948407254065e-06,
+      "loss": 3.6735,
+      "step": 1842
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.289917066174887e-06,
+      "loss": 3.58,
+      "step": 1843
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.28888503290771e-06,
+      "loss": 3.5447,
+      "step": 1844
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.287852307619056e-06,
+      "loss": 3.556,
+      "step": 1845
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.286818890475561e-06,
+      "loss": 3.573,
+      "step": 1846
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.285784781643965e-06,
+      "loss": 3.6216,
+      "step": 1847
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.284749981291127e-06,
+      "loss": 3.6579,
+      "step": 1848
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.28371448958401e-06,
+      "loss": 3.6143,
+      "step": 1849
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.2826783066897e-06,
+      "loss": 3.617,
+      "step": 1850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.28164143277538e-06,
+      "loss": 3.4128,
+      "step": 1851
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.280603868008358e-06,
+      "loss": 3.5375,
+      "step": 1852
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.279565612556043e-06,
+      "loss": 3.6613,
+      "step": 1853
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.278526666585961e-06,
+      "loss": 3.6867,
+      "step": 1854
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.277487030265751e-06,
+      "loss": 3.7181,
+      "step": 1855
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.27644670376316e-06,
+      "loss": 3.5627,
+      "step": 1856
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.275405687246043e-06,
+      "loss": 3.4984,
+      "step": 1857
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.274363980882375e-06,
+      "loss": 3.6026,
+      "step": 1858
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.273321584840235e-06,
+      "loss": 3.647,
+      "step": 1859
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.272278499287817e-06,
+      "loss": 3.5827,
+      "step": 1860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.271234724393425e-06,
+      "loss": 3.6031,
+      "step": 1861
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.270190260325475e-06,
+      "loss": 3.4466,
+      "step": 1862
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.269145107252495e-06,
+      "loss": 3.6462,
+      "step": 1863
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.26809926534312e-06,
+      "loss": 3.6019,
+      "step": 1864
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.2670527347661e-06,
+      "loss": 3.7169,
+      "step": 1865
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.266005515690297e-06,
+      "loss": 3.6116,
+      "step": 1866
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.264957608284679e-06,
+      "loss": 3.5458,
+      "step": 1867
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.263909012718331e-06,
+      "loss": 3.5983,
+      "step": 1868
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.262859729160442e-06,
+      "loss": 3.4931,
+      "step": 1869
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.261809757780323e-06,
+      "loss": 3.5733,
+      "step": 1870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.260759098747385e-06,
+      "loss": 3.4844,
+      "step": 1871
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.259707752231156e-06,
+      "loss": 3.4786,
+      "step": 1872
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.25865571840127e-06,
+      "loss": 3.6589,
+      "step": 1873
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.257602997427479e-06,
+      "loss": 3.5876,
+      "step": 1874
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.25654958947964e-06,
+      "loss": 3.5808,
+      "step": 1875
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.255495494727722e-06,
+      "loss": 3.4774,
+      "step": 1876
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.254440713341809e-06,
+      "loss": 3.5101,
+      "step": 1877
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.253385245492089e-06,
+      "loss": 3.591,
+      "step": 1878
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.252329091348863e-06,
+      "loss": 3.6224,
+      "step": 1879
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.251272251082549e-06,
+      "loss": 3.6541,
+      "step": 1880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.250214724863666e-06,
+      "loss": 3.5523,
+      "step": 1881
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.249156512862851e-06,
+      "loss": 3.577,
+      "step": 1882
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.248097615250849e-06,
+      "loss": 3.5574,
+      "step": 1883
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.247038032198513e-06,
+      "loss": 3.6302,
+      "step": 1884
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.245977763876812e-06,
+      "loss": 3.6002,
+      "step": 1885
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.244916810456822e-06,
+      "loss": 3.5088,
+      "step": 1886
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.243855172109727e-06,
+      "loss": 3.6197,
+      "step": 1887
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.24279284900683e-06,
+      "loss": 3.6503,
+      "step": 1888
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.241729841319535e-06,
+      "loss": 3.6201,
+      "step": 1889
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.240666149219363e-06,
+      "loss": 3.5346,
+      "step": 1890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.239601772877943e-06,
+      "loss": 3.5819,
+      "step": 1891
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.238536712467015e-06,
+      "loss": 3.6951,
+      "step": 1892
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.237470968158429e-06,
+      "loss": 3.6024,
+      "step": 1893
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.236404540124143e-06,
+      "loss": 3.4689,
+      "step": 1894
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.23533742853623e-06,
+      "loss": 3.487,
+      "step": 1895
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.234269633566871e-06,
+      "loss": 3.6622,
+      "step": 1896
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.233201155388354e-06,
+      "loss": 3.5505,
+      "step": 1897
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.232131994173084e-06,
+      "loss": 3.4764,
+      "step": 1898
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.23106215009357e-06,
+      "loss": 3.5185,
+      "step": 1899
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.229991623322437e-06,
+      "loss": 3.5435,
+      "step": 1900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.228920414032414e-06,
+      "loss": 3.5502,
+      "step": 1901
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.227848522396345e-06,
+      "loss": 3.5845,
+      "step": 1902
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.22677594858718e-06,
+      "loss": 3.5582,
+      "step": 1903
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.225702692777982e-06,
+      "loss": 3.5275,
+      "step": 1904
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.224628755141927e-06,
+      "loss": 3.6643,
+      "step": 1905
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.223554135852292e-06,
+      "loss": 3.432,
+      "step": 1906
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.222478835082471e-06,
+      "loss": 3.4473,
+      "step": 1907
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.221402853005968e-06,
+      "loss": 3.5178,
+      "step": 1908
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.220326189796394e-06,
+      "loss": 3.5974,
+      "step": 1909
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.219248845627472e-06,
+      "loss": 3.6293,
+      "step": 1910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.218170820673032e-06,
+      "loss": 3.5774,
+      "step": 1911
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.217092115107016e-06,
+      "loss": 3.5229,
+      "step": 1912
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.216012729103476e-06,
+      "loss": 3.5639,
+      "step": 1913
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.214932662836576e-06,
+      "loss": 3.5927,
+      "step": 1914
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.213851916480583e-06,
+      "loss": 3.5775,
+      "step": 1915
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.21277049020988e-06,
+      "loss": 3.6401,
+      "step": 1916
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.211688384198958e-06,
+      "loss": 3.5781,
+      "step": 1917
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.210605598622419e-06,
+      "loss": 3.653,
+      "step": 1918
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.20952213365497e-06,
+      "loss": 3.5331,
+      "step": 1919
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.208437989471428e-06,
+      "loss": 3.5797,
+      "step": 1920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.207353166246729e-06,
+      "loss": 3.5305,
+      "step": 1921
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.206267664155906e-06,
+      "loss": 3.5958,
+      "step": 1922
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.205181483374111e-06,
+      "loss": 3.5996,
+      "step": 1923
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.2040946240766e-06,
+      "loss": 3.5488,
+      "step": 1924
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.20300708643874e-06,
+      "loss": 3.6308,
+      "step": 1925
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.20191887063601e-06,
+      "loss": 3.6293,
+      "step": 1926
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.200829976843992e-06,
+      "loss": 3.6071,
+      "step": 1927
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.199740405238386e-06,
+      "loss": 3.576,
+      "step": 1928
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.198650155994994e-06,
+      "loss": 3.6012,
+      "step": 1929
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.19755922928973e-06,
+      "loss": 3.6008,
+      "step": 1930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.19646762529862e-06,
+      "loss": 3.726,
+      "step": 1931
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.195375344197794e-06,
+      "loss": 3.7055,
+      "step": 1932
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.194282386163496e-06,
+      "loss": 3.535,
+      "step": 1933
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.193188751372078e-06,
+      "loss": 3.4441,
+      "step": 1934
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.192094439999997e-06,
+      "loss": 3.6377,
+      "step": 1935
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.190999452223826e-06,
+      "loss": 3.5556,
+      "step": 1936
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.189903788220243e-06,
+      "loss": 3.6281,
+      "step": 1937
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.188807448166035e-06,
+      "loss": 3.5413,
+      "step": 1938
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.1877104322381e-06,
+      "loss": 3.64,
+      "step": 1939
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.186612740613443e-06,
+      "loss": 3.5202,
+      "step": 1940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.18551437346918e-06,
+      "loss": 3.5964,
+      "step": 1941
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.184415330982535e-06,
+      "loss": 3.5966,
+      "step": 1942
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.183315613330841e-06,
+      "loss": 3.6954,
+      "step": 1943
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.182215220691539e-06,
+      "loss": 3.6607,
+      "step": 1944
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.181114153242181e-06,
+      "loss": 3.6124,
+      "step": 1945
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.180012411160428e-06,
+      "loss": 3.6293,
+      "step": 1946
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.178909994624046e-06,
+      "loss": 3.6577,
+      "step": 1947
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.177806903810913e-06,
+      "loss": 3.4774,
+      "step": 1948
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.176703138899015e-06,
+      "loss": 3.5726,
+      "step": 1949
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.175598700066448e-06,
+      "loss": 3.5752,
+      "step": 1950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.174493587491416e-06,
+      "loss": 3.5973,
+      "step": 1951
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.173387801352232e-06,
+      "loss": 3.5083,
+      "step": 1952
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.172281341827313e-06,
+      "loss": 3.528,
+      "step": 1953
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.171174209095194e-06,
+      "loss": 3.5963,
+      "step": 1954
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.170066403334512e-06,
+      "loss": 3.5582,
+      "step": 1955
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.168957924724014e-06,
+      "loss": 3.5424,
+      "step": 1956
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.167848773442554e-06,
+      "loss": 3.5205,
+      "step": 1957
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.166738949669095e-06,
+      "loss": 3.6264,
+      "step": 1958
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.165628453582714e-06,
+      "loss": 3.6018,
+      "step": 1959
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.16451728536259e-06,
+      "loss": 3.5669,
+      "step": 1960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.16340544518801e-06,
+      "loss": 3.6343,
+      "step": 1961
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.162292933238375e-06,
+      "loss": 3.5405,
+      "step": 1962
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.16117974969319e-06,
+      "loss": 3.5179,
+      "step": 1963
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.160065894732073e-06,
+      "loss": 3.5695,
+      "step": 1964
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.158951368534743e-06,
+      "loss": 3.618,
+      "step": 1965
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.157836171281032e-06,
+      "loss": 3.5685,
+      "step": 1966
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.15672030315088e-06,
+      "loss": 3.5994,
+      "step": 1967
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.155603764324336e-06,
+      "loss": 3.6269,
+      "step": 1968
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.154486554981556e-06,
+      "loss": 3.6186,
+      "step": 1969
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.153368675302804e-06,
+      "loss": 3.5683,
+      "step": 1970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.152250125468451e-06,
+      "loss": 3.5879,
+      "step": 1971
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.15113090565898e-06,
+      "loss": 3.5949,
+      "step": 1972
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.150011016054978e-06,
+      "loss": 3.5518,
+      "step": 1973
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.148890456837143e-06,
+      "loss": 3.6633,
+      "step": 1974
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.147769228186277e-06,
+      "loss": 3.6096,
+      "step": 1975
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.146647330283296e-06,
+      "loss": 3.6542,
+      "step": 1976
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.145524763309219e-06,
+      "loss": 3.6183,
+      "step": 1977
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.144401527445176e-06,
+      "loss": 3.5048,
+      "step": 1978
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.143277622872405e-06,
+      "loss": 3.5178,
+      "step": 1979
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.142153049772247e-06,
+      "loss": 3.5561,
+      "step": 1980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.141027808326155e-06,
+      "loss": 3.5748,
+      "step": 1981
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.139901898715692e-06,
+      "loss": 3.5753,
+      "step": 1982
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.138775321122524e-06,
+      "loss": 3.5905,
+      "step": 1983
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.137648075728428e-06,
+      "loss": 3.6011,
+      "step": 1984
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.136520162715288e-06,
+      "loss": 3.6766,
+      "step": 1985
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.135391582265092e-06,
+      "loss": 3.6714,
+      "step": 1986
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.134262334559943e-06,
+      "loss": 3.5809,
+      "step": 1987
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.133132419782048e-06,
+      "loss": 3.6637,
+      "step": 1988
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.132001838113715e-06,
+      "loss": 3.607,
+      "step": 1989
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.130870589737374e-06,
+      "loss": 3.4624,
+      "step": 1990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.129738674835551e-06,
+      "loss": 3.5599,
+      "step": 1991
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.128606093590882e-06,
+      "loss": 3.5256,
+      "step": 1992
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.127472846186115e-06,
+      "loss": 3.6087,
+      "step": 1993
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.126338932804098e-06,
+      "loss": 3.5525,
+      "step": 1994
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.125204353627792e-06,
+      "loss": 3.5623,
+      "step": 1995
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.124069108840266e-06,
+      "loss": 3.482,
+      "step": 1996
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.12293319862469e-06,
+      "loss": 3.5851,
+      "step": 1997
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.12179662316435e-06,
+      "loss": 3.6714,
+      "step": 1998
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.120659382642633e-06,
+      "loss": 3.5125,
+      "step": 1999
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.119521477243039e-06,
+      "loss": 3.5189,
+      "step": 2000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.118382907149164e-06,
+      "loss": 3.5863,
+      "step": 2001
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.117243672544727e-06,
+      "loss": 3.631,
+      "step": 2002
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.11610377361354e-06,
+      "loss": 3.5731,
+      "step": 2003
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.114963210539531e-06,
+      "loss": 3.6429,
+      "step": 2004
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.113821983506735e-06,
+      "loss": 3.4692,
+      "step": 2005
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.112680092699286e-06,
+      "loss": 3.4931,
+      "step": 2006
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.111537538301436e-06,
+      "loss": 3.5615,
+      "step": 2007
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.110394320497535e-06,
+      "loss": 3.6205,
+      "step": 2008
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.109250439472046e-06,
+      "loss": 3.4608,
+      "step": 2009
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.108105895409535e-06,
+      "loss": 3.543,
+      "step": 2010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.10696068849468e-06,
+      "loss": 3.5756,
+      "step": 2011
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.105814818912258e-06,
+      "loss": 3.5115,
+      "step": 2012
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.104668286847162e-06,
+      "loss": 3.5735,
+      "step": 2013
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.103521092484386e-06,
+      "loss": 3.6059,
+      "step": 2014
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.102373236009033e-06,
+      "loss": 3.6578,
+      "step": 2015
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.101224717606309e-06,
+      "loss": 3.5702,
+      "step": 2016
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.100075537461537e-06,
+      "loss": 3.5516,
+      "step": 2017
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.098925695760132e-06,
+      "loss": 3.657,
+      "step": 2018
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.09777519268763e-06,
+      "loss": 3.6417,
+      "step": 2019
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.096624028429664e-06,
+      "loss": 3.5492,
+      "step": 2020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.095472203171975e-06,
+      "loss": 3.4884,
+      "step": 2021
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.09431971710042e-06,
+      "loss": 3.6105,
+      "step": 2022
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.093166570400947e-06,
+      "loss": 3.6159,
+      "step": 2023
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.092012763259623e-06,
+      "loss": 3.5648,
+      "step": 2024
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.090858295862616e-06,
+      "loss": 3.5349,
+      "step": 2025
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.089703168396205e-06,
+      "loss": 3.5829,
+      "step": 2026
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.08854738104677e-06,
+      "loss": 3.5248,
+      "step": 2027
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.0873909340008e-06,
+      "loss": 3.5901,
+      "step": 2028
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.08623382744489e-06,
+      "loss": 3.6081,
+      "step": 2029
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.085076061565743e-06,
+      "loss": 3.5969,
+      "step": 2030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.083917636550165e-06,
+      "loss": 3.5887,
+      "step": 2031
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.082758552585076e-06,
+      "loss": 3.5706,
+      "step": 2032
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.08159880985749e-06,
+      "loss": 3.5591,
+      "step": 2033
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.080438408554539e-06,
+      "loss": 3.5517,
+      "step": 2034
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.079277348863454e-06,
+      "loss": 3.5901,
+      "step": 2035
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.078115630971573e-06,
+      "loss": 3.5454,
+      "step": 2036
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.076953255066348e-06,
+      "loss": 3.5499,
+      "step": 2037
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.075790221335325e-06,
+      "loss": 3.612,
+      "step": 2038
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.074626529966164e-06,
+      "loss": 3.6073,
+      "step": 2039
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.07346218114663e-06,
+      "loss": 3.5885,
+      "step": 2040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.072297175064595e-06,
+      "loss": 3.5469,
+      "step": 2041
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.071131511908031e-06,
+      "loss": 3.4909,
+      "step": 2042
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.069965191865025e-06,
+      "loss": 3.6792,
+      "step": 2043
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.068798215123761e-06,
+      "loss": 3.608,
+      "step": 2044
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.067630581872537e-06,
+      "loss": 3.5046,
+      "step": 2045
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.066462292299752e-06,
+      "loss": 3.6628,
+      "step": 2046
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.06529334659391e-06,
+      "loss": 3.5689,
+      "step": 2047
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.064123744943628e-06,
+      "loss": 3.6773,
+      "step": 2048
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.062953487537622e-06,
+      "loss": 3.6665,
+      "step": 2049
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.061782574564713e-06,
+      "loss": 3.4647,
+      "step": 2050
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.060611006213833e-06,
+      "loss": 3.5672,
+      "step": 2051
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.059438782674016e-06,
+      "loss": 3.6887,
+      "step": 2052
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.058265904134405e-06,
+      "loss": 3.5206,
+      "step": 2053
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.057092370784244e-06,
+      "loss": 3.5698,
+      "step": 2054
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.055918182812887e-06,
+      "loss": 3.4452,
+      "step": 2055
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.054743340409791e-06,
+      "loss": 3.542,
+      "step": 2056
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.053567843764522e-06,
+      "loss": 3.6889,
+      "step": 2057
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.052391693066744e-06,
+      "loss": 3.445,
+      "step": 2058
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.051214888506237e-06,
+      "loss": 3.4835,
+      "step": 2059
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.050037430272879e-06,
+      "loss": 3.5808,
+      "step": 2060
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.048859318556656e-06,
+      "loss": 3.4636,
+      "step": 2061
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.047680553547656e-06,
+      "loss": 3.6025,
+      "step": 2062
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.04650113543608e-06,
+      "loss": 3.6347,
+      "step": 2063
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.045321064412226e-06,
+      "loss": 3.5443,
+      "step": 2064
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.044140340666505e-06,
+      "loss": 3.6184,
+      "step": 2065
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.042958964389428e-06,
+      "loss": 3.6299,
+      "step": 2066
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.041776935771612e-06,
+      "loss": 3.5819,
+      "step": 2067
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.040594255003778e-06,
+      "loss": 3.7408,
+      "step": 2068
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.039410922276759e-06,
+      "loss": 3.6101,
+      "step": 2069
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.038226937781484e-06,
+      "loss": 3.5084,
+      "step": 2070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.037042301708995e-06,
+      "loss": 3.46,
+      "step": 2071
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.035857014250435e-06,
+      "loss": 3.6373,
+      "step": 2072
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.034671075597053e-06,
+      "loss": 3.5985,
+      "step": 2073
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.0334844859402e-06,
+      "loss": 3.5475,
+      "step": 2074
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.032297245471337e-06,
+      "loss": 3.6271,
+      "step": 2075
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.031109354382031e-06,
+      "loss": 3.5639,
+      "step": 2076
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.029920812863946e-06,
+      "loss": 3.544,
+      "step": 2077
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.028731621108857e-06,
+      "loss": 3.5591,
+      "step": 2078
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.027541779308647e-06,
+      "loss": 3.5927,
+      "step": 2079
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.026351287655294e-06,
+      "loss": 3.5508,
+      "step": 2080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.02516014634089e-06,
+      "loss": 3.6394,
+      "step": 2081
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.023968355557627e-06,
+      "loss": 3.7006,
+      "step": 2082
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.022775915497805e-06,
+      "loss": 3.4855,
+      "step": 2083
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.021582826353825e-06,
+      "loss": 3.6176,
+      "step": 2084
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.020389088318195e-06,
+      "loss": 3.5859,
+      "step": 2085
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.01919470158353e-06,
+      "loss": 3.6573,
+      "step": 2086
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.01799966634254e-06,
+      "loss": 3.6486,
+      "step": 2087
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.016803982788056e-06,
+      "loss": 3.5955,
+      "step": 2088
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.015607651112998e-06,
+      "loss": 3.5371,
+      "step": 2089
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.014410671510399e-06,
+      "loss": 3.6246,
+      "step": 2090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.013213044173394e-06,
+      "loss": 3.5488,
+      "step": 2091
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.012014769295222e-06,
+      "loss": 3.4892,
+      "step": 2092
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.01081584706923e-06,
+      "loss": 3.5217,
+      "step": 2093
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.009616277688867e-06,
+      "loss": 3.5362,
+      "step": 2094
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.00841606134768e-06,
+      "loss": 3.5463,
+      "step": 2095
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.007215198239334e-06,
+      "loss": 3.5312,
+      "step": 2096
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.006013688557588e-06,
+      "loss": 3.6357,
+      "step": 2097
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.004811532496308e-06,
+      "loss": 3.5486,
+      "step": 2098
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.003608730249465e-06,
+      "loss": 3.5147,
+      "step": 2099
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.002405282011135e-06,
+      "loss": 3.5406,
+      "step": 2100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.001201187975497e-06,
+      "loss": 3.5993,
+      "step": 2101
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.999996448336834e-06,
+      "loss": 3.5602,
+      "step": 2102
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.99879106328953e-06,
+      "loss": 3.555,
+      "step": 2103
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.997585033028082e-06,
+      "loss": 3.5112,
+      "step": 2104
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.996378357747085e-06,
+      "loss": 3.5518,
+      "step": 2105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.995171037641235e-06,
+      "loss": 3.5874,
+      "step": 2106
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.993963072905338e-06,
+      "loss": 3.5283,
+      "step": 2107
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.992754463734303e-06,
+      "loss": 3.6068,
+      "step": 2108
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.991545210323143e-06,
+      "loss": 3.5823,
+      "step": 2109
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.990335312866969e-06,
+      "loss": 3.601,
+      "step": 2110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.989124771561006e-06,
+      "loss": 3.6365,
+      "step": 2111
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.987913586600573e-06,
+      "loss": 3.6225,
+      "step": 2112
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.986701758181103e-06,
+      "loss": 3.4653,
+      "step": 2113
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.98548928649812e-06,
+      "loss": 3.5796,
+      "step": 2114
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.984276171747266e-06,
+      "loss": 3.4472,
+      "step": 2115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.983062414124276e-06,
+      "loss": 3.6087,
+      "step": 2116
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.981848013824995e-06,
+      "loss": 3.4869,
+      "step": 2117
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.980632971045365e-06,
+      "loss": 3.5038,
+      "step": 2118
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.97941728598144e-06,
+      "loss": 3.5214,
+      "step": 2119
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.978200958829373e-06,
+      "loss": 3.396,
+      "step": 2120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.976983989785421e-06,
+      "loss": 3.5669,
+      "step": 2121
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.975766379045945e-06,
+      "loss": 3.7016,
+      "step": 2122
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.974548126807407e-06,
+      "loss": 3.5167,
+      "step": 2123
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.973329233266377e-06,
+      "loss": 3.5733,
+      "step": 2124
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.972109698619525e-06,
+      "loss": 3.6701,
+      "step": 2125
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.970889523063627e-06,
+      "loss": 3.6505,
+      "step": 2126
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.969668706795561e-06,
+      "loss": 3.5949,
+      "step": 2127
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.968447250012307e-06,
+      "loss": 3.5726,
+      "step": 2128
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.967225152910953e-06,
+      "loss": 3.6247,
+      "step": 2129
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.966002415688684e-06,
+      "loss": 3.6438,
+      "step": 2130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.964779038542793e-06,
+      "loss": 3.5668,
+      "step": 2131
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.963555021670674e-06,
+      "loss": 3.4518,
+      "step": 2132
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.962330365269825e-06,
+      "loss": 3.5476,
+      "step": 2133
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.961105069537846e-06,
+      "loss": 3.5551,
+      "step": 2134
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.959879134672445e-06,
+      "loss": 3.6562,
+      "step": 2135
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.958652560871426e-06,
+      "loss": 3.5932,
+      "step": 2136
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.957425348332701e-06,
+      "loss": 3.68,
+      "step": 2137
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.956197497254284e-06,
+      "loss": 3.628,
+      "step": 2138
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.95496900783429e-06,
+      "loss": 3.5658,
+      "step": 2139
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.953739880270939e-06,
+      "loss": 3.71,
+      "step": 2140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.952510114762554e-06,
+      "loss": 3.5877,
+      "step": 2141
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.951279711507562e-06,
+      "loss": 3.5277,
+      "step": 2142
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.950048670704488e-06,
+      "loss": 3.5771,
+      "step": 2143
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.948816992551966e-06,
+      "loss": 3.5566,
+      "step": 2144
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.947584677248729e-06,
+      "loss": 3.6758,
+      "step": 2145
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.946351724993613e-06,
+      "loss": 3.623,
+      "step": 2146
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.94511813598556e-06,
+      "loss": 3.6129,
+      "step": 2147
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.943883910423612e-06,
+      "loss": 3.5846,
+      "step": 2148
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.942649048506913e-06,
+      "loss": 3.6017,
+      "step": 2149
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.94141355043471e-06,
+      "loss": 3.5378,
+      "step": 2150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.940177416406357e-06,
+      "loss": 3.5644,
+      "step": 2151
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.938940646621302e-06,
+      "loss": 3.5003,
+      "step": 2152
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.937703241279106e-06,
+      "loss": 3.5508,
+      "step": 2153
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.936465200579422e-06,
+      "loss": 3.5507,
+      "step": 2154
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.935226524722013e-06,
+      "loss": 3.6736,
+      "step": 2155
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.933987213906745e-06,
+      "loss": 3.5101,
+      "step": 2156
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.932747268333582e-06,
+      "loss": 3.5151,
+      "step": 2157
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.931506688202588e-06,
+      "loss": 3.5867,
+      "step": 2158
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.930265473713939e-06,
+      "loss": 3.6661,
+      "step": 2159
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.929023625067902e-06,
+      "loss": 3.5873,
+      "step": 2160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.92778114246486e-06,
+      "loss": 3.5667,
+      "step": 2161
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.926538026105285e-06,
+      "loss": 3.663,
+      "step": 2162
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.925294276189756e-06,
+      "loss": 3.6447,
+      "step": 2163
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.924049892918958e-06,
+      "loss": 3.5406,
+      "step": 2164
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.922804876493674e-06,
+      "loss": 3.6533,
+      "step": 2165
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.921559227114789e-06,
+      "loss": 3.4641,
+      "step": 2166
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.920312944983293e-06,
+      "loss": 3.6167,
+      "step": 2167
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.919066030300277e-06,
+      "loss": 3.5709,
+      "step": 2168
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.91781848326693e-06,
+      "loss": 3.553,
+      "step": 2169
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.91657030408455e-06,
+      "loss": 3.5018,
+      "step": 2170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.915321492954534e-06,
+      "loss": 3.5185,
+      "step": 2171
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.914072050078377e-06,
+      "loss": 3.5959,
+      "step": 2172
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.912821975657683e-06,
+      "loss": 3.5022,
+      "step": 2173
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.911571269894154e-06,
+      "loss": 3.6363,
+      "step": 2174
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.910319932989594e-06,
+      "loss": 3.5964,
+      "step": 2175
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.909067965145905e-06,
+      "loss": 3.539,
+      "step": 2176
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.9078153665651e-06,
+      "loss": 3.574,
+      "step": 2177
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.906562137449288e-06,
+      "loss": 3.5919,
+      "step": 2178
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.905308278000678e-06,
+      "loss": 3.6116,
+      "step": 2179
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.904053788421586e-06,
+      "loss": 3.6095,
+      "step": 2180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.902798668914422e-06,
+      "loss": 3.4549,
+      "step": 2181
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.901542919681709e-06,
+      "loss": 3.615,
+      "step": 2182
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.900286540926062e-06,
+      "loss": 3.5336,
+      "step": 2183
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.899029532850199e-06,
+      "loss": 3.5211,
+      "step": 2184
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.897771895656942e-06,
+      "loss": 3.6048,
+      "step": 2185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.896513629549214e-06,
+      "loss": 3.5647,
+      "step": 2186
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.895254734730041e-06,
+      "loss": 3.654,
+      "step": 2187
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.893995211402547e-06,
+      "loss": 3.5037,
+      "step": 2188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.892735059769956e-06,
+      "loss": 3.4959,
+      "step": 2189
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.891474280035601e-06,
+      "loss": 3.5422,
+      "step": 2190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.89021287240291e-06,
+      "loss": 3.672,
+      "step": 2191
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.888950837075413e-06,
+      "loss": 3.705,
+      "step": 2192
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.887688174256746e-06,
+      "loss": 3.6494,
+      "step": 2193
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.886424884150637e-06,
+      "loss": 3.4889,
+      "step": 2194
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.885160966960925e-06,
+      "loss": 3.5786,
+      "step": 2195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.883896422891545e-06,
+      "loss": 3.5792,
+      "step": 2196
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.882631252146531e-06,
+      "loss": 3.6565,
+      "step": 2197
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.881365454930027e-06,
+      "loss": 3.6336,
+      "step": 2198
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.880099031446268e-06,
+      "loss": 3.59,
+      "step": 2199
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.878831981899593e-06,
+      "loss": 3.6041,
+      "step": 2200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.87756430649445e-06,
+      "loss": 3.5903,
+      "step": 2201
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.876296005435374e-06,
+      "loss": 3.5968,
+      "step": 2202
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.875027078927014e-06,
+      "loss": 3.6244,
+      "step": 2203
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.87375752717411e-06,
+      "loss": 3.5084,
+      "step": 2204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.872487350381509e-06,
+      "loss": 3.5163,
+      "step": 2205
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.871216548754157e-06,
+      "loss": 3.5992,
+      "step": 2206
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.869945122497101e-06,
+      "loss": 3.6805,
+      "step": 2207
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.868673071815487e-06,
+      "loss": 3.5925,
+      "step": 2208
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.867400396914565e-06,
+      "loss": 3.6521,
+      "step": 2209
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.866127097999684e-06,
+      "loss": 3.5513,
+      "step": 2210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.86485317527629e-06,
+      "loss": 3.5253,
+      "step": 2211
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.86357862894994e-06,
+      "loss": 3.5045,
+      "step": 2212
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.862303459226281e-06,
+      "loss": 3.601,
+      "step": 2213
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.861027666311064e-06,
+      "loss": 3.6026,
+      "step": 2214
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.859751250410143e-06,
+      "loss": 3.6265,
+      "step": 2215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.85847421172947e-06,
+      "loss": 3.5724,
+      "step": 2216
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.857196550475097e-06,
+      "loss": 3.5027,
+      "step": 2217
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.855918266853178e-06,
+      "loss": 3.5206,
+      "step": 2218
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.85463936106997e-06,
+      "loss": 3.539,
+      "step": 2219
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.853359833331822e-06,
+      "loss": 3.6251,
+      "step": 2220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.852079683845194e-06,
+      "loss": 3.6375,
+      "step": 2221
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.850798912816637e-06,
+      "loss": 3.5641,
+      "step": 2222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.84951752045281e-06,
+      "loss": 3.5796,
+      "step": 2223
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.848235506960466e-06,
+      "loss": 3.508,
+      "step": 2224
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.84695287254646e-06,
+      "loss": 3.5342,
+      "step": 2225
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.84566961741775e-06,
+      "loss": 3.4671,
+      "step": 2226
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.844385741781394e-06,
+      "loss": 3.5813,
+      "step": 2227
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.843101245844543e-06,
+      "loss": 3.7404,
+      "step": 2228
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.841816129814458e-06,
+      "loss": 3.6477,
+      "step": 2229
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.840530393898494e-06,
+      "loss": 3.4147,
+      "step": 2230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.839244038304106e-06,
+      "loss": 3.5976,
+      "step": 2231
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.837957063238853e-06,
+      "loss": 3.608,
+      "step": 2232
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.83666946891039e-06,
+      "loss": 3.6458,
+      "step": 2233
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.835381255526474e-06,
+      "loss": 3.6388,
+      "step": 2234
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.834092423294958e-06,
+      "loss": 3.5814,
+      "step": 2235
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.832802972423803e-06,
+      "loss": 3.5541,
+      "step": 2236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.831512903121063e-06,
+      "loss": 3.559,
+      "step": 2237
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.83022221559489e-06,
+      "loss": 3.5534,
+      "step": 2238
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.828930910053545e-06,
+      "loss": 3.5562,
+      "step": 2239
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.82763898670538e-06,
+      "loss": 3.5689,
+      "step": 2240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.826346445758852e-06,
+      "loss": 3.47,
+      "step": 2241
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.825053287422512e-06,
+      "loss": 3.5886,
+      "step": 2242
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.823759511905017e-06,
+      "loss": 3.572,
+      "step": 2243
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.82246511941512e-06,
+      "loss": 3.6124,
+      "step": 2244
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.821170110161674e-06,
+      "loss": 3.6122,
+      "step": 2245
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.81987448435363e-06,
+      "loss": 3.5121,
+      "step": 2246
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.818578242200042e-06,
+      "loss": 3.5637,
+      "step": 2247
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.817281383910064e-06,
+      "loss": 3.5663,
+      "step": 2248
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.815983909692941e-06,
+      "loss": 3.4717,
+      "step": 2249
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.814685819758032e-06,
+      "loss": 3.5372,
+      "step": 2250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.813387114314778e-06,
+      "loss": 3.6008,
+      "step": 2251
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.812087793572731e-06,
+      "loss": 3.6336,
+      "step": 2252
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.810787857741542e-06,
+      "loss": 3.6436,
+      "step": 2253
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.809487307030956e-06,
+      "loss": 3.7188,
+      "step": 2254
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.808186141650822e-06,
+      "loss": 3.5893,
+      "step": 2255
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.806884361811086e-06,
+      "loss": 3.5597,
+      "step": 2256
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.80558196772179e-06,
+      "loss": 3.4418,
+      "step": 2257
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.804278959593082e-06,
+      "loss": 3.6226,
+      "step": 2258
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.802975337635202e-06,
+      "loss": 3.5993,
+      "step": 2259
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.801671102058495e-06,
+      "loss": 3.6672,
+      "step": 2260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.800366253073402e-06,
+      "loss": 3.5702,
+      "step": 2261
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.79906079089046e-06,
+      "loss": 3.4887,
+      "step": 2262
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.797754715720314e-06,
+      "loss": 3.7433,
+      "step": 2263
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.7964480277737e-06,
+      "loss": 3.4815,
+      "step": 2264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.795140727261453e-06,
+      "loss": 3.5197,
+      "step": 2265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.79383281439451e-06,
+      "loss": 3.5512,
+      "step": 2266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.792524289383909e-06,
+      "loss": 3.5112,
+      "step": 2267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.791215152440778e-06,
+      "loss": 3.5599,
+      "step": 2268
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.789905403776353e-06,
+      "loss": 3.595,
+      "step": 2269
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.788595043601963e-06,
+      "loss": 3.5048,
+      "step": 2270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.787284072129039e-06,
+      "loss": 3.503,
+      "step": 2271
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.78597248956911e-06,
+      "loss": 3.57,
+      "step": 2272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.7846602961338e-06,
+      "loss": 3.564,
+      "step": 2273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.783347492034839e-06,
+      "loss": 3.5696,
+      "step": 2274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.782034077484044e-06,
+      "loss": 3.6152,
+      "step": 2275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.780720052693346e-06,
+      "loss": 3.4771,
+      "step": 2276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.77940541787476e-06,
+      "loss": 3.5776,
+      "step": 2277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.778090173240405e-06,
+      "loss": 3.5278,
+      "step": 2278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.776774319002504e-06,
+      "loss": 3.4659,
+      "step": 2279
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.775457855373367e-06,
+      "loss": 3.6274,
+      "step": 2280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.774140782565413e-06,
+      "loss": 3.5333,
+      "step": 2281
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.772823100791152e-06,
+      "loss": 3.6372,
+      "step": 2282
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.771504810263196e-06,
+      "loss": 3.5962,
+      "step": 2283
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.770185911194253e-06,
+      "loss": 3.5739,
+      "step": 2284
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.768866403797134e-06,
+      "loss": 3.4967,
+      "step": 2285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.76754628828474e-06,
+      "loss": 3.5638,
+      "step": 2286
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.766225564870078e-06,
+      "loss": 3.6042,
+      "step": 2287
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.764904233766249e-06,
+      "loss": 3.5957,
+      "step": 2288
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.76358229518645e-06,
+      "loss": 3.5646,
+      "step": 2289
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.762259749343984e-06,
+      "loss": 3.5614,
+      "step": 2290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.760936596452242e-06,
+      "loss": 3.5913,
+      "step": 2291
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.75961283672472e-06,
+      "loss": 3.5079,
+      "step": 2292
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.75828847037501e-06,
+      "loss": 3.4178,
+      "step": 2293
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.756963497616801e-06,
+      "loss": 3.5176,
+      "step": 2294
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.75563791866388e-06,
+      "loss": 3.5939,
+      "step": 2295
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.754311733730133e-06,
+      "loss": 3.5749,
+      "step": 2296
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.752984943029543e-06,
+      "loss": 3.6537,
+      "step": 2297
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.75165754677619e-06,
+      "loss": 3.5012,
+      "step": 2298
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.750329545184253e-06,
+      "loss": 3.5479,
+      "step": 2299
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.749000938468006e-06,
+      "loss": 3.6074,
+      "step": 2300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.747671726841826e-06,
+      "loss": 3.5973,
+      "step": 2301
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.746341910520184e-06,
+      "loss": 3.5486,
+      "step": 2302
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.745011489717647e-06,
+      "loss": 3.5444,
+      "step": 2303
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.743680464648882e-06,
+      "loss": 3.5483,
+      "step": 2304
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.742348835528656e-06,
+      "loss": 3.5632,
+      "step": 2305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.741016602571826e-06,
+      "loss": 3.5102,
+      "step": 2306
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.739683765993352e-06,
+      "loss": 3.5666,
+      "step": 2307
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.738350326008291e-06,
+      "loss": 3.6723,
+      "step": 2308
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.737016282831797e-06,
+      "loss": 3.5949,
+      "step": 2309
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.735681636679121e-06,
+      "loss": 3.6804,
+      "step": 2310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.73434638776561e-06,
+      "loss": 3.6756,
+      "step": 2311
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.73301053630671e-06,
+      "loss": 3.58,
+      "step": 2312
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.731674082517965e-06,
+      "loss": 3.5959,
+      "step": 2313
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.730337026615011e-06,
+      "loss": 3.6837,
+      "step": 2314
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.728999368813591e-06,
+      "loss": 3.5081,
+      "step": 2315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.727661109329535e-06,
+      "loss": 3.6351,
+      "step": 2316
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.726322248378775e-06,
+      "loss": 3.5949,
+      "step": 2317
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.724982786177338e-06,
+      "loss": 3.6552,
+      "step": 2318
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.723642722941352e-06,
+      "loss": 3.4646,
+      "step": 2319
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.72230205888704e-06,
+      "loss": 3.6004,
+      "step": 2320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.720960794230718e-06,
+      "loss": 3.6749,
+      "step": 2321
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.719618929188802e-06,
+      "loss": 3.5833,
+      "step": 2322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.718276463977808e-06,
+      "loss": 3.6785,
+      "step": 2323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.716933398814345e-06,
+      "loss": 3.5929,
+      "step": 2324
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.715589733915117e-06,
+      "loss": 3.5475,
+      "step": 2325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.714245469496932e-06,
+      "loss": 3.6746,
+      "step": 2326
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.712900605776686e-06,
+      "loss": 3.6069,
+      "step": 2327
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.711555142971377e-06,
+      "loss": 3.6158,
+      "step": 2328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.710209081298102e-06,
+      "loss": 3.4882,
+      "step": 2329
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.708862420974045e-06,
+      "loss": 3.5244,
+      "step": 2330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.707515162216496e-06,
+      "loss": 3.596,
+      "step": 2331
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.70616730524284e-06,
+      "loss": 3.7118,
+      "step": 2332
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.704818850270554e-06,
+      "loss": 3.5866,
+      "step": 2333
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.703469797517216e-06,
+      "loss": 3.5889,
+      "step": 2334
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.702120147200497e-06,
+      "loss": 3.5973,
+      "step": 2335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.700769899538168e-06,
+      "loss": 3.561,
+      "step": 2336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.699419054748094e-06,
+      "loss": 3.5842,
+      "step": 2337
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.698067613048236e-06,
+      "loss": 3.5816,
+      "step": 2338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.696715574656654e-06,
+      "loss": 3.6439,
+      "step": 2339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.6953629397915e-06,
+      "loss": 3.6058,
+      "step": 2340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.694009708671028e-06,
+      "loss": 3.6342,
+      "step": 2341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.692655881513582e-06,
+      "loss": 3.5927,
+      "step": 2342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.691301458537607e-06,
+      "loss": 3.4529,
+      "step": 2343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.689946439961642e-06,
+      "loss": 3.5654,
+      "step": 2344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.688590826004322e-06,
+      "loss": 3.5551,
+      "step": 2345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.687234616884377e-06,
+      "loss": 3.4488,
+      "step": 2346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.685877812820637e-06,
+      "loss": 3.6196,
+      "step": 2347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.684520414032023e-06,
+      "loss": 3.583,
+      "step": 2348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.683162420737558e-06,
+      "loss": 3.5679,
+      "step": 2349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.681803833156354e-06,
+      "loss": 3.5641,
+      "step": 2350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.680444651507624e-06,
+      "loss": 3.5621,
+      "step": 2351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.679084876010672e-06,
+      "loss": 3.5725,
+      "step": 2352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.677724506884906e-06,
+      "loss": 3.5191,
+      "step": 2353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.676363544349822e-06,
+      "loss": 3.6966,
+      "step": 2354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.675001988625013e-06,
+      "loss": 3.6465,
+      "step": 2355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.67363983993017e-06,
+      "loss": 3.6708,
+      "step": 2356
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.672277098485082e-06,
+      "loss": 3.5733,
+      "step": 2357
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.670913764509626e-06,
+      "loss": 3.5785,
+      "step": 2358
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.669549838223782e-06,
+      "loss": 3.6105,
+      "step": 2359
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.668185319847622e-06,
+      "loss": 3.6531,
+      "step": 2360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.666820209601313e-06,
+      "loss": 3.5401,
+      "step": 2361
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.66545450770512e-06,
+      "loss": 3.6561,
+      "step": 2362
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.6640882143794e-06,
+      "loss": 3.5747,
+      "step": 2363
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.662721329844611e-06,
+      "loss": 3.5595,
+      "step": 2364
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.661353854321301e-06,
+      "loss": 3.579,
+      "step": 2365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.659985788030114e-06,
+      "loss": 3.5661,
+      "step": 2366
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.658617131191794e-06,
+      "loss": 3.4356,
+      "step": 2367
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.657247884027175e-06,
+      "loss": 3.6252,
+      "step": 2368
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.655878046757187e-06,
+      "loss": 3.5055,
+      "step": 2369
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.654507619602858e-06,
+      "loss": 3.5197,
+      "step": 2370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.65313660278531e-06,
+      "loss": 3.5642,
+      "step": 2371
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.65176499652576e-06,
+      "loss": 3.5012,
+      "step": 2372
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.650392801045517e-06,
+      "loss": 3.5806,
+      "step": 2373
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.64902001656599e-06,
+      "loss": 3.5757,
+      "step": 2374
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.647646643308682e-06,
+      "loss": 3.5698,
+      "step": 2375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.646272681495186e-06,
+      "loss": 3.5664,
+      "step": 2376
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.644898131347199e-06,
+      "loss": 3.51,
+      "step": 2377
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.643522993086504e-06,
+      "loss": 3.5465,
+      "step": 2378
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.642147266934985e-06,
+      "loss": 3.5876,
+      "step": 2379
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.640770953114617e-06,
+      "loss": 3.6318,
+      "step": 2380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.639394051847472e-06,
+      "loss": 3.5668,
+      "step": 2381
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.638016563355716e-06,
+      "loss": 3.5809,
+      "step": 2382
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.636638487861611e-06,
+      "loss": 3.6656,
+      "step": 2383
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.63525982558751e-06,
+      "loss": 3.5791,
+      "step": 2384
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.633880576755866e-06,
+      "loss": 3.6253,
+      "step": 2385
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.632500741589223e-06,
+      "loss": 3.6492,
+      "step": 2386
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.631120320310223e-06,
+      "loss": 3.5361,
+      "step": 2387
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.629739313141594e-06,
+      "loss": 3.5898,
+      "step": 2388
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.62835772030617e-06,
+      "loss": 3.4773,
+      "step": 2389
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.626975542026875e-06,
+      "loss": 3.429,
+      "step": 2390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.625592778526722e-06,
+      "loss": 3.5938,
+      "step": 2391
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.624209430028826e-06,
+      "loss": 3.5311,
+      "step": 2392
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.622825496756394e-06,
+      "loss": 3.6192,
+      "step": 2393
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.621440978932724e-06,
+      "loss": 3.5562,
+      "step": 2394
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.620055876781213e-06,
+      "loss": 3.6022,
+      "step": 2395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.61867019052535e-06,
+      "loss": 3.61,
+      "step": 2396
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.617283920388721e-06,
+      "loss": 3.537,
+      "step": 2397
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.615897066595001e-06,
+      "loss": 3.6408,
+      "step": 2398
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.614509629367962e-06,
+      "loss": 3.6826,
+      "step": 2399
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.613121608931472e-06,
+      "loss": 3.5201,
+      "step": 2400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.61173300550949e-06,
+      "loss": 3.4682,
+      "step": 2401
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.610343819326071e-06,
+      "loss": 3.5144,
+      "step": 2402
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.608954050605363e-06,
+      "loss": 3.5989,
+      "step": 2403
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.607563699571608e-06,
+      "loss": 3.665,
+      "step": 2404
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.606172766449144e-06,
+      "loss": 3.6167,
+      "step": 2405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.6047812514624e-06,
+      "loss": 3.5298,
+      "step": 2406
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.6033891548359e-06,
+      "loss": 3.5769,
+      "step": 2407
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.60199647679426e-06,
+      "loss": 3.5197,
+      "step": 2408
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.600603217562198e-06,
+      "loss": 3.5854,
+      "step": 2409
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.599209377364513e-06,
+      "loss": 3.5965,
+      "step": 2410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.597814956426106e-06,
+      "loss": 3.5969,
+      "step": 2411
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.596419954971973e-06,
+      "loss": 3.7134,
+      "step": 2412
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.595024373227198e-06,
+      "loss": 3.6008,
+      "step": 2413
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.593628211416964e-06,
+      "loss": 3.5205,
+      "step": 2414
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.592231469766541e-06,
+      "loss": 3.64,
+      "step": 2415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.590834148501298e-06,
+      "loss": 3.6141,
+      "step": 2416
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.589436247846699e-06,
+      "loss": 3.6138,
+      "step": 2417
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.588037768028295e-06,
+      "loss": 3.526,
+      "step": 2418
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.586638709271733e-06,
+      "loss": 3.551,
+      "step": 2419
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.585239071802758e-06,
+      "loss": 3.6643,
+      "step": 2420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.583838855847202e-06,
+      "loss": 3.5474,
+      "step": 2421
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.582438061630993e-06,
+      "loss": 3.6066,
+      "step": 2422
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.581036689380156e-06,
+      "loss": 3.5776,
+      "step": 2423
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.579634739320802e-06,
+      "loss": 3.7028,
+      "step": 2424
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.578232211679139e-06,
+      "loss": 3.4881,
+      "step": 2425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.57682910668147e-06,
+      "loss": 3.5842,
+      "step": 2426
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.575425424554187e-06,
+      "loss": 3.6196,
+      "step": 2427
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.574021165523779e-06,
+      "loss": 3.484,
+      "step": 2428
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.572616329816827e-06,
+      "loss": 3.6415,
+      "step": 2429
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.571210917660002e-06,
+      "loss": 3.6082,
+      "step": 2430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.569804929280072e-06,
+      "loss": 3.6039,
+      "step": 2431
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.568398364903898e-06,
+      "loss": 3.6023,
+      "step": 2432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.566991224758429e-06,
+      "loss": 3.4951,
+      "step": 2433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.565583509070715e-06,
+      "loss": 3.5357,
+      "step": 2434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.564175218067889e-06,
+      "loss": 3.4663,
+      "step": 2435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.562766351977183e-06,
+      "loss": 3.5203,
+      "step": 2436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.561356911025924e-06,
+      "loss": 3.5493,
+      "step": 2437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.559946895441527e-06,
+      "loss": 3.6309,
+      "step": 2438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.558536305451498e-06,
+      "loss": 3.6263,
+      "step": 2439
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.557125141283444e-06,
+      "loss": 3.6601,
+      "step": 2440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.555713403165058e-06,
+      "loss": 3.4794,
+      "step": 2441
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.554301091324126e-06,
+      "loss": 3.6922,
+      "step": 2442
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.552888205988527e-06,
+      "loss": 3.5831,
+      "step": 2443
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.551474747386236e-06,
+      "loss": 3.6163,
+      "step": 2444
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.550060715745316e-06,
+      "loss": 3.6094,
+      "step": 2445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.548646111293926e-06,
+      "loss": 3.5468,
+      "step": 2446
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.547230934260313e-06,
+      "loss": 3.5555,
+      "step": 2447
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.54581518487282e-06,
+      "loss": 3.5606,
+      "step": 2448
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.544398863359882e-06,
+      "loss": 3.5887,
+      "step": 2449
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.542981969950027e-06,
+      "loss": 3.4902,
+      "step": 2450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.54156450487187e-06,
+      "loss": 3.5946,
+      "step": 2451
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.54014646835413e-06,
+      "loss": 3.6073,
+      "step": 2452
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.5387278606256e-06,
+      "loss": 3.6042,
+      "step": 2453
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.537308681915184e-06,
+      "loss": 3.5561,
+      "step": 2454
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.535888932451866e-06,
+      "loss": 3.5104,
+      "step": 2455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.534468612464727e-06,
+      "loss": 3.6846,
+      "step": 2456
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.533047722182937e-06,
+      "loss": 3.5539,
+      "step": 2457
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.53162626183576e-06,
+      "loss": 3.5751,
+      "step": 2458
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.530204231652557e-06,
+      "loss": 3.6845,
+      "step": 2459
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.528781631862772e-06,
+      "loss": 3.6929,
+      "step": 2460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.527358462695943e-06,
+      "loss": 3.6369,
+      "step": 2461
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.525934724381703e-06,
+      "loss": 3.5557,
+      "step": 2462
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.524510417149778e-06,
+      "loss": 3.5001,
+      "step": 2463
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.523085541229981e-06,
+      "loss": 3.6416,
+      "step": 2464
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.521660096852218e-06,
+      "loss": 3.4373,
+      "step": 2465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.52023408424649e-06,
+      "loss": 3.636,
+      "step": 2466
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.518807503642885e-06,
+      "loss": 3.5658,
+      "step": 2467
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.517380355271588e-06,
+      "loss": 3.6511,
+      "step": 2468
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.51595263936287e-06,
+      "loss": 3.5783,
+      "step": 2469
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.514524356147099e-06,
+      "loss": 3.5775,
+      "step": 2470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.513095505854728e-06,
+      "loss": 3.4571,
+      "step": 2471
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.511666088716309e-06,
+      "loss": 3.562,
+      "step": 2472
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.51023610496248e-06,
+      "loss": 3.5354,
+      "step": 2473
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.508805554823971e-06,
+      "loss": 3.621,
+      "step": 2474
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.507374438531606e-06,
+      "loss": 3.7211,
+      "step": 2475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.5059427563163e-06,
+      "loss": 3.5753,
+      "step": 2476
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.504510508409057e-06,
+      "loss": 3.497,
+      "step": 2477
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.503077695040974e-06,
+      "loss": 3.4837,
+      "step": 2478
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.501644316443239e-06,
+      "loss": 3.5862,
+      "step": 2479
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.500210372847128e-06,
+      "loss": 3.5727,
+      "step": 2480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.498775864484014e-06,
+      "loss": 3.5518,
+      "step": 2481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.497340791585358e-06,
+      "loss": 3.5178,
+      "step": 2482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.495905154382712e-06,
+      "loss": 3.554,
+      "step": 2483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.494468953107717e-06,
+      "loss": 3.6614,
+      "step": 2484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.493032187992113e-06,
+      "loss": 3.7004,
+      "step": 2485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.49159485926772e-06,
+      "loss": 3.6259,
+      "step": 2486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.490156967166456e-06,
+      "loss": 3.6094,
+      "step": 2487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.488718511920331e-06,
+      "loss": 3.5567,
+      "step": 2488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.487279493761438e-06,
+      "loss": 3.7432,
+      "step": 2489
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.48583991292197e-06,
+      "loss": 3.5053,
+      "step": 2490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.484399769634204e-06,
+      "loss": 3.4778,
+      "step": 2491
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.482959064130512e-06,
+      "loss": 3.5447,
+      "step": 2492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.481517796643353e-06,
+      "loss": 3.5279,
+      "step": 2493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.480075967405284e-06,
+      "loss": 3.5922,
+      "step": 2494
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.478633576648942e-06,
+      "loss": 3.6412,
+      "step": 2495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.47719062460706e-06,
+      "loss": 3.5983,
+      "step": 2496
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.475747111512468e-06,
+      "loss": 3.5221,
+      "step": 2497
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.474303037598074e-06,
+      "loss": 3.7463,
+      "step": 2498
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.472858403096884e-06,
+      "loss": 3.6575,
+      "step": 2499
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.471413208241993e-06,
+      "loss": 3.5724,
+      "step": 2500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.46996745326659e-06,
+      "loss": 3.5463,
+      "step": 2501
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.468521138403946e-06,
+      "loss": 3.499,
+      "step": 2502
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.467074263887429e-06,
+      "loss": 3.6753,
+      "step": 2503
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.465626829950497e-06,
+      "loss": 3.4174,
+      "step": 2504
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.464178836826694e-06,
+      "loss": 3.5851,
+      "step": 2505
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.462730284749661e-06,
+      "loss": 3.5084,
+      "step": 2506
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.46128117395312e-06,
+      "loss": 3.6364,
+      "step": 2507
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.459831504670893e-06,
+      "loss": 3.5186,
+      "step": 2508
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.458381277136885e-06,
+      "loss": 3.5293,
+      "step": 2509
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.456930491585094e-06,
+      "loss": 3.5497,
+      "step": 2510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.455479148249609e-06,
+      "loss": 3.6893,
+      "step": 2511
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.454027247364606e-06,
+      "loss": 3.639,
+      "step": 2512
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.452574789164352e-06,
+      "loss": 3.6262,
+      "step": 2513
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.451121773883206e-06,
+      "loss": 3.5259,
+      "step": 2514
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.449668201755617e-06,
+      "loss": 3.6079,
+      "step": 2515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.448214073016117e-06,
+      "loss": 3.5258,
+      "step": 2516
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.446759387899338e-06,
+      "loss": 3.499,
+      "step": 2517
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.445304146639992e-06,
+      "loss": 3.5902,
+      "step": 2518
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.443848349472892e-06,
+      "loss": 3.5127,
+      "step": 2519
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.442391996632928e-06,
+      "loss": 3.6069,
+      "step": 2520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.440935088355089e-06,
+      "loss": 3.5917,
+      "step": 2521
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.43947762487445e-06,
+      "loss": 3.6788,
+      "step": 2522
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.438019606426178e-06,
+      "loss": 3.5723,
+      "step": 2523
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.436561033245523e-06,
+      "loss": 3.6094,
+      "step": 2524
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.435101905567833e-06,
+      "loss": 3.6324,
+      "step": 2525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.43364222362854e-06,
+      "loss": 3.6362,
+      "step": 2526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.432181987663168e-06,
+      "loss": 3.5805,
+      "step": 2527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.430721197907328e-06,
+      "loss": 3.6422,
+      "step": 2528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.429259854596725e-06,
+      "loss": 3.6002,
+      "step": 2529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.427797957967146e-06,
+      "loss": 3.5312,
+      "step": 2530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.426335508254473e-06,
+      "loss": 3.6679,
+      "step": 2531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.424872505694676e-06,
+      "loss": 3.6761,
+      "step": 2532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.423408950523816e-06,
+      "loss": 3.5508,
+      "step": 2533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.421944842978036e-06,
+      "loss": 3.5491,
+      "step": 2534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.420480183293576e-06,
+      "loss": 3.4759,
+      "step": 2535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.419014971706763e-06,
+      "loss": 3.4787,
+      "step": 2536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.41754920845401e-06,
+      "loss": 3.5515,
+      "step": 2537
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.416082893771823e-06,
+      "loss": 3.6134,
+      "step": 2538
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.414616027896796e-06,
+      "loss": 3.4877,
+      "step": 2539
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.41314861106561e-06,
+      "loss": 3.677,
+      "step": 2540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.411680643515035e-06,
+      "loss": 3.5186,
+      "step": 2541
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.410212125481931e-06,
+      "loss": 3.4999,
+      "step": 2542
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.40874305720325e-06,
+      "loss": 3.6957,
+      "step": 2543
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.407273438916027e-06,
+      "loss": 3.4733,
+      "step": 2544
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.405803270857387e-06,
+      "loss": 3.526,
+      "step": 2545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.404332553264548e-06,
+      "loss": 3.5902,
+      "step": 2546
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.402861286374811e-06,
+      "loss": 3.6606,
+      "step": 2547
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.40138947042557e-06,
+      "loss": 3.7087,
+      "step": 2548
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.399917105654306e-06,
+      "loss": 3.6332,
+      "step": 2549
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.39844419229859e-06,
+      "loss": 3.5872,
+      "step": 2550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.396970730596074e-06,
+      "loss": 3.6019,
+      "step": 2551
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.395496720784511e-06,
+      "loss": 3.5933,
+      "step": 2552
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.394022163101732e-06,
+      "loss": 3.5576,
+      "step": 2553
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.392547057785662e-06,
+      "loss": 3.5877,
+      "step": 2554
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.39107140507431e-06,
+      "loss": 3.5727,
+      "step": 2555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.389595205205781e-06,
+      "loss": 3.6469,
+      "step": 2556
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.388118458418259e-06,
+      "loss": 3.6763,
+      "step": 2557
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.386641164950023e-06,
+      "loss": 3.6356,
+      "step": 2558
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.385163325039437e-06,
+      "loss": 3.5068,
+      "step": 2559
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.383684938924952e-06,
+      "loss": 3.6318,
+      "step": 2560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.382206006845111e-06,
+      "loss": 3.5881,
+      "step": 2561
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.380726529038541e-06,
+      "loss": 3.6061,
+      "step": 2562
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.37924650574396e-06,
+      "loss": 3.4196,
+      "step": 2563
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.377765937200178e-06,
+      "loss": 3.5114,
+      "step": 2564
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.37628482364608e-06,
+      "loss": 3.5389,
+      "step": 2565
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.374803165320653e-06,
+      "loss": 3.6459,
+      "step": 2566
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.373320962462965e-06,
+      "loss": 3.5914,
+      "step": 2567
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.37183821531217e-06,
+      "loss": 3.5756,
+      "step": 2568
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.370354924107515e-06,
+      "loss": 3.613,
+      "step": 2569
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.368871089088333e-06,
+      "loss": 3.5767,
+      "step": 2570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.367386710494042e-06,
+      "loss": 3.6275,
+      "step": 2571
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.365901788564152e-06,
+      "loss": 3.6018,
+      "step": 2572
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.364416323538255e-06,
+      "loss": 3.491,
+      "step": 2573
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.36293031565604e-06,
+      "loss": 3.5205,
+      "step": 2574
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.361443765157274e-06,
+      "loss": 3.6516,
+      "step": 2575
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.359956672281816e-06,
+      "loss": 3.5378,
+      "step": 2576
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.358469037269612e-06,
+      "loss": 3.5635,
+      "step": 2577
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.356980860360694e-06,
+      "loss": 3.6537,
+      "step": 2578
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.355492141795185e-06,
+      "loss": 3.5382,
+      "step": 2579
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.354002881813293e-06,
+      "loss": 3.6462,
+      "step": 2580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.35251308065531e-06,
+      "loss": 3.6404,
+      "step": 2581
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.351022738561624e-06,
+      "loss": 3.5567,
+      "step": 2582
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.349531855772701e-06,
+      "loss": 3.4482,
+      "step": 2583
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.3480404325291e-06,
+      "loss": 3.5672,
+      "step": 2584
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.346548469071468e-06,
+      "loss": 3.5651,
+      "step": 2585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.345055965640532e-06,
+      "loss": 3.4525,
+      "step": 2586
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.343562922477115e-06,
+      "loss": 3.591,
+      "step": 2587
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.342069339822118e-06,
+      "loss": 3.5431,
+      "step": 2588
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.34057521791654e-06,
+      "loss": 3.5306,
+      "step": 2589
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.339080557001458e-06,
+      "loss": 3.61,
+      "step": 2590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.337585357318039e-06,
+      "loss": 3.5616,
+      "step": 2591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.336089619107536e-06,
+      "loss": 3.5754,
+      "step": 2592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.334593342611292e-06,
+      "loss": 3.5579,
+      "step": 2593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.333096528070733e-06,
+      "loss": 3.5425,
+      "step": 2594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.331599175727374e-06,
+      "loss": 3.5709,
+      "step": 2595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.330101285822816e-06,
+      "loss": 3.6233,
+      "step": 2596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.328602858598747e-06,
+      "loss": 3.5805,
+      "step": 2597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.327103894296944e-06,
+      "loss": 3.675,
+      "step": 2598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.325604393159263e-06,
+      "loss": 3.496,
+      "step": 2599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.324104355427659e-06,
+      "loss": 3.4602,
+      "step": 2600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.32260378134416e-06,
+      "loss": 3.487,
+      "step": 2601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.32110267115089e-06,
+      "loss": 3.5985,
+      "step": 2602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.319601025090058e-06,
+      "loss": 3.4983,
+      "step": 2603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.318098843403955e-06,
+      "loss": 3.6574,
+      "step": 2604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.316596126334964e-06,
+      "loss": 3.576,
+      "step": 2605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.31509287412555e-06,
+      "loss": 3.5302,
+      "step": 2606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.313589087018267e-06,
+      "loss": 3.7204,
+      "step": 2607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.312084765255754e-06,
+      "loss": 3.6403,
+      "step": 2608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.310579909080739e-06,
+      "loss": 3.5467,
+      "step": 2609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.309074518736032e-06,
+      "loss": 3.5886,
+      "step": 2610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.30756859446453e-06,
+      "loss": 3.5,
+      "step": 2611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.30606213650922e-06,
+      "loss": 3.5011,
+      "step": 2612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.304555145113172e-06,
+      "loss": 3.6371,
+      "step": 2613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.303047620519539e-06,
+      "loss": 3.6442,
+      "step": 2614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.301539562971568e-06,
+      "loss": 3.6253,
+      "step": 2615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.300030972712588e-06,
+      "loss": 3.582,
+      "step": 2616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.29852184998601e-06,
+      "loss": 3.5035,
+      "step": 2617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.297012195035336e-06,
+      "loss": 3.5858,
+      "step": 2618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.295502008104151e-06,
+      "loss": 3.5802,
+      "step": 2619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.29399128943613e-06,
+      "loss": 3.6367,
+      "step": 2620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.292480039275028e-06,
+      "loss": 3.673,
+      "step": 2621
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.290968257864691e-06,
+      "loss": 3.6286,
+      "step": 2622
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.289455945449047e-06,
+      "loss": 3.4942,
+      "step": 2623
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.287943102272112e-06,
+      "loss": 3.6096,
+      "step": 2624
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.286429728577987e-06,
+      "loss": 3.6697,
+      "step": 2625
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.284915824610855e-06,
+      "loss": 3.5219,
+      "step": 2626
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.283401390614992e-06,
+      "loss": 3.4739,
+      "step": 2627
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.281886426834753e-06,
+      "loss": 3.5439,
+      "step": 2628
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.280370933514583e-06,
+      "loss": 3.5489,
+      "step": 2629
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.278854910899008e-06,
+      "loss": 3.5944,
+      "step": 2630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.277338359232645e-06,
+      "loss": 3.5636,
+      "step": 2631
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.275821278760188e-06,
+      "loss": 3.6133,
+      "step": 2632
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.274303669726427e-06,
+      "loss": 3.6415,
+      "step": 2633
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.272785532376227e-06,
+      "loss": 3.6937,
+      "step": 2634
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.271266866954546e-06,
+      "loss": 3.6393,
+      "step": 2635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.269747673706423e-06,
+      "loss": 3.7391,
+      "step": 2636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.268227952876984e-06,
+      "loss": 3.5683,
+      "step": 2637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.266707704711438e-06,
+      "loss": 3.5128,
+      "step": 2638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.265186929455082e-06,
+      "loss": 3.6123,
+      "step": 2639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.263665627353297e-06,
+      "loss": 3.657,
+      "step": 2640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.262143798651546e-06,
+      "loss": 3.5271,
+      "step": 2641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.260621443595382e-06,
+      "loss": 3.4826,
+      "step": 2642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.25909856243044e-06,
+      "loss": 3.6177,
+      "step": 2643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.25757515540244e-06,
+      "loss": 3.6453,
+      "step": 2644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.256051222757188e-06,
+      "loss": 3.5094,
+      "step": 2645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.254526764740572e-06,
+      "loss": 3.422,
+      "step": 2646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.25300178159857e-06,
+      "loss": 3.588,
+      "step": 2647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.251476273577239e-06,
+      "loss": 3.6872,
+      "step": 2648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.249950240922724e-06,
+      "loss": 3.5237,
+      "step": 2649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.248423683881252e-06,
+      "loss": 3.5565,
+      "step": 2650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.246896602699141e-06,
+      "loss": 3.6575,
+      "step": 2651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.245368997622786e-06,
+      "loss": 3.5774,
+      "step": 2652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.243840868898669e-06,
+      "loss": 3.4754,
+      "step": 2653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.24231221677336e-06,
+      "loss": 3.6396,
+      "step": 2654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.240783041493507e-06,
+      "loss": 3.5537,
+      "step": 2655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.239253343305849e-06,
+      "loss": 3.6384,
+      "step": 2656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.237723122457204e-06,
+      "loss": 3.5097,
+      "step": 2657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.236192379194476e-06,
+      "loss": 3.5203,
+      "step": 2658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.234661113764659e-06,
+      "loss": 3.4956,
+      "step": 2659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.233129326414819e-06,
+      "loss": 3.5319,
+      "step": 2660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.23159701739212e-06,
+      "loss": 3.5787,
+      "step": 2661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.230064186943801e-06,
+      "loss": 3.5226,
+      "step": 2662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.228530835317185e-06,
+      "loss": 3.4915,
+      "step": 2663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.226996962759686e-06,
+      "loss": 3.6589,
+      "step": 2664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.225462569518796e-06,
+      "loss": 3.644,
+      "step": 2665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.223927655842093e-06,
+      "loss": 3.5997,
+      "step": 2666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.22239222197724e-06,
+      "loss": 3.5108,
+      "step": 2667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.220856268171979e-06,
+      "loss": 3.5248,
+      "step": 2668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.219319794674146e-06,
+      "loss": 3.4344,
+      "step": 2669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.217782801731649e-06,
+      "loss": 3.6074,
+      "step": 2670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.216245289592486e-06,
+      "loss": 3.5272,
+      "step": 2671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.21470725850474e-06,
+      "loss": 3.5945,
+      "step": 2672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.213168708716577e-06,
+      "loss": 3.5415,
+      "step": 2673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.211629640476242e-06,
+      "loss": 3.6722,
+      "step": 2674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.210090054032071e-06,
+      "loss": 3.5886,
+      "step": 2675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.208549949632476e-06,
+      "loss": 3.5372,
+      "step": 2676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.20700932752596e-06,
+      "loss": 3.5263,
+      "step": 2677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.2054681879611e-06,
+      "loss": 3.6096,
+      "step": 2678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.203926531186571e-06,
+      "loss": 3.6636,
+      "step": 2679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.202384357451117e-06,
+      "loss": 3.5668,
+      "step": 2680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.200841667003572e-06,
+      "loss": 3.5384,
+      "step": 2681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.199298460092855e-06,
+      "loss": 3.7262,
+      "step": 2682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.197754736967964e-06,
+      "loss": 3.5193,
+      "step": 2683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.19621049787798e-06,
+      "loss": 3.5322,
+      "step": 2684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.194665743072076e-06,
+      "loss": 3.5067,
+      "step": 2685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.193120472799496e-06,
+      "loss": 3.6577,
+      "step": 2686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.191574687309574e-06,
+      "loss": 3.5548,
+      "step": 2687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.190028386851729e-06,
+      "loss": 3.5795,
+      "step": 2688
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.188481571675458e-06,
+      "loss": 3.6724,
+      "step": 2689
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.18693424203034e-06,
+      "loss": 3.6075,
+      "step": 2690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.185386398166046e-06,
+      "loss": 3.6437,
+      "step": 2691
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.183838040332321e-06,
+      "loss": 3.6274,
+      "step": 2692
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.182289168778997e-06,
+      "loss": 3.5623,
+      "step": 2693
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.180739783755989e-06,
+      "loss": 3.714,
+      "step": 2694
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.17918988551329e-06,
+      "loss": 3.4677,
+      "step": 2695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.177639474300984e-06,
+      "loss": 3.5262,
+      "step": 2696
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.17608855036923e-06,
+      "loss": 3.7565,
+      "step": 2697
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.174537113968275e-06,
+      "loss": 3.6326,
+      "step": 2698
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.172985165348448e-06,
+      "loss": 3.6294,
+      "step": 2699
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.171432704760159e-06,
+      "loss": 3.6893,
+      "step": 2700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.169879732453896e-06,
+      "loss": 3.5169,
+      "step": 2701
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.168326248680243e-06,
+      "loss": 3.454,
+      "step": 2702
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.166772253689852e-06,
+      "loss": 3.5795,
+      "step": 2703
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.165217747733465e-06,
+      "loss": 3.5901,
+      "step": 2704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.163662731061908e-06,
+      "loss": 3.5283,
+      "step": 2705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.162107203926082e-06,
+      "loss": 3.5589,
+      "step": 2706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.160551166576978e-06,
+      "loss": 3.5507,
+      "step": 2707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.158994619265665e-06,
+      "loss": 3.4924,
+      "step": 2708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.157437562243298e-06,
+      "loss": 3.6629,
+      "step": 2709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.155879995761106e-06,
+      "loss": 3.5919,
+      "step": 2710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.154321920070415e-06,
+      "loss": 3.5961,
+      "step": 2711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.152763335422612e-06,
+      "loss": 3.6717,
+      "step": 2712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.151204242069189e-06,
+      "loss": 3.6025,
+      "step": 2713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.149644640261704e-06,
+      "loss": 3.6563,
+      "step": 2714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.148084530251805e-06,
+      "loss": 3.5659,
+      "step": 2715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.146523912291217e-06,
+      "loss": 3.5493,
+      "step": 2716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.14496278663175e-06,
+      "loss": 3.4726,
+      "step": 2717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.143401153525296e-06,
+      "loss": 3.503,
+      "step": 2718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.141839013223826e-06,
+      "loss": 3.6747,
+      "step": 2719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.140276365979397e-06,
+      "loss": 3.4861,
+      "step": 2720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.138713212044147e-06,
+      "loss": 3.4837,
+      "step": 2721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.13714955167029e-06,
+      "loss": 3.5931,
+      "step": 2722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.135585385110129e-06,
+      "loss": 3.5343,
+      "step": 2723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.134020712616049e-06,
+      "loss": 3.5154,
+      "step": 2724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.132455534440506e-06,
+      "loss": 3.6802,
+      "step": 2725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.130889850836053e-06,
+      "loss": 3.5133,
+      "step": 2726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.12932366205531e-06,
+      "loss": 3.5822,
+      "step": 2727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.127756968350991e-06,
+      "loss": 3.5719,
+      "step": 2728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.12618976997588e-06,
+      "loss": 3.4859,
+      "step": 2729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.124622067182852e-06,
+      "loss": 3.6001,
+      "step": 2730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.123053860224859e-06,
+      "loss": 3.616,
+      "step": 2731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.121485149354933e-06,
+      "loss": 3.5888,
+      "step": 2732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.119915934826192e-06,
+      "loss": 3.6968,
+      "step": 2733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.11834621689183e-06,
+      "loss": 3.4632,
+      "step": 2734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.116775995805125e-06,
+      "loss": 3.5512,
+      "step": 2735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.115205271819436e-06,
+      "loss": 3.5722,
+      "step": 2736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.1136340451882e-06,
+      "loss": 3.6344,
+      "step": 2737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.112062316164945e-06,
+      "loss": 3.4908,
+      "step": 2738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.110490085003267e-06,
+      "loss": 3.617,
+      "step": 2739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.108917351956853e-06,
+      "loss": 3.5654,
+      "step": 2740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.107344117279463e-06,
+      "loss": 3.4626,
+      "step": 2741
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.105770381224946e-06,
+      "loss": 3.5851,
+      "step": 2742
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.104196144047224e-06,
+      "loss": 3.5635,
+      "step": 2743
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.10262140600031e-06,
+      "loss": 3.5057,
+      "step": 2744
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.101046167338284e-06,
+      "loss": 3.7551,
+      "step": 2745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.09947042831532e-06,
+      "loss": 3.6741,
+      "step": 2746
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.097894189185664e-06,
+      "loss": 3.4798,
+      "step": 2747
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.096317450203648e-06,
+      "loss": 3.6139,
+      "step": 2748
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.094740211623682e-06,
+      "loss": 3.5589,
+      "step": 2749
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.093162473700255e-06,
+      "loss": 3.5627,
+      "step": 2750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.091584236687941e-06,
+      "loss": 3.6279,
+      "step": 2751
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.09000550084139e-06,
+      "loss": 3.5367,
+      "step": 2752
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.08842626641534e-06,
+      "loss": 3.6412,
+      "step": 2753
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.086846533664597e-06,
+      "loss": 3.6052,
+      "step": 2754
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.085266302844058e-06,
+      "loss": 3.7209,
+      "step": 2755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.083685574208694e-06,
+      "loss": 3.7207,
+      "step": 2756
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.082104348013567e-06,
+      "loss": 3.4659,
+      "step": 2757
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.080522624513802e-06,
+      "loss": 3.5228,
+      "step": 2758
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.07894040396462e-06,
+      "loss": 3.625,
+      "step": 2759
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.077357686621311e-06,
+      "loss": 3.5713,
+      "step": 2760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.075774472739258e-06,
+      "loss": 3.6204,
+      "step": 2761
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.074190762573907e-06,
+      "loss": 3.532,
+      "step": 2762
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.072606556380799e-06,
+      "loss": 3.5604,
+      "step": 2763
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.071021854415548e-06,
+      "loss": 3.6056,
+      "step": 2764
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.069436656933849e-06,
+      "loss": 3.6032,
+      "step": 2765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.067850964191476e-06,
+      "loss": 3.4837,
+      "step": 2766
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.066264776444285e-06,
+      "loss": 3.6373,
+      "step": 2767
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.064678093948215e-06,
+      "loss": 3.5931,
+      "step": 2768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.063090916959275e-06,
+      "loss": 3.4752,
+      "step": 2769
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.061503245733562e-06,
+      "loss": 3.6203,
+      "step": 2770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.059915080527251e-06,
+      "loss": 3.4833,
+      "step": 2771
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.058326421596595e-06,
+      "loss": 3.55,
+      "step": 2772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.056737269197931e-06,
+      "loss": 3.5875,
+      "step": 2773
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.055147623587669e-06,
+      "loss": 3.5029,
+      "step": 2774
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.053557485022302e-06,
+      "loss": 3.6291,
+      "step": 2775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.051966853758401e-06,
+      "loss": 3.6578,
+      "step": 2776
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.050375730052622e-06,
+      "loss": 3.4615,
+      "step": 2777
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.048784114161696e-06,
+      "loss": 3.6052,
+      "step": 2778
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.047192006342432e-06,
+      "loss": 3.4028,
+      "step": 2779
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.04559940685172e-06,
+      "loss": 3.5973,
+      "step": 2780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.044006315946531e-06,
+      "loss": 3.6198,
+      "step": 2781
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.042412733883915e-06,
+      "loss": 3.6657,
+      "step": 2782
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.040818660920996e-06,
+      "loss": 3.566,
+      "step": 2783
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.039224097314987e-06,
+      "loss": 3.6716,
+      "step": 2784
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.037629043323169e-06,
+      "loss": 3.5012,
+      "step": 2785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.03603349920291e-06,
+      "loss": 3.6741,
+      "step": 2786
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.034437465211655e-06,
+      "loss": 3.5564,
+      "step": 2787
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.032840941606927e-06,
+      "loss": 3.5389,
+      "step": 2788
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.031243928646331e-06,
+      "loss": 3.5693,
+      "step": 2789
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.029646426587547e-06,
+      "loss": 3.5711,
+      "step": 2790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.028048435688333e-06,
+      "loss": 3.5746,
+      "step": 2791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.026449956206535e-06,
+      "loss": 3.4791,
+      "step": 2792
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.024850988400064e-06,
+      "loss": 3.5484,
+      "step": 2793
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.023251532526923e-06,
+      "loss": 3.5807,
+      "step": 2794
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.021651588845185e-06,
+      "loss": 3.5263,
+      "step": 2795
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.020051157613003e-06,
+      "loss": 3.599,
+      "step": 2796
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.018450239088615e-06,
+      "loss": 3.4778,
+      "step": 2797
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.016848833530328e-06,
+      "loss": 3.4841,
+      "step": 2798
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.015246941196538e-06,
+      "loss": 3.4895,
+      "step": 2799
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.013644562345708e-06,
+      "loss": 3.6155,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.01204169723639e-06,
+      "loss": 3.6191,
+      "step": 2801
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.010438346127208e-06,
+      "loss": 3.5973,
+      "step": 2802
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.008834509276864e-06,
+      "loss": 3.649,
+      "step": 2803
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.007230186944146e-06,
+      "loss": 3.5734,
+      "step": 2804
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.00562537938791e-06,
+      "loss": 3.6562,
+      "step": 2805
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.004020086867101e-06,
+      "loss": 3.5512,
+      "step": 2806
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.002414309640734e-06,
+      "loss": 3.687,
+      "step": 2807
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.000808047967903e-06,
+      "loss": 3.5946,
+      "step": 2808
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.999201302107783e-06,
+      "loss": 3.6325,
+      "step": 2809
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.997594072319625e-06,
+      "loss": 3.523,
+      "step": 2810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.995986358862763e-06,
+      "loss": 3.6456,
+      "step": 2811
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.994378161996604e-06,
+      "loss": 3.4611,
+      "step": 2812
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.992769481980632e-06,
+      "loss": 3.663,
+      "step": 2813
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.99116031907441e-06,
+      "loss": 3.5504,
+      "step": 2814
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.989550673537588e-06,
+      "loss": 3.4776,
+      "step": 2815
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.987940545629878e-06,
+      "loss": 3.6024,
+      "step": 2816
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.986329935611078e-06,
+      "loss": 3.4896,
+      "step": 2817
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.98471884374107e-06,
+      "loss": 3.6205,
+      "step": 2818
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.983107270279805e-06,
+      "loss": 3.5328,
+      "step": 2819
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.981495215487309e-06,
+      "loss": 3.5071,
+      "step": 2820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.979882679623696e-06,
+      "loss": 3.5418,
+      "step": 2821
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.97826966294915e-06,
+      "loss": 3.5259,
+      "step": 2822
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.976656165723935e-06,
+      "loss": 3.5376,
+      "step": 2823
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.975042188208396e-06,
+      "loss": 3.5995,
+      "step": 2824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.973427730662946e-06,
+      "loss": 3.4791,
+      "step": 2825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.971812793348085e-06,
+      "loss": 3.6052,
+      "step": 2826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.970197376524389e-06,
+      "loss": 3.6121,
+      "step": 2827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.968581480452505e-06,
+      "loss": 3.7566,
+      "step": 2828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.966965105393165e-06,
+      "loss": 3.5215,
+      "step": 2829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.96534825160717e-06,
+      "loss": 3.5013,
+      "step": 2830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.963730919355409e-06,
+      "loss": 3.5183,
+      "step": 2831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.962113108898838e-06,
+      "loss": 3.4272,
+      "step": 2832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.960494820498498e-06,
+      "loss": 3.5003,
+      "step": 2833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.9588760544155e-06,
+      "loss": 3.6151,
+      "step": 2834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.957256810911036e-06,
+      "loss": 3.5253,
+      "step": 2835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.955637090246378e-06,
+      "loss": 3.5753,
+      "step": 2836
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.954016892682869e-06,
+      "loss": 3.5759,
+      "step": 2837
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.952396218481931e-06,
+      "loss": 3.5609,
+      "step": 2838
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.950775067905066e-06,
+      "loss": 3.6691,
+      "step": 2839
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.949153441213848e-06,
+      "loss": 3.6297,
+      "step": 2840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.947531338669932e-06,
+      "loss": 3.5685,
+      "step": 2841
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.945908760535045e-06,
+      "loss": 3.5558,
+      "step": 2842
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.944285707070999e-06,
+      "loss": 3.6114,
+      "step": 2843
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.942662178539672e-06,
+      "loss": 3.5714,
+      "step": 2844
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.941038175203027e-06,
+      "loss": 3.5955,
+      "step": 2845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.9394136973231e-06,
+      "loss": 3.659,
+      "step": 2846
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.937788745162005e-06,
+      "loss": 3.5464,
+      "step": 2847
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.936163318981931e-06,
+      "loss": 3.5372,
+      "step": 2848
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.934537419045144e-06,
+      "loss": 3.5724,
+      "step": 2849
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.932911045613986e-06,
+      "loss": 3.636,
+      "step": 2850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.931284198950878e-06,
+      "loss": 3.5195,
+      "step": 2851
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.929656879318315e-06,
+      "loss": 3.5107,
+      "step": 2852
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.928029086978868e-06,
+      "loss": 3.5938,
+      "step": 2853
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.926400822195184e-06,
+      "loss": 3.4936,
+      "step": 2854
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.92477208522999e-06,
+      "loss": 3.5336,
+      "step": 2855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.923142876346086e-06,
+      "loss": 3.5888,
+      "step": 2856
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.921513195806346e-06,
+      "loss": 3.5713,
+      "step": 2857
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.919883043873725e-06,
+      "loss": 3.5766,
+      "step": 2858
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.918252420811252e-06,
+      "loss": 3.5331,
+      "step": 2859
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.916621326882031e-06,
+      "loss": 3.5542,
+      "step": 2860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.91498976234924e-06,
+      "loss": 3.7087,
+      "step": 2861
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.913357727476141e-06,
+      "loss": 3.5993,
+      "step": 2862
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.911725222526063e-06,
+      "loss": 3.6308,
+      "step": 2863
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.910092247762416e-06,
+      "loss": 3.5535,
+      "step": 2864
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.908458803448685e-06,
+      "loss": 3.6611,
+      "step": 2865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.906824889848426e-06,
+      "loss": 3.5161,
+      "step": 2866
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.905190507225278e-06,
+      "loss": 3.5307,
+      "step": 2867
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.903555655842954e-06,
+      "loss": 3.6279,
+      "step": 2868
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.901920335965237e-06,
+      "loss": 3.6026,
+      "step": 2869
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.900284547855992e-06,
+      "loss": 3.5951,
+      "step": 2870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.898648291779156e-06,
+      "loss": 3.6779,
+      "step": 2871
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.897011567998743e-06,
+      "loss": 3.6412,
+      "step": 2872
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.895374376778843e-06,
+      "loss": 3.5291,
+      "step": 2873
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.893736718383622e-06,
+      "loss": 3.5534,
+      "step": 2874
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.892098593077316e-06,
+      "loss": 3.5814,
+      "step": 2875
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.890460001124242e-06,
+      "loss": 3.5178,
+      "step": 2876
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.888820942788793e-06,
+      "loss": 3.6367,
+      "step": 2877
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.88718141833543e-06,
+      "loss": 3.6056,
+      "step": 2878
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.885541428028698e-06,
+      "loss": 3.6177,
+      "step": 2879
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.883900972133213e-06,
+      "loss": 3.6672,
+      "step": 2880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.882260050913664e-06,
+      "loss": 3.6015,
+      "step": 2881
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.880618664634819e-06,
+      "loss": 3.5856,
+      "step": 2882
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.87897681356152e-06,
+      "loss": 3.5655,
+      "step": 2883
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.877334497958682e-06,
+      "loss": 3.5571,
+      "step": 2884
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.875691718091295e-06,
+      "loss": 3.6018,
+      "step": 2885
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.874048474224432e-06,
+      "loss": 3.555,
+      "step": 2886
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.872404766623226e-06,
+      "loss": 3.666,
+      "step": 2887
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.870760595552897e-06,
+      "loss": 3.608,
+      "step": 2888
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.869115961278735e-06,
+      "loss": 3.4242,
+      "step": 2889
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.867470864066107e-06,
+      "loss": 3.5588,
+      "step": 2890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.865825304180452e-06,
+      "loss": 3.5264,
+      "step": 2891
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.864179281887285e-06,
+      "loss": 3.5518,
+      "step": 2892
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.862532797452196e-06,
+      "loss": 3.5389,
+      "step": 2893
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.860885851140847e-06,
+      "loss": 3.5611,
+      "step": 2894
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.85923844321898e-06,
+      "loss": 3.5134,
+      "step": 2895
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.857590573952403e-06,
+      "loss": 3.6829,
+      "step": 2896
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.855942243607008e-06,
+      "loss": 3.5285,
+      "step": 2897
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.854293452448759e-06,
+      "loss": 3.688,
+      "step": 2898
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.852644200743685e-06,
+      "loss": 3.6078,
+      "step": 2899
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.8509944887579e-06,
+      "loss": 3.5976,
+      "step": 2900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.849344316757593e-06,
+      "loss": 3.5761,
+      "step": 2901
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.847693685009016e-06,
+      "loss": 3.534,
+      "step": 2902
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.846042593778506e-06,
+      "loss": 3.5417,
+      "step": 2903
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.84439104333247e-06,
+      "loss": 3.6087,
+      "step": 2904
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.84273903393739e-06,
+      "loss": 3.5072,
+      "step": 2905
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.841086565859819e-06,
+      "loss": 3.5769,
+      "step": 2906
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.83943363936639e-06,
+      "loss": 3.5198,
+      "step": 2907
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.837780254723804e-06,
+      "loss": 3.5467,
+      "step": 2908
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.836126412198842e-06,
+      "loss": 3.6281,
+      "step": 2909
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.834472112058348e-06,
+      "loss": 3.5407,
+      "step": 2910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.832817354569256e-06,
+      "loss": 3.5327,
+      "step": 2911
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.831162139998557e-06,
+      "loss": 3.5644,
+      "step": 2912
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.82950646861333e-06,
+      "loss": 3.5506,
+      "step": 2913
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.827850340680717e-06,
+      "loss": 3.7216,
+      "step": 2914
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.826193756467941e-06,
+      "loss": 3.5979,
+      "step": 2915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.824536716242293e-06,
+      "loss": 3.5092,
+      "step": 2916
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.822879220271142e-06,
+      "loss": 3.5317,
+      "step": 2917
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.82122126882193e-06,
+      "loss": 3.6119,
+      "step": 2918
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.819562862162169e-06,
+      "loss": 3.6188,
+      "step": 2919
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.81790400055945e-06,
+      "loss": 3.5887,
+      "step": 2920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.816244684281426e-06,
+      "loss": 3.4658,
+      "step": 2921
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.814584913595841e-06,
+      "loss": 3.4786,
+      "step": 2922
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.812924688770498e-06,
+      "loss": 3.5155,
+      "step": 2923
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.81126401007328e-06,
+      "loss": 3.5453,
+      "step": 2924
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.80960287777214e-06,
+      "loss": 3.4976,
+      "step": 2925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.807941292135107e-06,
+      "loss": 3.5061,
+      "step": 2926
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.806279253430278e-06,
+      "loss": 3.643,
+      "step": 2927
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.80461676192583e-06,
+      "loss": 3.6274,
+      "step": 2928
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.80295381789001e-06,
+      "loss": 3.459,
+      "step": 2929
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.801290421591136e-06,
+      "loss": 3.6944,
+      "step": 2930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.799626573297605e-06,
+      "loss": 3.6067,
+      "step": 2931
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.797962273277877e-06,
+      "loss": 3.5828,
+      "step": 2932
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.796297521800494e-06,
+      "loss": 3.6502,
+      "step": 2933
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.794632319134065e-06,
+      "loss": 3.6604,
+      "step": 2934
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.792966665547278e-06,
+      "loss": 3.6326,
+      "step": 2935
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.791300561308888e-06,
+      "loss": 3.5032,
+      "step": 2936
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.789634006687725e-06,
+      "loss": 3.5536,
+      "step": 2937
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.787967001952689e-06,
+      "loss": 3.6458,
+      "step": 2938
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.786299547372761e-06,
+      "loss": 3.7063,
+      "step": 2939
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.784631643216982e-06,
+      "loss": 3.5256,
+      "step": 2940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.782963289754478e-06,
+      "loss": 3.6032,
+      "step": 2941
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.781294487254436e-06,
+      "loss": 3.6183,
+      "step": 2942
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.779625235986124e-06,
+      "loss": 3.5651,
+      "step": 2943
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.77795553621888e-06,
+      "loss": 3.5628,
+      "step": 2944
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.776285388222114e-06,
+      "loss": 3.6355,
+      "step": 2945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.774614792265308e-06,
+      "loss": 3.552,
+      "step": 2946
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.772943748618016e-06,
+      "loss": 3.5702,
+      "step": 2947
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.771272257549864e-06,
+      "loss": 3.6797,
+      "step": 2948
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.769600319330553e-06,
+      "loss": 3.5751,
+      "step": 2949
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.76792793422985e-06,
+      "loss": 3.5762,
+      "step": 2950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.766255102517603e-06,
+      "loss": 3.6204,
+      "step": 2951
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.764581824463725e-06,
+      "loss": 3.4172,
+      "step": 2952
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.762908100338205e-06,
+      "loss": 3.4529,
+      "step": 2953
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.761233930411098e-06,
+      "loss": 3.5342,
+      "step": 2954
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.759559314952541e-06,
+      "loss": 3.6621,
+      "step": 2955
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.757884254232733e-06,
+      "loss": 3.6158,
+      "step": 2956
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.756208748521949e-06,
+      "loss": 3.5477,
+      "step": 2957
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.754532798090537e-06,
+      "loss": 3.5236,
+      "step": 2958
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.752856403208914e-06,
+      "loss": 3.6298,
+      "step": 2959
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.751179564147574e-06,
+      "loss": 3.5534,
+      "step": 2960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.749502281177074e-06,
+      "loss": 3.3687,
+      "step": 2961
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.74782455456805e-06,
+      "loss": 3.597,
+      "step": 2962
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.746146384591208e-06,
+      "loss": 3.6068,
+      "step": 2963
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.744467771517322e-06,
+      "loss": 3.6556,
+      "step": 2964
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.742788715617243e-06,
+      "loss": 3.6126,
+      "step": 2965
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.741109217161886e-06,
+      "loss": 3.5918,
+      "step": 2966
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.739429276422245e-06,
+      "loss": 3.5389,
+      "step": 2967
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.737748893669384e-06,
+      "loss": 3.5549,
+      "step": 2968
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.736068069174434e-06,
+      "loss": 3.6883,
+      "step": 2969
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.7343868032086e-06,
+      "loss": 3.6006,
+      "step": 2970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.73270509604316e-06,
+      "loss": 3.5252,
+      "step": 2971
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.731022947949458e-06,
+      "loss": 3.5604,
+      "step": 2972
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.729340359198917e-06,
+      "loss": 3.4809,
+      "step": 2973
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.727657330063023e-06,
+      "loss": 3.5451,
+      "step": 2974
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.725973860813338e-06,
+      "loss": 3.5524,
+      "step": 2975
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.724289951721495e-06,
+      "loss": 3.6382,
+      "step": 2976
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.722605603059193e-06,
+      "loss": 3.527,
+      "step": 2977
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.720920815098207e-06,
+      "loss": 3.51,
+      "step": 2978
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.719235588110382e-06,
+      "loss": 3.5406,
+      "step": 2979
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.717549922367635e-06,
+      "loss": 3.4952,
+      "step": 2980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.715863818141948e-06,
+      "loss": 3.6117,
+      "step": 2981
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.714177275705382e-06,
+      "loss": 3.6117,
+      "step": 2982
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.712490295330062e-06,
+      "loss": 3.5685,
+      "step": 2983
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.710802877288188e-06,
+      "loss": 3.4913,
+      "step": 2984
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.709115021852025e-06,
+      "loss": 3.5239,
+      "step": 2985
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.707426729293917e-06,
+      "loss": 3.5673,
+      "step": 2986
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.70573799988627e-06,
+      "loss": 3.604,
+      "step": 2987
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.704048833901567e-06,
+      "loss": 3.716,
+      "step": 2988
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.702359231612358e-06,
+      "loss": 3.5187,
+      "step": 2989
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.700669193291264e-06,
+      "loss": 3.5447,
+      "step": 2990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.698978719210976e-06,
+      "loss": 3.6292,
+      "step": 2991
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.697287809644258e-06,
+      "loss": 3.576,
+      "step": 2992
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.695596464863942e-06,
+      "loss": 3.6035,
+      "step": 2993
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.693904685142927e-06,
+      "loss": 3.5965,
+      "step": 2994
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.692212470754188e-06,
+      "loss": 3.5501,
+      "step": 2995
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.690519821970768e-06,
+      "loss": 3.5398,
+      "step": 2996
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.688826739065779e-06,
+      "loss": 3.5764,
+      "step": 2997
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.687133222312403e-06,
+      "loss": 3.4793,
+      "step": 2998
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.685439271983895e-06,
+      "loss": 3.5091,
+      "step": 2999
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.683744888353577e-06,
+      "loss": 3.5818,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.68205007169484e-06,
+      "loss": 3.6423,
+      "step": 3001
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.680354822281148e-06,
+      "loss": 3.5514,
+      "step": 3002
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.678659140386034e-06,
+      "loss": 3.5158,
+      "step": 3003
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.6769630262831e-06,
+      "loss": 3.527,
+      "step": 3004
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.675266480246017e-06,
+      "loss": 3.5566,
+      "step": 3005
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.673569502548528e-06,
+      "loss": 3.5205,
+      "step": 3006
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.671872093464441e-06,
+      "loss": 3.5764,
+      "step": 3007
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.67017425326764e-06,
+      "loss": 3.5937,
+      "step": 3008
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.668475982232075e-06,
+      "loss": 3.5523,
+      "step": 3009
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.666777280631766e-06,
+      "loss": 3.5467,
+      "step": 3010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.6650781487408e-06,
+      "loss": 3.5097,
+      "step": 3011
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.663378586833342e-06,
+      "loss": 3.5054,
+      "step": 3012
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.661678595183612e-06,
+      "loss": 3.6437,
+      "step": 3013
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.659978174065912e-06,
+      "loss": 3.5802,
+      "step": 3014
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.65827732375461e-06,
+      "loss": 3.6316,
+      "step": 3015
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.65657604452414e-06,
+      "loss": 3.5652,
+      "step": 3016
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.654874336649007e-06,
+      "loss": 3.53,
+      "step": 3017
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.653172200403787e-06,
+      "loss": 3.5607,
+      "step": 3018
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.651469636063122e-06,
+      "loss": 3.5298,
+      "step": 3019
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.649766643901725e-06,
+      "loss": 3.55,
+      "step": 3020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.64806322419438e-06,
+      "loss": 3.6388,
+      "step": 3021
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.646359377215938e-06,
+      "loss": 3.4942,
+      "step": 3022
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.644655103241314e-06,
+      "loss": 3.5764,
+      "step": 3023
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.6429504025455e-06,
+      "loss": 3.556,
+      "step": 3024
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.641245275403551e-06,
+      "loss": 3.4827,
+      "step": 3025
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.639539722090597e-06,
+      "loss": 3.6653,
+      "step": 3026
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.637833742881829e-06,
+      "loss": 3.5378,
+      "step": 3027
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.636127338052513e-06,
+      "loss": 3.6334,
+      "step": 3028
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.63442050787798e-06,
+      "loss": 3.5043,
+      "step": 3029
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.632713252633632e-06,
+      "loss": 3.509,
+      "step": 3030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.63100557259494e-06,
+      "loss": 3.5183,
+      "step": 3031
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.62929746803744e-06,
+      "loss": 3.6492,
+      "step": 3032
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.6275889392367375e-06,
+      "loss": 3.4842,
+      "step": 3033
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.625879986468509e-06,
+      "loss": 3.6512,
+      "step": 3034
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.624170610008499e-06,
+      "loss": 3.6337,
+      "step": 3035
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.6224608101325185e-06,
+      "loss": 3.5545,
+      "step": 3036
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.620750587116447e-06,
+      "loss": 3.5287,
+      "step": 3037
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.619039941236233e-06,
+      "loss": 3.5733,
+      "step": 3038
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.617328872767894e-06,
+      "loss": 3.5591,
+      "step": 3039
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.615617381987515e-06,
+      "loss": 3.5428,
+      "step": 3040
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.613905469171247e-06,
+      "loss": 3.4809,
+      "step": 3041
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.612193134595313e-06,
+      "loss": 3.66,
+      "step": 3042
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.610480378536002e-06,
+      "loss": 3.5012,
+      "step": 3043
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.60876720126967e-06,
+      "loss": 3.5645,
+      "step": 3044
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.607053603072745e-06,
+      "loss": 3.59,
+      "step": 3045
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.605339584221716e-06,
+      "loss": 3.598,
+      "step": 3046
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.603625144993145e-06,
+      "loss": 3.507,
+      "step": 3047
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.601910285663663e-06,
+      "loss": 3.6405,
+      "step": 3048
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.6001950065099625e-06,
+      "loss": 3.6437,
+      "step": 3049
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.598479307808813e-06,
+      "loss": 3.4897,
+      "step": 3050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.596763189837042e-06,
+      "loss": 3.6799,
+      "step": 3051
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5950466528715515e-06,
+      "loss": 3.5286,
+      "step": 3052
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.593329697189307e-06,
+      "loss": 3.5224,
+      "step": 3053
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.591612323067344e-06,
+      "loss": 3.5263,
+      "step": 3054
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.589894530782766e-06,
+      "loss": 3.5723,
+      "step": 3055
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.588176320612741e-06,
+      "loss": 3.6376,
+      "step": 3056
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.586457692834508e-06,
+      "loss": 3.5142,
+      "step": 3057
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.584738647725369e-06,
+      "loss": 3.6537,
+      "step": 3058
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.583019185562696e-06,
+      "loss": 3.5938,
+      "step": 3059
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5812993066239325e-06,
+      "loss": 3.646,
+      "step": 3060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.57957901118658e-06,
+      "loss": 3.612,
+      "step": 3061
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.577858299528214e-06,
+      "loss": 3.4947,
+      "step": 3062
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.576137171926477e-06,
+      "loss": 3.5856,
+      "step": 3063
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.574415628659073e-06,
+      "loss": 3.5515,
+      "step": 3064
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.572693670003781e-06,
+      "loss": 3.6058,
+      "step": 3065
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.570971296238441e-06,
+      "loss": 3.5863,
+      "step": 3066
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.56924850764096e-06,
+      "loss": 3.5316,
+      "step": 3067
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5675253044893185e-06,
+      "loss": 3.4864,
+      "step": 3068
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.565801687061555e-06,
+      "loss": 3.6256,
+      "step": 3069
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.564077655635781e-06,
+      "loss": 3.6037,
+      "step": 3070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.562353210490174e-06,
+      "loss": 3.4251,
+      "step": 3071
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5606283519029745e-06,
+      "loss": 3.5677,
+      "step": 3072
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5589030801524955e-06,
+      "loss": 3.5694,
+      "step": 3073
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5571773955171124e-06,
+      "loss": 3.583,
+      "step": 3074
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.555451298275267e-06,
+      "loss": 3.6769,
+      "step": 3075
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.55372478870547e-06,
+      "loss": 3.5111,
+      "step": 3076
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.551997867086298e-06,
+      "loss": 3.6254,
+      "step": 3077
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.550270533696394e-06,
+      "loss": 3.6265,
+      "step": 3078
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.548542788814467e-06,
+      "loss": 3.4756,
+      "step": 3079
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.546814632719293e-06,
+      "loss": 3.5913,
+      "step": 3080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.545086065689713e-06,
+      "loss": 3.4775,
+      "step": 3081
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.543357088004635e-06,
+      "loss": 3.515,
+      "step": 3082
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5416276999430345e-06,
+      "loss": 3.5612,
+      "step": 3083
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.539897901783951e-06,
+      "loss": 3.5666,
+      "step": 3084
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.538167693806494e-06,
+      "loss": 3.4427,
+      "step": 3085
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.536437076289833e-06,
+      "loss": 3.5529,
+      "step": 3086
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.534706049513212e-06,
+      "loss": 3.6213,
+      "step": 3087
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.53297461375593e-06,
+      "loss": 3.5876,
+      "step": 3088
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.531242769297363e-06,
+      "loss": 3.6464,
+      "step": 3089
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.529510516416947e-06,
+      "loss": 3.4994,
+      "step": 3090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.527777855394183e-06,
+      "loss": 3.6143,
+      "step": 3091
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.526044786508642e-06,
+      "loss": 3.5827,
+      "step": 3092
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.524311310039958e-06,
+      "loss": 3.4966,
+      "step": 3093
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.52257742626783e-06,
+      "loss": 3.5488,
+      "step": 3094
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.520843135472026e-06,
+      "loss": 3.5362,
+      "step": 3095
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.519108437932379e-06,
+      "loss": 3.5016,
+      "step": 3096
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.5173733339287834e-06,
+      "loss": 3.5183,
+      "step": 3097
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.515637823741203e-06,
+      "loss": 3.4577,
+      "step": 3098
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.513901907649668e-06,
+      "loss": 3.5169,
+      "step": 3099
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.512165585934272e-06,
+      "loss": 3.4337,
+      "step": 3100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.510428858875173e-06,
+      "loss": 3.639,
+      "step": 3101
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.508691726752597e-06,
+      "loss": 3.4712,
+      "step": 3102
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.506954189846833e-06,
+      "loss": 3.4206,
+      "step": 3103
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.505216248438238e-06,
+      "loss": 3.5982,
+      "step": 3104
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.503477902807233e-06,
+      "loss": 3.5435,
+      "step": 3105
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.501739153234304e-06,
+      "loss": 3.6388,
+      "step": 3106
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 3.5925,
+      "step": 3107
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.498260443384941e-06,
+      "loss": 3.5789,
+      "step": 3108
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.4965204836698066e-06,
+      "loss": 3.5719,
+      "step": 3109
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.494780121135343e-06,
+      "loss": 3.6062,
+      "step": 3110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.493039356062361e-06,
+      "loss": 3.5272,
+      "step": 3111
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.4912981887317394e-06,
+      "loss": 3.4262,
+      "step": 3112
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.489556619424418e-06,
+      "loss": 3.6051,
+      "step": 3113
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.4878146484214034e-06,
+      "loss": 3.5322,
+      "step": 3114
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.486072276003766e-06,
+      "loss": 3.6208,
+      "step": 3115
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.484329502452643e-06,
+      "loss": 3.5672,
+      "step": 3116
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.482586328049234e-06,
+      "loss": 3.5541,
+      "step": 3117
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.480842753074804e-06,
+      "loss": 3.6413,
+      "step": 3118
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.479098777810684e-06,
+      "loss": 3.6912,
+      "step": 3119
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.477354402538266e-06,
+      "loss": 3.648,
+      "step": 3120
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4756096275390115e-06,
+      "loss": 3.5541,
+      "step": 3121
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.473864453094442e-06,
+      "loss": 3.6046,
+      "step": 3122
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4721188794861465e-06,
+      "loss": 3.7377,
+      "step": 3123
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.470372906995777e-06,
+      "loss": 3.5992,
+      "step": 3124
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4686265359050515e-06,
+      "loss": 3.603,
+      "step": 3125
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.466879766495749e-06,
+      "loss": 3.6564,
+      "step": 3126
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.465132599049716e-06,
+      "loss": 3.5064,
+      "step": 3127
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4633850338488615e-06,
+      "loss": 3.5961,
+      "step": 3128
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.46163707117516e-06,
+      "loss": 3.5969,
+      "step": 3129
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.459888711310647e-06,
+      "loss": 3.5603,
+      "step": 3130
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.458139954537427e-06,
+      "loss": 3.6465,
+      "step": 3131
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.456390801137665e-06,
+      "loss": 3.633,
+      "step": 3132
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.45464125139359e-06,
+      "loss": 3.4789,
+      "step": 3133
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.452891305587497e-06,
+      "loss": 3.5848,
+      "step": 3134
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.451140964001744e-06,
+      "loss": 3.6226,
+      "step": 3135
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.449390226918754e-06,
+      "loss": 3.5928,
+      "step": 3136
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.447639094621008e-06,
+      "loss": 3.5966,
+      "step": 3137
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.4458875673910595e-06,
+      "loss": 3.5451,
+      "step": 3138
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.444135645511517e-06,
+      "loss": 3.5078,
+      "step": 3139
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.442383329265063e-06,
+      "loss": 3.5738,
+      "step": 3140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.440630618934433e-06,
+      "loss": 3.6108,
+      "step": 3141
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.438877514802433e-06,
+      "loss": 3.5936,
+      "step": 3142
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.43712401715193e-06,
+      "loss": 3.64,
+      "step": 3143
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.435370126265855e-06,
+      "loss": 3.5064,
+      "step": 3144
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.433615842427202e-06,
+      "loss": 3.5583,
+      "step": 3145
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.431861165919028e-06,
+      "loss": 3.6447,
+      "step": 3146
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.430106097024456e-06,
+      "loss": 3.5614,
+      "step": 3147
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.428350636026669e-06,
+      "loss": 3.5407,
+      "step": 3148
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.426594783208914e-06,
+      "loss": 3.5857,
+      "step": 3149
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.424838538854505e-06,
+      "loss": 3.6642,
+      "step": 3150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.423081903246814e-06,
+      "loss": 3.5272,
+      "step": 3151
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.421324876669277e-06,
+      "loss": 3.5519,
+      "step": 3152
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.419567459405396e-06,
+      "loss": 3.5637,
+      "step": 3153
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.417809651738735e-06,
+      "loss": 3.5909,
+      "step": 3154
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.416051453952917e-06,
+      "loss": 3.5622,
+      "step": 3155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.414292866331634e-06,
+      "loss": 3.5366,
+      "step": 3156
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.412533889158638e-06,
+      "loss": 3.601,
+      "step": 3157
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.410774522717743e-06,
+      "loss": 3.5637,
+      "step": 3158
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.409014767292828e-06,
+      "loss": 3.6867,
+      "step": 3159
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.407254623167833e-06,
+      "loss": 3.5538,
+      "step": 3160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.405494090626761e-06,
+      "loss": 3.4793,
+      "step": 3161
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.403733169953679e-06,
+      "loss": 3.5415,
+      "step": 3162
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.401971861432714e-06,
+      "loss": 3.5148,
+      "step": 3163
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.400210165348059e-06,
+      "loss": 3.6657,
+      "step": 3164
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3984480819839656e-06,
+      "loss": 3.6194,
+      "step": 3165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.396685611624752e-06,
+      "loss": 3.5893,
+      "step": 3166
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.394922754554795e-06,
+      "loss": 3.5382,
+      "step": 3167
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3931595110585384e-06,
+      "loss": 3.6869,
+      "step": 3168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.391395881420482e-06,
+      "loss": 3.4866,
+      "step": 3169
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.389631865925196e-06,
+      "loss": 3.5367,
+      "step": 3170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.387867464857304e-06,
+      "loss": 3.5372,
+      "step": 3171
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.386102678501498e-06,
+      "loss": 3.5495,
+      "step": 3172
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3843375071425315e-06,
+      "loss": 3.6412,
+      "step": 3173
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.382571951065216e-06,
+      "loss": 3.6722,
+      "step": 3174
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.380806010554431e-06,
+      "loss": 3.6285,
+      "step": 3175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.379039685895114e-06,
+      "loss": 3.6463,
+      "step": 3176
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.377272977372265e-06,
+      "loss": 3.6555,
+      "step": 3177
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.375505885270946e-06,
+      "loss": 3.545,
+      "step": 3178
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.373738409876283e-06,
+      "loss": 3.5193,
+      "step": 3179
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.371970551473462e-06,
+      "loss": 3.6105,
+      "step": 3180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.370202310347729e-06,
+      "loss": 3.4707,
+      "step": 3181
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3684336867843975e-06,
+      "loss": 3.4455,
+      "step": 3182
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.366664681068836e-06,
+      "loss": 3.5355,
+      "step": 3183
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.364895293486478e-06,
+      "loss": 3.6982,
+      "step": 3184
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.363125524322818e-06,
+      "loss": 3.5928,
+      "step": 3185
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.361355373863415e-06,
+      "loss": 3.5368,
+      "step": 3186
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3595848423938835e-06,
+      "loss": 3.6114,
+      "step": 3187
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.3578139301999044e-06,
+      "loss": 3.5113,
+      "step": 3188
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.356042637567218e-06,
+      "loss": 3.5928,
+      "step": 3189
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.354270964781627e-06,
+      "loss": 3.6181,
+      "step": 3190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.352498912128996e-06,
+      "loss": 3.7087,
+      "step": 3191
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.350726479895246e-06,
+      "loss": 3.5727,
+      "step": 3192
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.348953668366367e-06,
+      "loss": 3.4921,
+      "step": 3193
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.347180477828404e-06,
+      "loss": 3.5706,
+      "step": 3194
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.345406908567465e-06,
+      "loss": 3.4663,
+      "step": 3195
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.343632960869722e-06,
+      "loss": 3.6202,
+      "step": 3196
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.341858635021404e-06,
+      "loss": 3.6014,
+      "step": 3197
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.340083931308802e-06,
+      "loss": 3.5891,
+      "step": 3198
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.338308850018269e-06,
+      "loss": 3.4689,
+      "step": 3199
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.336533391436218e-06,
+      "loss": 3.6176,
+      "step": 3200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.334757555849126e-06,
+      "loss": 3.6212,
+      "step": 3201
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.332981343543524e-06,
+      "loss": 3.4991,
+      "step": 3202
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.331204754806011e-06,
+      "loss": 3.598,
+      "step": 3203
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.329427789923243e-06,
+      "loss": 3.6054,
+      "step": 3204
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.327650449181935e-06,
+      "loss": 3.5147,
+      "step": 3205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.32587273286887e-06,
+      "loss": 3.5388,
+      "step": 3206
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.324094641270882e-06,
+      "loss": 3.4413,
+      "step": 3207
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.322316174674872e-06,
+      "loss": 3.6762,
+      "step": 3208
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.320537333367801e-06,
+      "loss": 3.5173,
+      "step": 3209
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.3187581176366864e-06,
+      "loss": 3.4665,
+      "step": 3210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.31697852776861e-06,
+      "loss": 3.5267,
+      "step": 3211
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.315198564050713e-06,
+      "loss": 3.5665,
+      "step": 3212
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.313418226770196e-06,
+      "loss": 3.5877,
+      "step": 3213
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.311637516214322e-06,
+      "loss": 3.5701,
+      "step": 3214
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.30985643267041e-06,
+      "loss": 3.6825,
+      "step": 3215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.308074976425845e-06,
+      "loss": 3.5968,
+      "step": 3216
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.306293147768068e-06,
+      "loss": 3.5511,
+      "step": 3217
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.30451094698458e-06,
+      "loss": 3.6826,
+      "step": 3218
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.302728374362944e-06,
+      "loss": 3.5569,
+      "step": 3219
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.300945430190782e-06,
+      "loss": 3.6401,
+      "step": 3220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.299162114755778e-06,
+      "loss": 3.4548,
+      "step": 3221
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.297378428345671e-06,
+      "loss": 3.5436,
+      "step": 3222
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.295594371248265e-06,
+      "loss": 3.6303,
+      "step": 3223
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2938099437514196e-06,
+      "loss": 3.5183,
+      "step": 3224
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.29202514614306e-06,
+      "loss": 3.508,
+      "step": 3225
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.290239978711163e-06,
+      "loss": 3.4129,
+      "step": 3226
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2884544417437745e-06,
+      "loss": 3.6025,
+      "step": 3227
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.28666853552899e-06,
+      "loss": 3.5699,
+      "step": 3228
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.284882260354973e-06,
+      "loss": 3.5625,
+      "step": 3229
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.283095616509941e-06,
+      "loss": 3.5032,
+      "step": 3230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.281308604282173e-06,
+      "loss": 3.5708,
+      "step": 3231
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.27952122396001e-06,
+      "loss": 3.5396,
+      "step": 3232
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2777334758318475e-06,
+      "loss": 3.6052,
+      "step": 3233
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.275945360186144e-06,
+      "loss": 3.5246,
+      "step": 3234
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.274156877311416e-06,
+      "loss": 3.4384,
+      "step": 3235
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2723680274962395e-06,
+      "loss": 3.6916,
+      "step": 3236
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.270578811029248e-06,
+      "loss": 3.6103,
+      "step": 3237
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.268789228199137e-06,
+      "loss": 3.5284,
+      "step": 3238
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2669992792946595e-06,
+      "loss": 3.557,
+      "step": 3239
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2652089646046285e-06,
+      "loss": 3.4432,
+      "step": 3240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2634182844179155e-06,
+      "loss": 3.589,
+      "step": 3241
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2616272390234496e-06,
+      "loss": 3.4757,
+      "step": 3242
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.259835828710221e-06,
+      "loss": 3.4173,
+      "step": 3243
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2580440537672766e-06,
+      "loss": 3.5801,
+      "step": 3244
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.256251914483727e-06,
+      "loss": 3.6159,
+      "step": 3245
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.254459411148734e-06,
+      "loss": 3.4889,
+      "step": 3246
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.252666544051523e-06,
+      "loss": 3.681,
+      "step": 3247
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.250873313481379e-06,
+      "loss": 3.511,
+      "step": 3248
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2490797197276415e-06,
+      "loss": 3.5693,
+      "step": 3249
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.247285763079714e-06,
+      "loss": 3.5997,
+      "step": 3250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.245491443827052e-06,
+      "loss": 3.5406,
+      "step": 3251
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.243696762259176e-06,
+      "loss": 3.6117,
+      "step": 3252
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.241901718665659e-06,
+      "loss": 3.5664,
+      "step": 3253
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.240106313336139e-06,
+      "loss": 3.626,
+      "step": 3254
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.238310546560305e-06,
+      "loss": 3.6569,
+      "step": 3255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.236514418627912e-06,
+      "loss": 3.5597,
+      "step": 3256
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.234717929828767e-06,
+      "loss": 3.4527,
+      "step": 3257
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.232921080452738e-06,
+      "loss": 3.5646,
+      "step": 3258
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.231123870789751e-06,
+      "loss": 3.4974,
+      "step": 3259
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.22932630112979e-06,
+      "loss": 3.4471,
+      "step": 3260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.227528371762897e-06,
+      "loss": 3.5176,
+      "step": 3261
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.225730082979171e-06,
+      "loss": 3.6252,
+      "step": 3262
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.223931435068771e-06,
+      "loss": 3.5692,
+      "step": 3263
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.222132428321914e-06,
+      "loss": 3.634,
+      "step": 3264
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2203330630288714e-06,
+      "loss": 3.4914,
+      "step": 3265
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2185333394799775e-06,
+      "loss": 3.7337,
+      "step": 3266
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.216733257965619e-06,
+      "loss": 3.5728,
+      "step": 3267
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.214932818776246e-06,
+      "loss": 3.5175,
+      "step": 3268
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.213132022202361e-06,
+      "loss": 3.5712,
+      "step": 3269
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.21133086853453e-06,
+      "loss": 3.4929,
+      "step": 3270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.209529358063369e-06,
+      "loss": 3.5873,
+      "step": 3271
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2077274910795605e-06,
+      "loss": 3.5247,
+      "step": 3272
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.205925267873836e-06,
+      "loss": 3.5274,
+      "step": 3273
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.204122688736989e-06,
+      "loss": 3.5672,
+      "step": 3274
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.2023197539598734e-06,
+      "loss": 3.5887,
+      "step": 3275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.200516463833392e-06,
+      "loss": 3.6017,
+      "step": 3276
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.198712818648512e-06,
+      "loss": 3.561,
+      "step": 3277
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.196908818696258e-06,
+      "loss": 3.5149,
+      "step": 3278
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.195104464267705e-06,
+      "loss": 3.5074,
+      "step": 3279
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.193299755653992e-06,
+      "loss": 3.4988,
+      "step": 3280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.191494693146314e-06,
+      "loss": 3.5055,
+      "step": 3281
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.189689277035921e-06,
+      "loss": 3.4982,
+      "step": 3282
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.18788350761412e-06,
+      "loss": 3.5759,
+      "step": 3283
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.1860773851722765e-06,
+      "loss": 3.6014,
+      "step": 3284
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.184270910001814e-06,
+      "loss": 3.4929,
+      "step": 3285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.18246408239421e-06,
+      "loss": 3.6402,
+      "step": 3286
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.180656902641001e-06,
+      "loss": 3.567,
+      "step": 3287
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.178849371033779e-06,
+      "loss": 3.6044,
+      "step": 3288
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.177041487864193e-06,
+      "loss": 3.4927,
+      "step": 3289
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.175233253423951e-06,
+      "loss": 3.4973,
+      "step": 3290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.173424668004814e-06,
+      "loss": 3.3521,
+      "step": 3291
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1716157318986025e-06,
+      "loss": 3.5182,
+      "step": 3292
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1698064453971915e-06,
+      "loss": 3.5238,
+      "step": 3293
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.167996808792515e-06,
+      "loss": 3.5097,
+      "step": 3294
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1661868223765595e-06,
+      "loss": 3.6143,
+      "step": 3295
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.164376486441374e-06,
+      "loss": 3.6443,
+      "step": 3296
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1625658012790565e-06,
+      "loss": 3.5867,
+      "step": 3297
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.160754767181768e-06,
+      "loss": 3.5533,
+      "step": 3298
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.158943384441723e-06,
+      "loss": 3.3831,
+      "step": 3299
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.15713165335119e-06,
+      "loss": 3.5308,
+      "step": 3300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.155319574202499e-06,
+      "loss": 3.5486,
+      "step": 3301
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.15350714728803e-06,
+      "loss": 3.5116,
+      "step": 3302
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.151694372900225e-06,
+      "loss": 3.4511,
+      "step": 3303
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1498812513315785e-06,
+      "loss": 3.5585,
+      "step": 3304
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.14806778287464e-06,
+      "loss": 3.5291,
+      "step": 3305
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.14625396782202e-06,
+      "loss": 3.5183,
+      "step": 3306
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.144439806466379e-06,
+      "loss": 3.5519,
+      "step": 3307
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.142625299100437e-06,
+      "loss": 3.5007,
+      "step": 3308
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.140810446016971e-06,
+      "loss": 3.4554,
+      "step": 3309
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.138995247508807e-06,
+      "loss": 3.6232,
+      "step": 3310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.137179703868835e-06,
+      "loss": 3.676,
+      "step": 3311
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.135363815389998e-06,
+      "loss": 3.6971,
+      "step": 3312
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.133547582365291e-06,
+      "loss": 3.641,
+      "step": 3313
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.13173100508777e-06,
+      "loss": 3.6723,
+      "step": 3314
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.129914083850541e-06,
+      "loss": 3.4604,
+      "step": 3315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.12809681894677e-06,
+      "loss": 3.5607,
+      "step": 3316
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.126279210669677e-06,
+      "loss": 3.5554,
+      "step": 3317
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1244612593125385e-06,
+      "loss": 3.526,
+      "step": 3318
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1226429651686825e-06,
+      "loss": 3.5955,
+      "step": 3319
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.120824328531497e-06,
+      "loss": 3.6274,
+      "step": 3320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.119005349694421e-06,
+      "loss": 3.5017,
+      "step": 3321
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.1171860289509545e-06,
+      "loss": 3.5282,
+      "step": 3322
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.115366366594645e-06,
+      "loss": 3.516,
+      "step": 3323
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.113546362919102e-06,
+      "loss": 3.4773,
+      "step": 3324
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.111726018217986e-06,
+      "loss": 3.4951,
+      "step": 3325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.109905332785015e-06,
+      "loss": 3.6249,
+      "step": 3326
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.108084306913959e-06,
+      "loss": 3.5989,
+      "step": 3327
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.106262940898646e-06,
+      "loss": 3.6962,
+      "step": 3328
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.104441235032957e-06,
+      "loss": 3.529,
+      "step": 3329
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.10261918961083e-06,
+      "loss": 3.5823,
+      "step": 3330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.100796804926254e-06,
+      "loss": 3.4628,
+      "step": 3331
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.098974081273277e-06,
+      "loss": 3.6589,
+      "step": 3332
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.097151018945998e-06,
+      "loss": 3.4721,
+      "step": 3333
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0953276182385724e-06,
+      "loss": 3.4908,
+      "step": 3334
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.093503879445212e-06,
+      "loss": 3.5992,
+      "step": 3335
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.091679802860179e-06,
+      "loss": 3.6926,
+      "step": 3336
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.089855388777793e-06,
+      "loss": 3.6382,
+      "step": 3337
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.088030637492429e-06,
+      "loss": 3.5246,
+      "step": 3338
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.086205549298513e-06,
+      "loss": 3.5475,
+      "step": 3339
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.084380124490527e-06,
+      "loss": 3.4873,
+      "step": 3340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.082554363363008e-06,
+      "loss": 3.565,
+      "step": 3341
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.080728266210547e-06,
+      "loss": 3.4979,
+      "step": 3342
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.078901833327789e-06,
+      "loss": 3.5698,
+      "step": 3343
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0770750650094335e-06,
+      "loss": 3.6804,
+      "step": 3344
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.075247961550232e-06,
+      "loss": 3.5881,
+      "step": 3345
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.073420523244995e-06,
+      "loss": 3.5986,
+      "step": 3346
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.071592750388581e-06,
+      "loss": 3.5536,
+      "step": 3347
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.069764643275906e-06,
+      "loss": 3.6132,
+      "step": 3348
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.067936202201941e-06,
+      "loss": 3.6426,
+      "step": 3349
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.066107427461707e-06,
+      "loss": 3.6189,
+      "step": 3350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.064278319350282e-06,
+      "loss": 3.5772,
+      "step": 3351
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.062448878162797e-06,
+      "loss": 3.5754,
+      "step": 3352
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.060619104194436e-06,
+      "loss": 3.6142,
+      "step": 3353
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.058788997740438e-06,
+      "loss": 3.7523,
+      "step": 3354
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.056958559096095e-06,
+      "loss": 3.5083,
+      "step": 3355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0551277885567505e-06,
+      "loss": 3.5316,
+      "step": 3356
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.053296686417806e-06,
+      "loss": 3.6079,
+      "step": 3357
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.051465252974712e-06,
+      "loss": 3.5247,
+      "step": 3358
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.049633488522976e-06,
+      "loss": 3.544,
+      "step": 3359
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.047801393358156e-06,
+      "loss": 3.557,
+      "step": 3360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.045968967775867e-06,
+      "loss": 3.6087,
+      "step": 3361
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.044136212071773e-06,
+      "loss": 3.5532,
+      "step": 3362
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0423031265415946e-06,
+      "loss": 3.5592,
+      "step": 3363
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0404697114811025e-06,
+      "loss": 3.6053,
+      "step": 3364
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.038635967186125e-06,
+      "loss": 3.5729,
+      "step": 3365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.0368018939525395e-06,
+      "loss": 3.6239,
+      "step": 3366
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.034967492076279e-06,
+      "loss": 3.6667,
+      "step": 3367
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.033132761853327e-06,
+      "loss": 3.6439,
+      "step": 3368
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.031297703579722e-06,
+      "loss": 3.6556,
+      "step": 3369
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.029462317551554e-06,
+      "loss": 3.7324,
+      "step": 3370
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.02762660406497e-06,
+      "loss": 3.5288,
+      "step": 3371
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.025790563416163e-06,
+      "loss": 3.5538,
+      "step": 3372
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.0239541959013845e-06,
+      "loss": 3.5795,
+      "step": 3373
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.022117501816936e-06,
+      "loss": 3.5528,
+      "step": 3374
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.020280481459172e-06,
+      "loss": 3.581,
+      "step": 3375
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.018443135124499e-06,
+      "loss": 3.6376,
+      "step": 3376
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.01660546310938e-06,
+      "loss": 3.5824,
+      "step": 3377
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.014767465710324e-06,
+      "loss": 3.5389,
+      "step": 3378
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.0129291432238985e-06,
+      "loss": 3.5398,
+      "step": 3379
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.0110904959467216e-06,
+      "loss": 3.5224,
+      "step": 3380
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.009251524175461e-06,
+      "loss": 3.5872,
+      "step": 3381
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.00741222820684e-06,
+      "loss": 3.5324,
+      "step": 3382
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.005572608337635e-06,
+      "loss": 3.5513,
+      "step": 3383
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.003732664864671e-06,
+      "loss": 3.5633,
+      "step": 3384
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.0018923980848265e-06,
+      "loss": 3.626,
+      "step": 3385
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.000051808295034e-06,
+      "loss": 3.5623,
+      "step": 3386
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.998210895792277e-06,
+      "loss": 3.5821,
+      "step": 3387
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.996369660873591e-06,
+      "loss": 3.6135,
+      "step": 3388
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.994528103836063e-06,
+      "loss": 3.4509,
+      "step": 3389
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.992686224976832e-06,
+      "loss": 3.5188,
+      "step": 3390
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.990844024593091e-06,
+      "loss": 3.6201,
+      "step": 3391
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.989001502982081e-06,
+      "loss": 3.6714,
+      "step": 3392
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9871586604411e-06,
+      "loss": 3.6426,
+      "step": 3393
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.985315497267492e-06,
+      "loss": 3.5925,
+      "step": 3394
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.983472013758656e-06,
+      "loss": 3.6239,
+      "step": 3395
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.981628210212044e-06,
+      "loss": 3.4473,
+      "step": 3396
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.979784086925159e-06,
+      "loss": 3.5706,
+      "step": 3397
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.97793964419555e-06,
+      "loss": 3.5515,
+      "step": 3398
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.976094882320826e-06,
+      "loss": 3.5039,
+      "step": 3399
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.974249801598643e-06,
+      "loss": 3.4906,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9724044023267066e-06,
+      "loss": 3.5866,
+      "step": 3401
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.970558684802779e-06,
+      "loss": 3.5779,
+      "step": 3402
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.968712649324671e-06,
+      "loss": 3.6087,
+      "step": 3403
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.966866296190243e-06,
+      "loss": 3.5423,
+      "step": 3404
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.965019625697408e-06,
+      "loss": 3.6458,
+      "step": 3405
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.963172638144131e-06,
+      "loss": 3.4732,
+      "step": 3406
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9613253338284305e-06,
+      "loss": 3.5153,
+      "step": 3407
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.959477713048369e-06,
+      "loss": 3.6777,
+      "step": 3408
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.957629776102068e-06,
+      "loss": 3.5659,
+      "step": 3409
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9557815232876934e-06,
+      "loss": 3.5752,
+      "step": 3410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.953932954903468e-06,
+      "loss": 3.6292,
+      "step": 3411
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.952084071247659e-06,
+      "loss": 3.6478,
+      "step": 3412
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.950234872618591e-06,
+      "loss": 3.526,
+      "step": 3413
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.948385359314636e-06,
+      "loss": 3.5224,
+      "step": 3414
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.946535531634216e-06,
+      "loss": 3.6174,
+      "step": 3415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.944685389875805e-06,
+      "loss": 3.5321,
+      "step": 3416
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.942834934337928e-06,
+      "loss": 3.5514,
+      "step": 3417
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.940984165319161e-06,
+      "loss": 3.5491,
+      "step": 3418
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.939133083118131e-06,
+      "loss": 3.6526,
+      "step": 3419
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9372816880335104e-06,
+      "loss": 3.5683,
+      "step": 3420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.93542998036403e-06,
+      "loss": 3.604,
+      "step": 3421
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.933577960408465e-06,
+      "loss": 3.5137,
+      "step": 3422
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.931725628465643e-06,
+      "loss": 3.5089,
+      "step": 3423
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.929872984834444e-06,
+      "loss": 3.4775,
+      "step": 3424
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.928020029813794e-06,
+      "loss": 3.7062,
+      "step": 3425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.926166763702673e-06,
+      "loss": 3.5922,
+      "step": 3426
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.924313186800109e-06,
+      "loss": 3.4689,
+      "step": 3427
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.922459299405181e-06,
+      "loss": 3.5531,
+      "step": 3428
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.920605101817018e-06,
+      "loss": 3.4857,
+      "step": 3429
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9187505943348e-06,
+      "loss": 3.561,
+      "step": 3430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.916895777257755e-06,
+      "loss": 3.5286,
+      "step": 3431
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.915040650885161e-06,
+      "loss": 3.5228,
+      "step": 3432
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.913185215516349e-06,
+      "loss": 3.5624,
+      "step": 3433
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.911329471450697e-06,
+      "loss": 3.515,
+      "step": 3434
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.909473418987633e-06,
+      "loss": 3.5804,
+      "step": 3435
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.907617058426637e-06,
+      "loss": 3.5306,
+      "step": 3436
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.9057603900672355e-06,
+      "loss": 3.5625,
+      "step": 3437
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.903903414209007e-06,
+      "loss": 3.6249,
+      "step": 3438
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.902046131151577e-06,
+      "loss": 3.5606,
+      "step": 3439
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.900188541194626e-06,
+      "loss": 3.6437,
+      "step": 3440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.898330644637876e-06,
+      "loss": 3.5235,
+      "step": 3441
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.896472441781107e-06,
+      "loss": 3.5368,
+      "step": 3442
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.894613932924141e-06,
+      "loss": 3.4772,
+      "step": 3443
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.892755118366855e-06,
+      "loss": 3.5469,
+      "step": 3444
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.890895998409172e-06,
+      "loss": 3.6137,
+      "step": 3445
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.889036573351066e-06,
+      "loss": 3.5546,
+      "step": 3446
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.887176843492559e-06,
+      "loss": 3.5684,
+      "step": 3447
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.885316809133722e-06,
+      "loss": 3.5038,
+      "step": 3448
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.883456470574676e-06,
+      "loss": 3.5129,
+      "step": 3449
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.881595828115593e-06,
+      "loss": 3.512,
+      "step": 3450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.879734882056688e-06,
+      "loss": 3.587,
+      "step": 3451
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.877873632698233e-06,
+      "loss": 3.4819,
+      "step": 3452
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.8760120803405425e-06,
+      "loss": 3.546,
+      "step": 3453
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.874150225283983e-06,
+      "loss": 3.5139,
+      "step": 3454
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.872288067828969e-06,
+      "loss": 3.7085,
+      "step": 3455
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.870425608275963e-06,
+      "loss": 3.5583,
+      "step": 3456
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.868562846925479e-06,
+      "loss": 3.6002,
+      "step": 3457
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.866699784078078e-06,
+      "loss": 3.5335,
+      "step": 3458
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.864836420034366e-06,
+      "loss": 3.6328,
+      "step": 3459
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8629727550950055e-06,
+      "loss": 3.6007,
+      "step": 3460
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8611087895607e-06,
+      "loss": 3.6295,
+      "step": 3461
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.859244523732206e-06,
+      "loss": 3.6626,
+      "step": 3462
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8573799579103264e-06,
+      "loss": 3.5788,
+      "step": 3463
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.855515092395915e-06,
+      "loss": 3.6572,
+      "step": 3464
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.853649927489869e-06,
+      "loss": 3.572,
+      "step": 3465
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8517844634931395e-06,
+      "loss": 3.5519,
+      "step": 3466
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8499187007067235e-06,
+      "loss": 3.5878,
+      "step": 3467
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.848052639431664e-06,
+      "loss": 3.4924,
+      "step": 3468
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.846186279969056e-06,
+      "loss": 3.5501,
+      "step": 3469
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.844319622620039e-06,
+      "loss": 3.5871,
+      "step": 3470
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.842452667685806e-06,
+      "loss": 3.6404,
+      "step": 3471
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.840585415467591e-06,
+      "loss": 3.5116,
+      "step": 3472
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.838717866266682e-06,
+      "loss": 3.5854,
+      "step": 3473
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.836850020384412e-06,
+      "loss": 3.5498,
+      "step": 3474
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8349818781221586e-06,
+      "loss": 3.6361,
+      "step": 3475
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.833113439781356e-06,
+      "loss": 3.4975,
+      "step": 3476
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.831244705663478e-06,
+      "loss": 3.6119,
+      "step": 3477
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.829375676070049e-06,
+      "loss": 3.5908,
+      "step": 3478
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.827506351302644e-06,
+      "loss": 3.574,
+      "step": 3479
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.825636731662879e-06,
+      "loss": 3.6047,
+      "step": 3480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.823766817452425e-06,
+      "loss": 3.4939,
+      "step": 3481
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8218966089729955e-06,
+      "loss": 3.5522,
+      "step": 3482
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.820026106526352e-06,
+      "loss": 3.5057,
+      "step": 3483
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.818155310414305e-06,
+      "loss": 3.5685,
+      "step": 3484
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.816284220938713e-06,
+      "loss": 3.5853,
+      "step": 3485
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.814412838401478e-06,
+      "loss": 3.5625,
+      "step": 3486
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.812541163104554e-06,
+      "loss": 3.6127,
+      "step": 3487
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.810669195349941e-06,
+      "loss": 3.5775,
+      "step": 3488
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.8087969354396824e-06,
+      "loss": 3.6271,
+      "step": 3489
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.806924383675872e-06,
+      "loss": 3.6316,
+      "step": 3490
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.805051540360655e-06,
+      "loss": 3.4906,
+      "step": 3491
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.803178405796213e-06,
+      "loss": 3.5512,
+      "step": 3492
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.801304980284784e-06,
+      "loss": 3.5462,
+      "step": 3493
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.799431264128648e-06,
+      "loss": 3.624,
+      "step": 3494
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.797557257630133e-06,
+      "loss": 3.6734,
+      "step": 3495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.795682961091616e-06,
+      "loss": 3.5432,
+      "step": 3496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.793808374815518e-06,
+      "loss": 3.436,
+      "step": 3497
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7919334991043085e-06,
+      "loss": 3.62,
+      "step": 3498
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.790058334260502e-06,
+      "loss": 3.5576,
+      "step": 3499
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.788182880586662e-06,
+      "loss": 3.6095,
+      "step": 3500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.786307138385397e-06,
+      "loss": 3.6017,
+      "step": 3501
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.78443110795936e-06,
+      "loss": 3.6878,
+      "step": 3502
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.782554789611256e-06,
+      "loss": 3.6172,
+      "step": 3503
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.780678183643832e-06,
+      "loss": 3.4784,
+      "step": 3504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.778801290359882e-06,
+      "loss": 3.5015,
+      "step": 3505
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.776924110062248e-06,
+      "loss": 3.6,
+      "step": 3506
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.775046643053817e-06,
+      "loss": 3.6702,
+      "step": 3507
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7731688896375235e-06,
+      "loss": 3.5857,
+      "step": 3508
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.771290850116345e-06,
+      "loss": 3.5984,
+      "step": 3509
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.769412524793311e-06,
+      "loss": 3.5975,
+      "step": 3510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7675339139714906e-06,
+      "loss": 3.5014,
+      "step": 3511
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.765655017954002e-06,
+      "loss": 3.6595,
+      "step": 3512
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.763775837044012e-06,
+      "loss": 3.5387,
+      "step": 3513
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.761896371544729e-06,
+      "loss": 3.4612,
+      "step": 3514
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.760016621759409e-06,
+      "loss": 3.5463,
+      "step": 3515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.758136587991354e-06,
+      "loss": 3.4757,
+      "step": 3516
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.756256270543912e-06,
+      "loss": 3.5625,
+      "step": 3517
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.754375669720477e-06,
+      "loss": 3.5582,
+      "step": 3518
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.752494785824487e-06,
+      "loss": 3.4929,
+      "step": 3519
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.750613619159428e-06,
+      "loss": 3.5862,
+      "step": 3520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.748732170028831e-06,
+      "loss": 3.5415,
+      "step": 3521
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.746850438736273e-06,
+      "loss": 3.5904,
+      "step": 3522
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.744968425585372e-06,
+      "loss": 3.5216,
+      "step": 3523
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.743086130879799e-06,
+      "loss": 3.4864,
+      "step": 3524
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.741203554923263e-06,
+      "loss": 3.5033,
+      "step": 3525
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7393206980195244e-06,
+      "loss": 3.5512,
+      "step": 3526
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.737437560472386e-06,
+      "loss": 3.5458,
+      "step": 3527
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.735554142585697e-06,
+      "loss": 3.5735,
+      "step": 3528
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.733670444663351e-06,
+      "loss": 3.468,
+      "step": 3529
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.731786467009286e-06,
+      "loss": 3.4859,
+      "step": 3530
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.729902209927485e-06,
+      "loss": 3.5553,
+      "step": 3531
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.728017673721982e-06,
+      "loss": 3.5005,
+      "step": 3532
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.726132858696847e-06,
+      "loss": 3.5954,
+      "step": 3533
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7242477651562e-06,
+      "loss": 3.4923,
+      "step": 3534
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.722362393404205e-06,
+      "loss": 3.5382,
+      "step": 3535
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.7204767437450725e-06,
+      "loss": 3.5443,
+      "step": 3536
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.718590816483056e-06,
+      "loss": 3.6618,
+      "step": 3537
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.716704611922453e-06,
+      "loss": 3.4225,
+      "step": 3538
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.714818130367609e-06,
+      "loss": 3.6083,
+      "step": 3539
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.712931372122911e-06,
+      "loss": 3.5688,
+      "step": 3540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.711044337492791e-06,
+      "loss": 3.6509,
+      "step": 3541
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.709157026781727e-06,
+      "loss": 3.656,
+      "step": 3542
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.707269440294241e-06,
+      "loss": 3.5738,
+      "step": 3543
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.7053815783349e-06,
+      "loss": 3.6124,
+      "step": 3544
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.703493441208313e-06,
+      "loss": 3.6129,
+      "step": 3545
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.7016050292191374e-06,
+      "loss": 3.6456,
+      "step": 3546
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.699716342672071e-06,
+      "loss": 3.3895,
+      "step": 3547
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.69782738187186e-06,
+      "loss": 3.574,
+      "step": 3548
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.695938147123291e-06,
+      "loss": 3.5791,
+      "step": 3549
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.694048638731196e-06,
+      "loss": 3.5388,
+      "step": 3550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.692158857000453e-06,
+      "loss": 3.5667,
+      "step": 3551
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.690268802235982e-06,
+      "loss": 3.5497,
+      "step": 3552
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.688378474742747e-06,
+      "loss": 3.5653,
+      "step": 3553
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.686487874825758e-06,
+      "loss": 3.6394,
+      "step": 3554
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.684597002790066e-06,
+      "loss": 3.4969,
+      "step": 3555
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.682705858940769e-06,
+      "loss": 3.5718,
+      "step": 3556
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.680814443583007e-06,
+      "loss": 3.5712,
+      "step": 3557
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6789227570219635e-06,
+      "loss": 3.6028,
+      "step": 3558
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.677030799562868e-06,
+      "loss": 3.6046,
+      "step": 3559
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6751385715109905e-06,
+      "loss": 3.5788,
+      "step": 3560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6732460731716475e-06,
+      "loss": 3.6425,
+      "step": 3561
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.671353304850197e-06,
+      "loss": 3.5987,
+      "step": 3562
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.669460266852042e-06,
+      "loss": 3.4587,
+      "step": 3563
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.66756695948263e-06,
+      "loss": 3.5227,
+      "step": 3564
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6656733830474475e-06,
+      "loss": 3.5902,
+      "step": 3565
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.66377953785203e-06,
+      "loss": 3.5046,
+      "step": 3566
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.661885424201952e-06,
+      "loss": 3.6485,
+      "step": 3567
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6599910424028346e-06,
+      "loss": 3.6419,
+      "step": 3568
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.65809639276034e-06,
+      "loss": 3.671,
+      "step": 3569
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.656201475580174e-06,
+      "loss": 3.525,
+      "step": 3570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.654306291168087e-06,
+      "loss": 3.4373,
+      "step": 3571
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.65241083982987e-06,
+      "loss": 3.5864,
+      "step": 3572
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.650515121871358e-06,
+      "loss": 3.5547,
+      "step": 3573
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.648619137598431e-06,
+      "loss": 3.5929,
+      "step": 3574
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.64672288731701e-06,
+      "loss": 3.5485,
+      "step": 3575
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.64482637133306e-06,
+      "loss": 3.56,
+      "step": 3576
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.642929589952587e-06,
+      "loss": 3.534,
+      "step": 3577
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.64103254348164e-06,
+      "loss": 3.5234,
+      "step": 3578
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.639135232226315e-06,
+      "loss": 3.5364,
+      "step": 3579
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.637237656492745e-06,
+      "loss": 3.5611,
+      "step": 3580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.635339816587109e-06,
+      "loss": 3.6402,
+      "step": 3581
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.633441712815628e-06,
+      "loss": 3.5625,
+      "step": 3582
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6315433454845655e-06,
+      "loss": 3.5148,
+      "step": 3583
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.629644714900228e-06,
+      "loss": 3.5722,
+      "step": 3584
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.627745821368962e-06,
+      "loss": 3.5232,
+      "step": 3585
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.625846665197158e-06,
+      "loss": 3.5986,
+      "step": 3586
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.623947246691253e-06,
+      "loss": 3.4657,
+      "step": 3587
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.62204756615772e-06,
+      "loss": 3.5999,
+      "step": 3588
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.620147623903076e-06,
+      "loss": 3.5208,
+      "step": 3589
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.618247420233883e-06,
+      "loss": 3.565,
+      "step": 3590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.616346955456743e-06,
+      "loss": 3.5691,
+      "step": 3591
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.614446229878297e-06,
+      "loss": 3.6363,
+      "step": 3592
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6125452438052375e-06,
+      "loss": 3.542,
+      "step": 3593
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.610643997544288e-06,
+      "loss": 3.4837,
+      "step": 3594
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.608742491402221e-06,
+      "loss": 3.5238,
+      "step": 3595
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.606840725685849e-06,
+      "loss": 3.5519,
+      "step": 3596
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6049387007020246e-06,
+      "loss": 3.5985,
+      "step": 3597
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.6030364167576465e-06,
+      "loss": 3.5763,
+      "step": 3598
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.60113387415965e-06,
+      "loss": 3.5885,
+      "step": 3599
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.599231073215017e-06,
+      "loss": 3.6573,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.597328014230766e-06,
+      "loss": 3.6547,
+      "step": 3601
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.595424697513963e-06,
+      "loss": 3.6188,
+      "step": 3602
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.593521123371711e-06,
+      "loss": 3.4945,
+      "step": 3603
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.591617292111156e-06,
+      "loss": 3.4948,
+      "step": 3604
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.589713204039488e-06,
+      "loss": 3.6211,
+      "step": 3605
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.587808859463931e-06,
+      "loss": 3.5774,
+      "step": 3606
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.585904258691759e-06,
+      "loss": 3.5768,
+      "step": 3607
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.583999402030283e-06,
+      "loss": 3.4783,
+      "step": 3608
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.582094289786856e-06,
+      "loss": 3.5875,
+      "step": 3609
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.580188922268873e-06,
+      "loss": 3.6837,
+      "step": 3610
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.578283299783768e-06,
+      "loss": 3.5011,
+      "step": 3611
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.576377422639018e-06,
+      "loss": 3.5011,
+      "step": 3612
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.574471291142143e-06,
+      "loss": 3.6073,
+      "step": 3613
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.572564905600698e-06,
+      "loss": 3.5033,
+      "step": 3614
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.570658266322283e-06,
+      "loss": 3.6281,
+      "step": 3615
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.568751373614541e-06,
+      "loss": 3.5179,
+      "step": 3616
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.566844227785153e-06,
+      "loss": 3.4714,
+      "step": 3617
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.56493682914184e-06,
+      "loss": 3.5902,
+      "step": 3618
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.563029177992366e-06,
+      "loss": 3.6278,
+      "step": 3619
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.561121274644537e-06,
+      "loss": 3.5939,
+      "step": 3620
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.559213119406193e-06,
+      "loss": 3.5635,
+      "step": 3621
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.557304712585222e-06,
+      "loss": 3.5892,
+      "step": 3622
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.55539605448955e-06,
+      "loss": 3.5694,
+      "step": 3623
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.553487145427143e-06,
+      "loss": 3.5455,
+      "step": 3624
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.551577985706008e-06,
+      "loss": 3.5993,
+      "step": 3625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.549668575634192e-06,
+      "loss": 3.4227,
+      "step": 3626
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.547758915519784e-06,
+      "loss": 3.5955,
+      "step": 3627
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.545849005670909e-06,
+      "loss": 3.622,
+      "step": 3628
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.543938846395739e-06,
+      "loss": 3.5734,
+      "step": 3629
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.542028438002479e-06,
+      "loss": 3.4933,
+      "step": 3630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.540117780799381e-06,
+      "loss": 3.6181,
+      "step": 3631
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.538206875094734e-06,
+      "loss": 3.6246,
+      "step": 3632
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.536295721196864e-06,
+      "loss": 3.4978,
+      "step": 3633
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.534384319414141e-06,
+      "loss": 3.5571,
+      "step": 3634
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.532472670054975e-06,
+      "loss": 3.5647,
+      "step": 3635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.530560773427816e-06,
+      "loss": 3.5171,
+      "step": 3636
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.52864862984115e-06,
+      "loss": 3.6419,
+      "step": 3637
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.526736239603508e-06,
+      "loss": 3.5488,
+      "step": 3638
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.52482360302346e-06,
+      "loss": 3.5164,
+      "step": 3639
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.522910720409611e-06,
+      "loss": 3.5371,
+      "step": 3640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.52099759207061e-06,
+      "loss": 3.6142,
+      "step": 3641
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.519084218315146e-06,
+      "loss": 3.5193,
+      "step": 3642
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.517170599451944e-06,
+      "loss": 3.6408,
+      "step": 3643
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.515256735789773e-06,
+      "loss": 3.5538,
+      "step": 3644
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.5133426276374386e-06,
+      "loss": 3.4803,
+      "step": 3645
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.511428275303786e-06,
+      "loss": 3.6588,
+      "step": 3646
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.509513679097702e-06,
+      "loss": 3.4972,
+      "step": 3647
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.507598839328108e-06,
+      "loss": 3.5621,
+      "step": 3648
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.505683756303971e-06,
+      "loss": 3.4688,
+      "step": 3649
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.503768430334294e-06,
+      "loss": 3.5174,
+      "step": 3650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.501852861728116e-06,
+      "loss": 3.6311,
+      "step": 3651
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.499937050794522e-06,
+      "loss": 3.5221,
+      "step": 3652
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.498020997842629e-06,
+      "loss": 3.6273,
+      "step": 3653
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.496104703181599e-06,
+      "loss": 3.5595,
+      "step": 3654
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.49418816712063e-06,
+      "loss": 3.5601,
+      "step": 3655
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.492271389968961e-06,
+      "loss": 3.4966,
+      "step": 3656
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.490354372035865e-06,
+      "loss": 3.5324,
+      "step": 3657
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.48843711363066e-06,
+      "loss": 3.4979,
+      "step": 3658
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.486519615062698e-06,
+      "loss": 3.5389,
+      "step": 3659
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.484601876641375e-06,
+      "loss": 3.4943,
+      "step": 3660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.48268389867612e-06,
+      "loss": 3.5956,
+      "step": 3661
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4807656814764046e-06,
+      "loss": 3.5863,
+      "step": 3662
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.478847225351734e-06,
+      "loss": 3.5008,
+      "step": 3663
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.47692853061166e-06,
+      "loss": 3.58,
+      "step": 3664
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.475009597565767e-06,
+      "loss": 3.4949,
+      "step": 3665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.473090426523679e-06,
+      "loss": 3.5993,
+      "step": 3666
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.471171017795057e-06,
+      "loss": 3.5499,
+      "step": 3667
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.469251371689606e-06,
+      "loss": 3.4159,
+      "step": 3668
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.467331488517061e-06,
+      "loss": 3.5421,
+      "step": 3669
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.465411368587202e-06,
+      "loss": 3.5689,
+      "step": 3670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.463491012209844e-06,
+      "loss": 3.5757,
+      "step": 3671
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.461570419694841e-06,
+      "loss": 3.6332,
+      "step": 3672
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.459649591352086e-06,
+      "loss": 3.5596,
+      "step": 3673
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.457728527491508e-06,
+      "loss": 3.6455,
+      "step": 3674
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.455807228423076e-06,
+      "loss": 3.3891,
+      "step": 3675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.453885694456794e-06,
+      "loss": 3.5918,
+      "step": 3676
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.451963925902708e-06,
+      "loss": 3.5176,
+      "step": 3677
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.450041923070897e-06,
+      "loss": 3.5362,
+      "step": 3678
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.448119686271486e-06,
+      "loss": 3.6959,
+      "step": 3679
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.446197215814627e-06,
+      "loss": 3.568,
+      "step": 3680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.444274512010518e-06,
+      "loss": 3.5252,
+      "step": 3681
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.44235157516939e-06,
+      "loss": 3.5451,
+      "step": 3682
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.440428405601514e-06,
+      "loss": 3.45,
+      "step": 3683
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.438505003617199e-06,
+      "loss": 3.5887,
+      "step": 3684
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.436581369526788e-06,
+      "loss": 3.5457,
+      "step": 3685
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.434657503640667e-06,
+      "loss": 3.5459,
+      "step": 3686
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.432733406269251e-06,
+      "loss": 3.5463,
+      "step": 3687
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.430809077723004e-06,
+      "loss": 3.5757,
+      "step": 3688
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.428884518312415e-06,
+      "loss": 3.4597,
+      "step": 3689
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.426959728348019e-06,
+      "loss": 3.4747,
+      "step": 3690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.425034708140385e-06,
+      "loss": 3.5346,
+      "step": 3691
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.423109458000121e-06,
+      "loss": 3.4738,
+      "step": 3692
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.421183978237867e-06,
+      "loss": 3.5955,
+      "step": 3693
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.419258269164306e-06,
+      "loss": 3.5457,
+      "step": 3694
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.417332331090156e-06,
+      "loss": 3.5825,
+      "step": 3695
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.41540616432617e-06,
+      "loss": 3.5726,
+      "step": 3696
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.41347976918314e-06,
+      "loss": 3.4214,
+      "step": 3697
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4115531459718936e-06,
+      "loss": 3.4801,
+      "step": 3698
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.409626295003298e-06,
+      "loss": 3.5319,
+      "step": 3699
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4076992165882536e-06,
+      "loss": 3.6075,
+      "step": 3700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.405771911037698e-06,
+      "loss": 3.5639,
+      "step": 3701
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.4038443786626096e-06,
+      "loss": 3.6303,
+      "step": 3702
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.401916619773996e-06,
+      "loss": 3.559,
+      "step": 3703
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.399988634682909e-06,
+      "loss": 3.5498,
+      "step": 3704
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.398060423700431e-06,
+      "loss": 3.6692,
+      "step": 3705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.396131987137685e-06,
+      "loss": 3.4902,
+      "step": 3706
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.394203325305827e-06,
+      "loss": 3.4541,
+      "step": 3707
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.392274438516053e-06,
+      "loss": 3.4211,
+      "step": 3708
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.390345327079591e-06,
+      "loss": 3.514,
+      "step": 3709
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.388415991307711e-06,
+      "loss": 3.6187,
+      "step": 3710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3864864315117114e-06,
+      "loss": 3.6649,
+      "step": 3711
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.384556648002934e-06,
+      "loss": 3.563,
+      "step": 3712
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.382626641092753e-06,
+      "loss": 3.5274,
+      "step": 3713
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.38069641109258e-06,
+      "loss": 3.7145,
+      "step": 3714
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.378765958313863e-06,
+      "loss": 3.602,
+      "step": 3715
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.376835283068083e-06,
+      "loss": 3.4593,
+      "step": 3716
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.374904385666758e-06,
+      "loss": 3.529,
+      "step": 3717
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.372973266421446e-06,
+      "loss": 3.5138,
+      "step": 3718
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.371041925643735e-06,
+      "loss": 3.6029,
+      "step": 3719
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.369110363645252e-06,
+      "loss": 3.5503,
+      "step": 3720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.367178580737661e-06,
+      "loss": 3.5133,
+      "step": 3721
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.365246577232656e-06,
+      "loss": 3.4795,
+      "step": 3722
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.363314353441974e-06,
+      "loss": 3.5135,
+      "step": 3723
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.36138190967738e-06,
+      "loss": 3.6159,
+      "step": 3724
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.35944924625068e-06,
+      "loss": 3.587,
+      "step": 3725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.357516363473713e-06,
+      "loss": 3.617,
+      "step": 3726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.355583261658357e-06,
+      "loss": 3.544,
+      "step": 3727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.353649941116519e-06,
+      "loss": 3.5451,
+      "step": 3728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.351716402160145e-06,
+      "loss": 3.6134,
+      "step": 3729
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.349782645101218e-06,
+      "loss": 3.5058,
+      "step": 3730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.347848670251753e-06,
+      "loss": 3.5104,
+      "step": 3731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3459144779238e-06,
+      "loss": 3.5718,
+      "step": 3732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.343980068429448e-06,
+      "loss": 3.4675,
+      "step": 3733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.342045442080818e-06,
+      "loss": 3.6393,
+      "step": 3734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.340110599190064e-06,
+      "loss": 3.5382,
+      "step": 3735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.338175540069379e-06,
+      "loss": 3.6053,
+      "step": 3736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.33624026503099e-06,
+      "loss": 3.5626,
+      "step": 3737
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.334304774387157e-06,
+      "loss": 3.4707,
+      "step": 3738
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.332369068450175e-06,
+      "loss": 3.4782,
+      "step": 3739
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.330433147532377e-06,
+      "loss": 3.5351,
+      "step": 3740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.328497011946127e-06,
+      "loss": 3.5291,
+      "step": 3741
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3265606620038256e-06,
+      "loss": 3.526,
+      "step": 3742
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.324624098017905e-06,
+      "loss": 3.6237,
+      "step": 3743
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3226873203008375e-06,
+      "loss": 3.6253,
+      "step": 3744
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.320750329165124e-06,
+      "loss": 3.5565,
+      "step": 3745
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.318813124923303e-06,
+      "loss": 3.5636,
+      "step": 3746
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3168757078879486e-06,
+      "loss": 3.5253,
+      "step": 3747
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.314938078371666e-06,
+      "loss": 3.5635,
+      "step": 3748
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.313000236687097e-06,
+      "loss": 3.5384,
+      "step": 3749
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.3110621831469155e-06,
+      "loss": 3.5243,
+      "step": 3750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.309123918063832e-06,
+      "loss": 3.6184,
+      "step": 3751
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.307185441750589e-06,
+      "loss": 3.6111,
+      "step": 3752
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.305246754519967e-06,
+      "loss": 3.5773,
+      "step": 3753
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.303307856684774e-06,
+      "loss": 3.5388,
+      "step": 3754
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.301368748557857e-06,
+      "loss": 3.5433,
+      "step": 3755
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.299429430452097e-06,
+      "loss": 3.5709,
+      "step": 3756
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.297489902680405e-06,
+      "loss": 3.5147,
+      "step": 3757
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.29555016555573e-06,
+      "loss": 3.5863,
+      "step": 3758
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.293610219391053e-06,
+      "loss": 3.5441,
+      "step": 3759
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.291670064499387e-06,
+      "loss": 3.6003,
+      "step": 3760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.289729701193783e-06,
+      "loss": 3.4813,
+      "step": 3761
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2877891297873205e-06,
+      "loss": 3.5563,
+      "step": 3762
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.285848350593118e-06,
+      "loss": 3.5731,
+      "step": 3763
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.283907363924322e-06,
+      "loss": 3.6191,
+      "step": 3764
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.281966170094116e-06,
+      "loss": 3.5533,
+      "step": 3765
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2800247694157165e-06,
+      "loss": 3.5466,
+      "step": 3766
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.278083162202374e-06,
+      "loss": 3.5833,
+      "step": 3767
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.276141348767369e-06,
+      "loss": 3.4973,
+      "step": 3768
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.27419932942402e-06,
+      "loss": 3.544,
+      "step": 3769
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2722571044856755e-06,
+      "loss": 3.5464,
+      "step": 3770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.270314674265716e-06,
+      "loss": 3.4627,
+      "step": 3771
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.268372039077559e-06,
+      "loss": 3.5293,
+      "step": 3772
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.266429199234652e-06,
+      "loss": 3.4811,
+      "step": 3773
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2644861550504795e-06,
+      "loss": 3.6036,
+      "step": 3774
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2625429068385535e-06,
+      "loss": 3.5597,
+      "step": 3775
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.260599454912422e-06,
+      "loss": 3.5998,
+      "step": 3776
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.258655799585666e-06,
+      "loss": 3.5558,
+      "step": 3777
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.256711941171898e-06,
+      "loss": 3.5739,
+      "step": 3778
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.254767879984763e-06,
+      "loss": 3.6289,
+      "step": 3779
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.252823616337943e-06,
+      "loss": 3.4632,
+      "step": 3780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.2508791505451476e-06,
+      "loss": 3.5958,
+      "step": 3781
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.248934482920119e-06,
+      "loss": 3.4303,
+      "step": 3782
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.246989613776637e-06,
+      "loss": 3.6203,
+      "step": 3783
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.24504454342851e-06,
+      "loss": 3.6352,
+      "step": 3784
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.243099272189578e-06,
+      "loss": 3.5804,
+      "step": 3785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.241153800373716e-06,
+      "loss": 3.5979,
+      "step": 3786
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.23920812829483e-06,
+      "loss": 3.5411,
+      "step": 3787
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.237262256266859e-06,
+      "loss": 3.526,
+      "step": 3788
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.235316184603775e-06,
+      "loss": 3.4683,
+      "step": 3789
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.233369913619579e-06,
+      "loss": 3.4838,
+      "step": 3790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.231423443628307e-06,
+      "loss": 3.6017,
+      "step": 3791
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.229476774944028e-06,
+      "loss": 3.5373,
+      "step": 3792
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.227529907880839e-06,
+      "loss": 3.5373,
+      "step": 3793
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.225582842752873e-06,
+      "loss": 3.6463,
+      "step": 3794
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.223635579874294e-06,
+      "loss": 3.5069,
+      "step": 3795
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.221688119559297e-06,
+      "loss": 3.4671,
+      "step": 3796
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.219740462122107e-06,
+      "loss": 3.4332,
+      "step": 3797
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.217792607876986e-06,
+      "loss": 3.4671,
+      "step": 3798
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.215844557138222e-06,
+      "loss": 3.6083,
+      "step": 3799
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.21389631022014e-06,
+      "loss": 3.6226,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.211947867437093e-06,
+      "loss": 3.5461,
+      "step": 3801
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.209999229103466e-06,
+      "loss": 3.5438,
+      "step": 3802
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.208050395533678e-06,
+      "loss": 3.5951,
+      "step": 3803
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.2061013670421765e-06,
+      "loss": 3.5267,
+      "step": 3804
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.204152143943442e-06,
+      "loss": 3.5689,
+      "step": 3805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.2022027265519856e-06,
+      "loss": 3.5903,
+      "step": 3806
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.200253115182352e-06,
+      "loss": 3.5876,
+      "step": 3807
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.198303310149114e-06,
+      "loss": 3.5925,
+      "step": 3808
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1963533117668785e-06,
+      "loss": 3.6007,
+      "step": 3809
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.194403120350279e-06,
+      "loss": 3.542,
+      "step": 3810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.192452736213988e-06,
+      "loss": 3.4994,
+      "step": 3811
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.190502159672701e-06,
+      "loss": 3.5316,
+      "step": 3812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.18855139104115e-06,
+      "loss": 3.464,
+      "step": 3813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.186600430634095e-06,
+      "loss": 3.6248,
+      "step": 3814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.184649278766328e-06,
+      "loss": 3.6857,
+      "step": 3815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.182697935752672e-06,
+      "loss": 3.6869,
+      "step": 3816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1807464019079795e-06,
+      "loss": 3.5913,
+      "step": 3817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.178794677547138e-06,
+      "loss": 3.5382,
+      "step": 3818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1768427629850584e-06,
+      "loss": 3.5938,
+      "step": 3819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.174890658536691e-06,
+      "loss": 3.4842,
+      "step": 3820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.172938364517009e-06,
+      "loss": 3.5695,
+      "step": 3821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.170985881241021e-06,
+      "loss": 3.5145,
+      "step": 3822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.169033209023766e-06,
+      "loss": 3.5717,
+      "step": 3823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.16708034818031e-06,
+      "loss": 3.6355,
+      "step": 3824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.165127299025753e-06,
+      "loss": 3.6129,
+      "step": 3825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.163174061875223e-06,
+      "loss": 3.5092,
+      "step": 3826
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.161220637043879e-06,
+      "loss": 3.6386,
+      "step": 3827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.159267024846911e-06,
+      "loss": 3.5149,
+      "step": 3828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.15731322559954e-06,
+      "loss": 3.5141,
+      "step": 3829
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.155359239617015e-06,
+      "loss": 3.5406,
+      "step": 3830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.153405067214616e-06,
+      "loss": 3.472,
+      "step": 3831
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.151450708707654e-06,
+      "loss": 3.5457,
+      "step": 3832
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1494961644114685e-06,
+      "loss": 3.5609,
+      "step": 3833
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1475414346414294e-06,
+      "loss": 3.5163,
+      "step": 3834
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1455865197129375e-06,
+      "loss": 3.4762,
+      "step": 3835
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1436314199414226e-06,
+      "loss": 3.5275,
+      "step": 3836
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.141676135642345e-06,
+      "loss": 3.6368,
+      "step": 3837
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.139720667131192e-06,
+      "loss": 3.5825,
+      "step": 3838
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1377650147234875e-06,
+      "loss": 3.6669,
+      "step": 3839
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.135809178734777e-06,
+      "loss": 3.5603,
+      "step": 3840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.13385315948064e-06,
+      "loss": 3.5703,
+      "step": 3841
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.131896957276685e-06,
+      "loss": 3.5694,
+      "step": 3842
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1299405724385485e-06,
+      "loss": 3.5723,
+      "step": 3843
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.127984005281899e-06,
+      "loss": 3.4956,
+      "step": 3844
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1260272561224344e-06,
+      "loss": 3.5945,
+      "step": 3845
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.124070325275877e-06,
+      "loss": 3.5638,
+      "step": 3846
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.122113213057985e-06,
+      "loss": 3.539,
+      "step": 3847
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.120155919784542e-06,
+      "loss": 3.5463,
+      "step": 3848
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.118198445771364e-06,
+      "loss": 3.5769,
+      "step": 3849
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.11624079133429e-06,
+      "loss": 3.5629,
+      "step": 3850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.114282956789195e-06,
+      "loss": 3.5002,
+      "step": 3851
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.112324942451977e-06,
+      "loss": 3.5701,
+      "step": 3852
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.11036674863857e-06,
+      "loss": 3.4838,
+      "step": 3853
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.10840837566493e-06,
+      "loss": 3.5938,
+      "step": 3854
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.106449823847047e-06,
+      "loss": 3.5751,
+      "step": 3855
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.104491093500937e-06,
+      "loss": 3.573,
+      "step": 3856
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.1025321849426455e-06,
+      "loss": 3.4456,
+      "step": 3857
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.100573098488247e-06,
+      "loss": 3.6661,
+      "step": 3858
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.098613834453845e-06,
+      "loss": 3.5786,
+      "step": 3859
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.096654393155571e-06,
+      "loss": 3.544,
+      "step": 3860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.094694774909586e-06,
+      "loss": 3.6939,
+      "step": 3861
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.092734980032077e-06,
+      "loss": 3.4532,
+      "step": 3862
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.0907750088392624e-06,
+      "loss": 3.5118,
+      "step": 3863
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.0888148616473895e-06,
+      "loss": 3.6089,
+      "step": 3864
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.0868545387727305e-06,
+      "loss": 3.4947,
+      "step": 3865
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.084894040531591e-06,
+      "loss": 3.4844,
+      "step": 3866
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.0829333672402976e-06,
+      "loss": 3.4735,
+      "step": 3867
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.080972519215214e-06,
+      "loss": 3.7179,
+      "step": 3868
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.0790114967727235e-06,
+      "loss": 3.5275,
+      "step": 3869
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.077050300229244e-06,
+      "loss": 3.6379,
+      "step": 3870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.075088929901218e-06,
+      "loss": 3.5295,
+      "step": 3871
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.073127386105118e-06,
+      "loss": 3.4438,
+      "step": 3872
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.071165669157442e-06,
+      "loss": 3.5504,
+      "step": 3873
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0692037793747194e-06,
+      "loss": 3.5305,
+      "step": 3874
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.067241717073505e-06,
+      "loss": 3.5974,
+      "step": 3875
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.065279482570382e-06,
+      "loss": 3.6092,
+      "step": 3876
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.063317076181962e-06,
+      "loss": 3.5072,
+      "step": 3877
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.061354498224882e-06,
+      "loss": 3.5549,
+      "step": 3878
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.05939174901581e-06,
+      "loss": 3.5233,
+      "step": 3879
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.05742882887144e-06,
+      "loss": 3.5989,
+      "step": 3880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.055465738108494e-06,
+      "loss": 3.5477,
+      "step": 3881
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.05350247704372e-06,
+      "loss": 3.5677,
+      "step": 3882
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.051539045993896e-06,
+      "loss": 3.5066,
+      "step": 3883
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.049575445275825e-06,
+      "loss": 3.4367,
+      "step": 3884
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.047611675206339e-06,
+      "loss": 3.5349,
+      "step": 3885
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.045647736102298e-06,
+      "loss": 3.5259,
+      "step": 3886
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.043683628280587e-06,
+      "loss": 3.5968,
+      "step": 3887
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0417193520581175e-06,
+      "loss": 3.5415,
+      "step": 3888
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0397549077518326e-06,
+      "loss": 3.6712,
+      "step": 3889
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0377902956787e-06,
+      "loss": 3.561,
+      "step": 3890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.035825516155712e-06,
+      "loss": 3.5761,
+      "step": 3891
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.033860569499892e-06,
+      "loss": 3.5443,
+      "step": 3892
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.031895456028289e-06,
+      "loss": 3.5539,
+      "step": 3893
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0299301760579776e-06,
+      "loss": 3.5331,
+      "step": 3894
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.027964729906061e-06,
+      "loss": 3.5469,
+      "step": 3895
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.025999117889667e-06,
+      "loss": 3.5301,
+      "step": 3896
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.024033340325954e-06,
+      "loss": 3.5444,
+      "step": 3897
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.022067397532103e-06,
+      "loss": 3.5184,
+      "step": 3898
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0201012898253244e-06,
+      "loss": 3.724,
+      "step": 3899
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.018135017522853e-06,
+      "loss": 3.488,
+      "step": 3900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0161685809419515e-06,
+      "loss": 3.44,
+      "step": 3901
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0142019803999104e-06,
+      "loss": 3.5292,
+      "step": 3902
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.012235216214044e-06,
+      "loss": 3.5458,
+      "step": 3903
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.010268288701695e-06,
+      "loss": 3.6366,
+      "step": 3904
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.008301198180232e-06,
+      "loss": 3.514,
+      "step": 3905
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.006333944967048e-06,
+      "loss": 3.5265,
+      "step": 3906
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.004366529379564e-06,
+      "loss": 3.4935,
+      "step": 3907
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.0023989517352284e-06,
+      "loss": 3.5817,
+      "step": 3908
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.000431212351513e-06,
+      "loss": 3.5655,
+      "step": 3909
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.998463311545919e-06,
+      "loss": 3.603,
+      "step": 3910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.996495249635969e-06,
+      "loss": 3.5249,
+      "step": 3911
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.994527026939217e-06,
+      "loss": 3.464,
+      "step": 3912
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.992558643773239e-06,
+      "loss": 3.4659,
+      "step": 3913
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.990590100455638e-06,
+      "loss": 3.4669,
+      "step": 3914
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.988621397304043e-06,
+      "loss": 3.5492,
+      "step": 3915
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.986652534636108e-06,
+      "loss": 3.5636,
+      "step": 3916
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.984683512769517e-06,
+      "loss": 3.5369,
+      "step": 3917
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.982714332021971e-06,
+      "loss": 3.562,
+      "step": 3918
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.980744992711207e-06,
+      "loss": 3.528,
+      "step": 3919
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.978775495154979e-06,
+      "loss": 3.5606,
+      "step": 3920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.976805839671072e-06,
+      "loss": 3.4975,
+      "step": 3921
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.974836026577292e-06,
+      "loss": 3.5442,
+      "step": 3922
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.972866056191475e-06,
+      "loss": 3.6285,
+      "step": 3923
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.970895928831481e-06,
+      "loss": 3.7206,
+      "step": 3924
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.968925644815191e-06,
+      "loss": 3.5126,
+      "step": 3925
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.966955204460517e-06,
+      "loss": 3.4961,
+      "step": 3926
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.964984608085396e-06,
+      "loss": 3.5614,
+      "step": 3927
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.963013856007784e-06,
+      "loss": 3.6694,
+      "step": 3928
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.961042948545671e-06,
+      "loss": 3.5353,
+      "step": 3929
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.959071886017064e-06,
+      "loss": 3.5362,
+      "step": 3930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.95710066874e-06,
+      "loss": 3.6482,
+      "step": 3931
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9551292970325394e-06,
+      "loss": 3.5026,
+      "step": 3932
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9531577712127665e-06,
+      "loss": 3.5095,
+      "step": 3933
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9511860915987925e-06,
+      "loss": 3.5801,
+      "step": 3934
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.949214258508753e-06,
+      "loss": 3.5704,
+      "step": 3935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.947242272260807e-06,
+      "loss": 3.4962,
+      "step": 3936
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.945270133173139e-06,
+      "loss": 3.6198,
+      "step": 3937
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.943297841563958e-06,
+      "loss": 3.4897,
+      "step": 3938
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9413253977514985e-06,
+      "loss": 3.4526,
+      "step": 3939
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9393528020540195e-06,
+      "loss": 3.6079,
+      "step": 3940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.937380054789802e-06,
+      "loss": 3.6083,
+      "step": 3941
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.935407156277154e-06,
+      "loss": 3.4353,
+      "step": 3942
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.933434106834407e-06,
+      "loss": 3.6084,
+      "step": 3943
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.931460906779918e-06,
+      "loss": 3.5272,
+      "step": 3944
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.929487556432066e-06,
+      "loss": 3.4407,
+      "step": 3945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.927514056109257e-06,
+      "loss": 3.5231,
+      "step": 3946
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.925540406129919e-06,
+      "loss": 3.6266,
+      "step": 3947
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.923566606812505e-06,
+      "loss": 3.5656,
+      "step": 3948
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9215926584754926e-06,
+      "loss": 3.5994,
+      "step": 3949
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.919618561437382e-06,
+      "loss": 3.5962,
+      "step": 3950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.917644316016701e-06,
+      "loss": 3.6837,
+      "step": 3951
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.915669922531995e-06,
+      "loss": 3.5185,
+      "step": 3952
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.913695381301838e-06,
+      "loss": 3.5284,
+      "step": 3953
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.911720692644828e-06,
+      "loss": 3.5899,
+      "step": 3954
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.909745856879584e-06,
+      "loss": 3.5121,
+      "step": 3955
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.907770874324752e-06,
+      "loss": 3.5882,
+      "step": 3956
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.905795745298999e-06,
+      "loss": 3.5414,
+      "step": 3957
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.903820470121016e-06,
+      "loss": 3.6001,
+      "step": 3958
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.901845049109519e-06,
+      "loss": 3.5238,
+      "step": 3959
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.899869482583247e-06,
+      "loss": 3.6896,
+      "step": 3960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8978937708609615e-06,
+      "loss": 3.6312,
+      "step": 3961
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.895917914261449e-06,
+      "loss": 3.6014,
+      "step": 3962
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.893941913103518e-06,
+      "loss": 3.4914,
+      "step": 3963
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.891965767706001e-06,
+      "loss": 3.4341,
+      "step": 3964
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8899894783877536e-06,
+      "loss": 3.5057,
+      "step": 3965
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.888013045467653e-06,
+      "loss": 3.5474,
+      "step": 3966
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.886036469264603e-06,
+      "loss": 3.599,
+      "step": 3967
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.884059750097529e-06,
+      "loss": 3.5033,
+      "step": 3968
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.882082888285379e-06,
+      "loss": 3.4621,
+      "step": 3969
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.880105884147122e-06,
+      "loss": 3.6182,
+      "step": 3970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.878128738001756e-06,
+      "loss": 3.6547,
+      "step": 3971
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.876151450168295e-06,
+      "loss": 3.5765,
+      "step": 3972
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.874174020965781e-06,
+      "loss": 3.6487,
+      "step": 3973
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8721964507132745e-06,
+      "loss": 3.6679,
+      "step": 3974
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.870218739729862e-06,
+      "loss": 3.5454,
+      "step": 3975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8682408883346535e-06,
+      "loss": 3.4847,
+      "step": 3976
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8662628968467764e-06,
+      "loss": 3.4854,
+      "step": 3977
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8642847655853875e-06,
+      "loss": 3.6324,
+      "step": 3978
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.86230649486966e-06,
+      "loss": 3.6104,
+      "step": 3979
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.860328085018796e-06,
+      "loss": 3.6574,
+      "step": 3980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.858349536352015e-06,
+      "loss": 3.5895,
+      "step": 3981
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.856370849188557e-06,
+      "loss": 3.5763,
+      "step": 3982
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8543920238476926e-06,
+      "loss": 3.5367,
+      "step": 3983
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8524130606487066e-06,
+      "loss": 3.5216,
+      "step": 3984
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.85043395991091e-06,
+      "loss": 3.4503,
+      "step": 3985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.848454721953637e-06,
+      "loss": 3.5269,
+      "step": 3986
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.84647534709624e-06,
+      "loss": 3.5566,
+      "step": 3987
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.844495835658097e-06,
+      "loss": 3.5912,
+      "step": 3988
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.842516187958606e-06,
+      "loss": 3.6242,
+      "step": 3989
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.840536404317187e-06,
+      "loss": 3.6427,
+      "step": 3990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.838556485053285e-06,
+      "loss": 3.4847,
+      "step": 3991
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8365764304863625e-06,
+      "loss": 3.5613,
+      "step": 3992
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.834596240935907e-06,
+      "loss": 3.5611,
+      "step": 3993
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.832615916721426e-06,
+      "loss": 3.6026,
+      "step": 3994
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.830635458162449e-06,
+      "loss": 3.5402,
+      "step": 3995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.828654865578529e-06,
+      "loss": 3.566,
+      "step": 3996
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.826674139289238e-06,
+      "loss": 3.5146,
+      "step": 3997
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.824693279614171e-06,
+      "loss": 3.468,
+      "step": 3998
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8227122868729454e-06,
+      "loss": 3.5361,
+      "step": 3999
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.820731161385197e-06,
+      "loss": 3.6763,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.818749903470585e-06,
+      "loss": 3.5819,
+      "step": 4001
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8167685134487915e-06,
+      "loss": 3.6714,
+      "step": 4002
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.814786991639517e-06,
+      "loss": 3.5416,
+      "step": 4003
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.812805338362485e-06,
+      "loss": 3.5467,
+      "step": 4004
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.81082355393744e-06,
+      "loss": 3.6289,
+      "step": 4005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.808841638684148e-06,
+      "loss": 3.4585,
+      "step": 4006
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.806859592922396e-06,
+      "loss": 3.5784,
+      "step": 4007
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.804877416971989e-06,
+      "loss": 3.5527,
+      "step": 4008
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.802895111152758e-06,
+      "loss": 3.3792,
+      "step": 4009
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.8009126757845526e-06,
+      "loss": 3.4425,
+      "step": 4010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.798930111187242e-06,
+      "loss": 3.4767,
+      "step": 4011
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.796947417680719e-06,
+      "loss": 3.5266,
+      "step": 4012
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.794964595584893e-06,
+      "loss": 3.5011,
+      "step": 4013
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.7929816452197005e-06,
+      "loss": 3.5761,
+      "step": 4014
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.790998566905093e-06,
+      "loss": 3.5636,
+      "step": 4015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.789015360961046e-06,
+      "loss": 3.5826,
+      "step": 4016
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.787032027707553e-06,
+      "loss": 3.6239,
+      "step": 4017
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.785048567464631e-06,
+      "loss": 3.7098,
+      "step": 4018
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.783064980552314e-06,
+      "loss": 3.542,
+      "step": 4019
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.781081267290659e-06,
+      "loss": 3.5893,
+      "step": 4020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.779097427999743e-06,
+      "loss": 3.5433,
+      "step": 4021
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.777113462999664e-06,
+      "loss": 3.6077,
+      "step": 4022
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.775129372610537e-06,
+      "loss": 3.5509,
+      "step": 4023
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.773145157152501e-06,
+      "loss": 3.6243,
+      "step": 4024
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.771160816945713e-06,
+      "loss": 3.3943,
+      "step": 4025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.769176352310352e-06,
+      "loss": 3.6103,
+      "step": 4026
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.767191763566616e-06,
+      "loss": 3.4283,
+      "step": 4027
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.765207051034722e-06,
+      "loss": 3.5722,
+      "step": 4028
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.763222215034908e-06,
+      "loss": 3.5697,
+      "step": 4029
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.761237255887433e-06,
+      "loss": 3.4494,
+      "step": 4030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.759252173912573e-06,
+      "loss": 3.6735,
+      "step": 4031
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.757266969430626e-06,
+      "loss": 3.5929,
+      "step": 4032
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.755281642761911e-06,
+      "loss": 3.5834,
+      "step": 4033
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.753296194226763e-06,
+      "loss": 3.505,
+      "step": 4034
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.751310624145539e-06,
+      "loss": 3.5278,
+      "step": 4035
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.749324932838617e-06,
+      "loss": 3.5358,
+      "step": 4036
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.747339120626391e-06,
+      "loss": 3.4685,
+      "step": 4037
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7453531878292755e-06,
+      "loss": 3.4646,
+      "step": 4038
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.743367134767709e-06,
+      "loss": 3.5225,
+      "step": 4039
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.741380961762142e-06,
+      "loss": 3.5042,
+      "step": 4040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.739394669133051e-06,
+      "loss": 3.5059,
+      "step": 4041
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.737408257200926e-06,
+      "loss": 3.532,
+      "step": 4042
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.735421726286281e-06,
+      "loss": 3.6874,
+      "step": 4043
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7334350767096465e-06,
+      "loss": 3.453,
+      "step": 4044
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.731448308791575e-06,
+      "loss": 3.6011,
+      "step": 4045
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7294614228526335e-06,
+      "loss": 3.5491,
+      "step": 4046
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.727474419213413e-06,
+      "loss": 3.5745,
+      "step": 4047
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.725487298194518e-06,
+      "loss": 3.5407,
+      "step": 4048
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.723500060116578e-06,
+      "loss": 3.5618,
+      "step": 4049
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.721512705300236e-06,
+      "loss": 3.5209,
+      "step": 4050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.719525234066159e-06,
+      "loss": 3.6456,
+      "step": 4051
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7175376467350275e-06,
+      "loss": 3.5163,
+      "step": 4052
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.715549943627545e-06,
+      "loss": 3.4894,
+      "step": 4053
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7135621250644325e-06,
+      "loss": 3.5006,
+      "step": 4054
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.711574191366427e-06,
+      "loss": 3.5758,
+      "step": 4055
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.709586142854287e-06,
+      "loss": 3.5739,
+      "step": 4056
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.7075979798487895e-06,
+      "loss": 3.4172,
+      "step": 4057
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.705609702670727e-06,
+      "loss": 3.5086,
+      "step": 4058
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.703621311640916e-06,
+      "loss": 3.5434,
+      "step": 4059
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.701632807080184e-06,
+      "loss": 3.5682,
+      "step": 4060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.699644189309384e-06,
+      "loss": 3.3954,
+      "step": 4061
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.697655458649383e-06,
+      "loss": 3.5255,
+      "step": 4062
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.695666615421069e-06,
+      "loss": 3.5617,
+      "step": 4063
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.693677659945343e-06,
+      "loss": 3.5356,
+      "step": 4064
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.691688592543129e-06,
+      "loss": 3.6231,
+      "step": 4065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.689699413535367e-06,
+      "loss": 3.5,
+      "step": 4066
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.687710123243017e-06,
+      "loss": 3.5151,
+      "step": 4067
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.685720721987056e-06,
+      "loss": 3.6545,
+      "step": 4068
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.683731210088477e-06,
+      "loss": 3.5297,
+      "step": 4069
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.681741587868292e-06,
+      "loss": 3.5293,
+      "step": 4070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.679751855647534e-06,
+      "loss": 3.673,
+      "step": 4071
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.677762013747248e-06,
+      "loss": 3.4893,
+      "step": 4072
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6757720624885004e-06,
+      "loss": 3.5505,
+      "step": 4073
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.673782002192374e-06,
+      "loss": 3.6869,
+      "step": 4074
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.671791833179969e-06,
+      "loss": 3.541,
+      "step": 4075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.669801555772406e-06,
+      "loss": 3.4446,
+      "step": 4076
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.667811170290819e-06,
+      "loss": 3.5896,
+      "step": 4077
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.665820677056363e-06,
+      "loss": 3.626,
+      "step": 4078
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.663830076390206e-06,
+      "loss": 3.5328,
+      "step": 4079
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.661839368613538e-06,
+      "loss": 3.5662,
+      "step": 4080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6598485540475645e-06,
+      "loss": 3.6251,
+      "step": 4081
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.657857633013507e-06,
+      "loss": 3.4992,
+      "step": 4082
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6558666058326065e-06,
+      "loss": 3.5117,
+      "step": 4083
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.653875472826117e-06,
+      "loss": 3.4457,
+      "step": 4084
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.651884234315316e-06,
+      "loss": 3.538,
+      "step": 4085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.649892890621491e-06,
+      "loss": 3.601,
+      "step": 4086
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.647901442065953e-06,
+      "loss": 3.6291,
+      "step": 4087
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6459098889700246e-06,
+      "loss": 3.5455,
+      "step": 4088
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.643918231655048e-06,
+      "loss": 3.452,
+      "step": 4089
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.641926470442384e-06,
+      "loss": 3.6392,
+      "step": 4090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.639934605653405e-06,
+      "loss": 3.5366,
+      "step": 4091
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.637942637609502e-06,
+      "loss": 3.6209,
+      "step": 4092
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.635950566632087e-06,
+      "loss": 3.5114,
+      "step": 4093
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.633958393042582e-06,
+      "loss": 3.5104,
+      "step": 4094
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6319661171624315e-06,
+      "loss": 3.6159,
+      "step": 4095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.629973739313094e-06,
+      "loss": 3.4561,
+      "step": 4096
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.627981259816041e-06,
+      "loss": 3.5341,
+      "step": 4097
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6259886789927665e-06,
+      "loss": 3.5826,
+      "step": 4098
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.623995997164777e-06,
+      "loss": 3.6067,
+      "step": 4099
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6220032146535974e-06,
+      "loss": 3.5111,
+      "step": 4100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.620010331780765e-06,
+      "loss": 3.5575,
+      "step": 4101
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.618017348867839e-06,
+      "loss": 3.4804,
+      "step": 4102
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.616024266236391e-06,
+      "loss": 3.5305,
+      "step": 4103
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.614031084208009e-06,
+      "loss": 3.536,
+      "step": 4104
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.612037803104297e-06,
+      "loss": 3.5638,
+      "step": 4105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.610044423246877e-06,
+      "loss": 3.6265,
+      "step": 4106
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.608050944957384e-06,
+      "loss": 3.5092,
+      "step": 4107
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.6060573685574704e-06,
+      "loss": 3.6372,
+      "step": 4108
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.604063694368807e-06,
+      "loss": 3.5363,
+      "step": 4109
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.602069922713072e-06,
+      "loss": 3.5318,
+      "step": 4110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.600076053911972e-06,
+      "loss": 3.5057,
+      "step": 4111
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.598082088287215e-06,
+      "loss": 3.5462,
+      "step": 4112
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.596088026160539e-06,
+      "loss": 3.5503,
+      "step": 4113
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.594093867853686e-06,
+      "loss": 3.4791,
+      "step": 4114
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.59209961368842e-06,
+      "loss": 3.576,
+      "step": 4115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.590105263986518e-06,
+      "loss": 3.422,
+      "step": 4116
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.588110819069772e-06,
+      "loss": 3.5461,
+      "step": 4117
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.586116279259992e-06,
+      "loss": 3.5423,
+      "step": 4118
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.584121644879001e-06,
+      "loss": 3.5822,
+      "step": 4119
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.582126916248636e-06,
+      "loss": 3.5927,
+      "step": 4120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.580132093690753e-06,
+      "loss": 3.522,
+      "step": 4121
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.578137177527221e-06,
+      "loss": 3.4289,
+      "step": 4122
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.576142168079925e-06,
+      "loss": 3.5531,
+      "step": 4123
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.574147065670762e-06,
+      "loss": 3.5811,
+      "step": 4124
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.572151870621648e-06,
+      "loss": 3.5385,
+      "step": 4125
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.570156583254513e-06,
+      "loss": 3.5022,
+      "step": 4126
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.568161203891302e-06,
+      "loss": 3.6122,
+      "step": 4127
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.56616573285397e-06,
+      "loss": 3.5037,
+      "step": 4128
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.564170170464493e-06,
+      "loss": 3.5181,
+      "step": 4129
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.562174517044862e-06,
+      "loss": 3.3627,
+      "step": 4130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.560178772917076e-06,
+      "loss": 3.4258,
+      "step": 4131
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.558182938403156e-06,
+      "loss": 3.5175,
+      "step": 4132
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.556187013825134e-06,
+      "loss": 3.5099,
+      "step": 4133
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5541909995050554e-06,
+      "loss": 3.517,
+      "step": 4134
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.552194895764984e-06,
+      "loss": 3.5528,
+      "step": 4135
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.550198702926994e-06,
+      "loss": 3.5646,
+      "step": 4136
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.548202421313177e-06,
+      "loss": 3.4864,
+      "step": 4137
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5462060512456366e-06,
+      "loss": 3.5811,
+      "step": 4138
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.544209593046491e-06,
+      "loss": 3.5556,
+      "step": 4139
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5422130470378756e-06,
+      "loss": 3.6086,
+      "step": 4140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.540216413541937e-06,
+      "loss": 3.6036,
+      "step": 4141
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.538219692880834e-06,
+      "loss": 3.6393,
+      "step": 4142
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.536222885376746e-06,
+      "loss": 3.5174,
+      "step": 4143
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.53422599135186e-06,
+      "loss": 3.5503,
+      "step": 4144
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.532229011128382e-06,
+      "loss": 3.5567,
+      "step": 4145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.530231945028525e-06,
+      "loss": 3.5512,
+      "step": 4146
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.528234793374524e-06,
+      "loss": 3.7422,
+      "step": 4147
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.526237556488622e-06,
+      "loss": 3.5439,
+      "step": 4148
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5242402346930775e-06,
+      "loss": 3.5937,
+      "step": 4149
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.522242828310166e-06,
+      "loss": 3.6293,
+      "step": 4150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.52024533766217e-06,
+      "loss": 3.5774,
+      "step": 4151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.518247763071392e-06,
+      "loss": 3.5745,
+      "step": 4152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.516250104860144e-06,
+      "loss": 3.5044,
+      "step": 4153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.514252363350751e-06,
+      "loss": 3.6122,
+      "step": 4154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5122545388655565e-06,
+      "loss": 3.6232,
+      "step": 4155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.510256631726912e-06,
+      "loss": 3.5664,
+      "step": 4156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.508258642257186e-06,
+      "loss": 3.5451,
+      "step": 4157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.506260570778758e-06,
+      "loss": 3.5626,
+      "step": 4158
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.504262417614019e-06,
+      "loss": 3.5103,
+      "step": 4159
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.502264183085379e-06,
+      "loss": 3.5487,
+      "step": 4160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5002658675152565e-06,
+      "loss": 3.5101,
+      "step": 4161
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.498267471226084e-06,
+      "loss": 3.5428,
+      "step": 4162
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.496268994540309e-06,
+      "loss": 3.5962,
+      "step": 4163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.494270437780388e-06,
+      "loss": 3.6314,
+      "step": 4164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.492271801268794e-06,
+      "loss": 3.5965,
+      "step": 4165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.49027308532801e-06,
+      "loss": 3.4912,
+      "step": 4166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.488274290280537e-06,
+      "loss": 3.5495,
+      "step": 4167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.486275416448881e-06,
+      "loss": 3.4806,
+      "step": 4168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.484276464155566e-06,
+      "loss": 3.56,
+      "step": 4169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.48227743372313e-06,
+      "loss": 3.531,
+      "step": 4170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.480278325474119e-06,
+      "loss": 3.5271,
+      "step": 4171
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.478279139731094e-06,
+      "loss": 3.5097,
+      "step": 4172
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.476279876816629e-06,
+      "loss": 3.5971,
+      "step": 4173
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.474280537053309e-06,
+      "loss": 3.563,
+      "step": 4174
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4722811207637326e-06,
+      "loss": 3.5626,
+      "step": 4175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.470281628270508e-06,
+      "loss": 3.563,
+      "step": 4176
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.468282059896262e-06,
+      "loss": 3.5166,
+      "step": 4177
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.466282415963626e-06,
+      "loss": 3.6201,
+      "step": 4178
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.464282696795249e-06,
+      "loss": 3.467,
+      "step": 4179
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.462282902713791e-06,
+      "loss": 3.527,
+      "step": 4180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.46028303404192e-06,
+      "loss": 3.4246,
+      "step": 4181
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.458283091102323e-06,
+      "loss": 3.5629,
+      "step": 4182
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.456283074217693e-06,
+      "loss": 3.5552,
+      "step": 4183
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.454282983710739e-06,
+      "loss": 3.575,
+      "step": 4184
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.452282819904179e-06,
+      "loss": 3.4886,
+      "step": 4185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.450282583120745e-06,
+      "loss": 3.5332,
+      "step": 4186
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.448282273683179e-06,
+      "loss": 3.461,
+      "step": 4187
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.446281891914236e-06,
+      "loss": 3.5511,
+      "step": 4188
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.444281438136682e-06,
+      "loss": 3.5518,
+      "step": 4189
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.442280912673294e-06,
+      "loss": 3.5199,
+      "step": 4190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.440280315846863e-06,
+      "loss": 3.484,
+      "step": 4191
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4382796479801896e-06,
+      "loss": 3.5184,
+      "step": 4192
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.436278909396086e-06,
+      "loss": 3.5928,
+      "step": 4193
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.434278100417374e-06,
+      "loss": 3.5481,
+      "step": 4194
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.432277221366891e-06,
+      "loss": 3.4801,
+      "step": 4195
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.430276272567485e-06,
+      "loss": 3.5001,
+      "step": 4196
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.42827525434201e-06,
+      "loss": 3.593,
+      "step": 4197
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.4262741670133366e-06,
+      "loss": 3.4586,
+      "step": 4198
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.424273010904345e-06,
+      "loss": 3.4813,
+      "step": 4199
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.422271786337927e-06,
+      "loss": 3.603,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.420270493636982e-06,
+      "loss": 3.4768,
+      "step": 4201
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.418269133124427e-06,
+      "loss": 3.5183,
+      "step": 4202
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.4162677051231825e-06,
+      "loss": 3.5388,
+      "step": 4203
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.414266209956186e-06,
+      "loss": 3.4977,
+      "step": 4204
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.412264647946382e-06,
+      "loss": 3.534,
+      "step": 4205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.410263019416729e-06,
+      "loss": 3.5442,
+      "step": 4206
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.4082613246901915e-06,
+      "loss": 3.4261,
+      "step": 4207
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.4062595640897504e-06,
+      "loss": 3.6749,
+      "step": 4208
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.404257737938392e-06,
+      "loss": 3.5184,
+      "step": 4209
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.402255846559118e-06,
+      "loss": 3.5082,
+      "step": 4210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.400253890274936e-06,
+      "loss": 3.5381,
+      "step": 4211
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.398251869408867e-06,
+      "loss": 3.5583,
+      "step": 4212
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.396249784283943e-06,
+      "loss": 3.619,
+      "step": 4213
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.394247635223203e-06,
+      "loss": 3.5053,
+      "step": 4214
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.392245422549699e-06,
+      "loss": 3.5973,
+      "step": 4215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.390243146586493e-06,
+      "loss": 3.6274,
+      "step": 4216
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.388240807656657e-06,
+      "loss": 3.5854,
+      "step": 4217
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.386238406083273e-06,
+      "loss": 3.559,
+      "step": 4218
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3842359421894315e-06,
+      "loss": 3.5642,
+      "step": 4219
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.382233416298237e-06,
+      "loss": 3.534,
+      "step": 4220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.380230828732801e-06,
+      "loss": 3.5404,
+      "step": 4221
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.378228179816244e-06,
+      "loss": 3.47,
+      "step": 4222
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3762254698717e-06,
+      "loss": 3.6452,
+      "step": 4223
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.37422269922231e-06,
+      "loss": 3.5498,
+      "step": 4224
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.372219868191225e-06,
+      "loss": 3.6449,
+      "step": 4225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3702169771016074e-06,
+      "loss": 3.6028,
+      "step": 4226
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3682140262766295e-06,
+      "loss": 3.6111,
+      "step": 4227
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.366211016039469e-06,
+      "loss": 3.5734,
+      "step": 4228
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.364207946713318e-06,
+      "loss": 3.5067,
+      "step": 4229
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.362204818621376e-06,
+      "loss": 3.468,
+      "step": 4230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.360201632086853e-06,
+      "loss": 3.5783,
+      "step": 4231
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.358198387432967e-06,
+      "loss": 3.4926,
+      "step": 4232
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.356195084982946e-06,
+      "loss": 3.655,
+      "step": 4233
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.354191725060028e-06,
+      "loss": 3.5811,
+      "step": 4234
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.35218830798746e-06,
+      "loss": 3.523,
+      "step": 4235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.350184834088499e-06,
+      "loss": 3.536,
+      "step": 4236
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.348181303686406e-06,
+      "loss": 3.5507,
+      "step": 4237
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.34617771710446e-06,
+      "loss": 3.5183,
+      "step": 4238
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.34417407466594e-06,
+      "loss": 3.6132,
+      "step": 4239
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.342170376694141e-06,
+      "loss": 3.6292,
+      "step": 4240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.340166623512364e-06,
+      "loss": 3.4625,
+      "step": 4241
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.338162815443919e-06,
+      "loss": 3.447,
+      "step": 4242
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.336158952812126e-06,
+      "loss": 3.5149,
+      "step": 4243
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.33415503594031e-06,
+      "loss": 3.619,
+      "step": 4244
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.332151065151811e-06,
+      "loss": 3.4619,
+      "step": 4245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.330147040769973e-06,
+      "loss": 3.4847,
+      "step": 4246
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.328142963118149e-06,
+      "loss": 3.4688,
+      "step": 4247
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.326138832519703e-06,
+      "loss": 3.6121,
+      "step": 4248
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.324134649298004e-06,
+      "loss": 3.4611,
+      "step": 4249
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.322130413776434e-06,
+      "loss": 3.5588,
+      "step": 4250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.32012612627838e-06,
+      "loss": 3.5457,
+      "step": 4251
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.318121787127237e-06,
+      "loss": 3.5666,
+      "step": 4252
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.316117396646413e-06,
+      "loss": 3.6924,
+      "step": 4253
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3141129551593194e-06,
+      "loss": 3.5038,
+      "step": 4254
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.312108462989376e-06,
+      "loss": 3.4954,
+      "step": 4255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3101039204600135e-06,
+      "loss": 3.5274,
+      "step": 4256
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.308099327894669e-06,
+      "loss": 3.5048,
+      "step": 4257
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.306094685616789e-06,
+      "loss": 3.4823,
+      "step": 4258
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3040899939498265e-06,
+      "loss": 3.4676,
+      "step": 4259
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.302085253217243e-06,
+      "loss": 3.6596,
+      "step": 4260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.3000804637425085e-06,
+      "loss": 3.6185,
+      "step": 4261
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2980756258491e-06,
+      "loss": 3.4937,
+      "step": 4262
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.296070739860502e-06,
+      "loss": 3.5725,
+      "step": 4263
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2940658061002105e-06,
+      "loss": 3.5638,
+      "step": 4264
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.292060824891722e-06,
+      "loss": 3.4018,
+      "step": 4265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2900557965585485e-06,
+      "loss": 3.5973,
+      "step": 4266
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.288050721424204e-06,
+      "loss": 3.6215,
+      "step": 4267
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.286045599812212e-06,
+      "loss": 3.5484,
+      "step": 4268
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.284040432046102e-06,
+      "loss": 3.6049,
+      "step": 4269
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.282035218449417e-06,
+      "loss": 3.5761,
+      "step": 4270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.280029959345698e-06,
+      "loss": 3.5508,
+      "step": 4271
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.278024655058502e-06,
+      "loss": 3.5235,
+      "step": 4272
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.276019305911387e-06,
+      "loss": 3.5058,
+      "step": 4273
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2740139122279225e-06,
+      "loss": 3.4278,
+      "step": 4274
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.27200847433168e-06,
+      "loss": 3.5204,
+      "step": 4275
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.270002992546245e-06,
+      "loss": 3.5929,
+      "step": 4276
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.267997467195205e-06,
+      "loss": 3.5859,
+      "step": 4277
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2659918986021575e-06,
+      "loss": 3.4845,
+      "step": 4278
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2639862870907035e-06,
+      "loss": 3.6313,
+      "step": 4279
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2619806329844545e-06,
+      "loss": 3.5664,
+      "step": 4280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.259974936607027e-06,
+      "loss": 3.5442,
+      "step": 4281
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.257969198282046e-06,
+      "loss": 3.4382,
+      "step": 4282
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.25596341833314e-06,
+      "loss": 3.5054,
+      "step": 4283
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.253957597083946e-06,
+      "loss": 3.5615,
+      "step": 4284
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.251951734858109e-06,
+      "loss": 3.5243,
+      "step": 4285
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.249945831979282e-06,
+      "loss": 3.5387,
+      "step": 4286
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.247939888771116e-06,
+      "loss": 3.5448,
+      "step": 4287
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.245933905557281e-06,
+      "loss": 3.5521,
+      "step": 4288
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.243927882661443e-06,
+      "loss": 3.5223,
+      "step": 4289
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2419218204072785e-06,
+      "loss": 3.5676,
+      "step": 4290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.239915719118472e-06,
+      "loss": 3.5115,
+      "step": 4291
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.237909579118713e-06,
+      "loss": 3.5196,
+      "step": 4292
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.235903400731693e-06,
+      "loss": 3.5855,
+      "step": 4293
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.233897184281118e-06,
+      "loss": 3.595,
+      "step": 4294
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.231890930090692e-06,
+      "loss": 3.5111,
+      "step": 4295
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.229884638484132e-06,
+      "loss": 3.6027,
+      "step": 4296
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.227878309785156e-06,
+      "loss": 3.5715,
+      "step": 4297
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.22587194431749e-06,
+      "loss": 3.622,
+      "step": 4298
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.223865542404865e-06,
+      "loss": 3.4136,
+      "step": 4299
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.221859104371019e-06,
+      "loss": 3.7229,
+      "step": 4300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2198526305396965e-06,
+      "loss": 3.4042,
+      "step": 4301
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.217846121234645e-06,
+      "loss": 3.5324,
+      "step": 4302
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.215839576779621e-06,
+      "loss": 3.5456,
+      "step": 4303
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.213832997498383e-06,
+      "loss": 3.7004,
+      "step": 4304
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2118263837147e-06,
+      "loss": 3.5491,
+      "step": 4305
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.209819735752342e-06,
+      "loss": 3.4493,
+      "step": 4306
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.2078130539350856e-06,
+      "loss": 3.4842,
+      "step": 4307
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.205806338586715e-06,
+      "loss": 3.5258,
+      "step": 4308
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.203799590031019e-06,
+      "loss": 3.6228,
+      "step": 4309
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.201792808591788e-06,
+      "loss": 3.4689,
+      "step": 4310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.199785994592823e-06,
+      "loss": 3.5996,
+      "step": 4311
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.197779148357928e-06,
+      "loss": 3.6092,
+      "step": 4312
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.195772270210911e-06,
+      "loss": 3.5674,
+      "step": 4313
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1937653604755885e-06,
+      "loss": 3.547,
+      "step": 4314
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.19175841947578e-06,
+      "loss": 3.5373,
+      "step": 4315
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1897514475353075e-06,
+      "loss": 3.6488,
+      "step": 4316
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.187744444978001e-06,
+      "loss": 3.5387,
+      "step": 4317
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.185737412127697e-06,
+      "loss": 3.4394,
+      "step": 4318
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.183730349308234e-06,
+      "loss": 3.549,
+      "step": 4319
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.181723256843456e-06,
+      "loss": 3.5428,
+      "step": 4320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.179716135057212e-06,
+      "loss": 3.6976,
+      "step": 4321
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.177708984273356e-06,
+      "loss": 3.5849,
+      "step": 4322
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.175701804815746e-06,
+      "loss": 3.4715,
+      "step": 4323
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.173694597008245e-06,
+      "loss": 3.5262,
+      "step": 4324
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.171687361174721e-06,
+      "loss": 3.6834,
+      "step": 4325
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.169680097639046e-06,
+      "loss": 3.5377,
+      "step": 4326
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.167672806725097e-06,
+      "loss": 3.6152,
+      "step": 4327
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.165665488756755e-06,
+      "loss": 3.5388,
+      "step": 4328
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.163658144057904e-06,
+      "loss": 3.4816,
+      "step": 4329
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.161650772952433e-06,
+      "loss": 3.4578,
+      "step": 4330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.15964337576424e-06,
+      "loss": 3.4419,
+      "step": 4331
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1576359528172194e-06,
+      "loss": 3.4983,
+      "step": 4332
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.155628504435275e-06,
+      "loss": 3.5894,
+      "step": 4333
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.153621030942313e-06,
+      "loss": 3.5267,
+      "step": 4334
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.151613532662244e-06,
+      "loss": 3.5296,
+      "step": 4335
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.149606009918982e-06,
+      "loss": 3.5789,
+      "step": 4336
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1475984630364475e-06,
+      "loss": 3.5905,
+      "step": 4337
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.145590892338559e-06,
+      "loss": 3.5877,
+      "step": 4338
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.143583298149246e-06,
+      "loss": 3.5099,
+      "step": 4339
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1415756807924365e-06,
+      "loss": 3.5053,
+      "step": 4340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.139568040592066e-06,
+      "loss": 3.5792,
+      "step": 4341
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.13756037787207e-06,
+      "loss": 3.6088,
+      "step": 4342
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1355526929563895e-06,
+      "loss": 3.5251,
+      "step": 4343
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.133544986168972e-06,
+      "loss": 3.5793,
+      "step": 4344
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1315372578337615e-06,
+      "loss": 3.5211,
+      "step": 4345
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.129529508274712e-06,
+      "loss": 3.4725,
+      "step": 4346
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.127521737815778e-06,
+      "loss": 3.5439,
+      "step": 4347
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.125513946780919e-06,
+      "loss": 3.5356,
+      "step": 4348
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1235061354940956e-06,
+      "loss": 3.6376,
+      "step": 4349
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.121498304279271e-06,
+      "loss": 3.5034,
+      "step": 4350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1194904534604165e-06,
+      "loss": 3.6385,
+      "step": 4351
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1174825833615015e-06,
+      "loss": 3.5944,
+      "step": 4352
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.115474694306502e-06,
+      "loss": 3.5726,
+      "step": 4353
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.113466786619394e-06,
+      "loss": 3.4712,
+      "step": 4354
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.111458860624159e-06,
+      "loss": 3.4387,
+      "step": 4355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.10945091664478e-06,
+      "loss": 3.5889,
+      "step": 4356
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.107442955005244e-06,
+      "loss": 3.4769,
+      "step": 4357
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.1054349760295395e-06,
+      "loss": 3.5647,
+      "step": 4358
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.103426980041659e-06,
+      "loss": 3.4229,
+      "step": 4359
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.101418967365597e-06,
+      "loss": 3.5186,
+      "step": 4360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.099410938325351e-06,
+      "loss": 3.605,
+      "step": 4361
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.09740289324492e-06,
+      "loss": 3.5024,
+      "step": 4362
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.09539483244831e-06,
+      "loss": 3.5216,
+      "step": 4363
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.093386756259524e-06,
+      "loss": 3.5604,
+      "step": 4364
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.09137866500257e-06,
+      "loss": 3.5162,
+      "step": 4365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.0893705590014564e-06,
+      "loss": 3.5068,
+      "step": 4366
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.087362438580198e-06,
+      "loss": 3.5938,
+      "step": 4367
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.085354304062808e-06,
+      "loss": 3.5923,
+      "step": 4368
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.083346155773306e-06,
+      "loss": 3.4821,
+      "step": 4369
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0813379940357096e-06,
+      "loss": 3.4996,
+      "step": 4370
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.07932981917404e-06,
+      "loss": 3.4662,
+      "step": 4371
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.077321631512322e-06,
+      "loss": 3.5499,
+      "step": 4372
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.07531343137458e-06,
+      "loss": 3.4775,
+      "step": 4373
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.073305219084843e-06,
+      "loss": 3.4807,
+      "step": 4374
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.07129699496714e-06,
+      "loss": 3.6037,
+      "step": 4375
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.069288759345502e-06,
+      "loss": 3.6269,
+      "step": 4376
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.067280512543964e-06,
+      "loss": 3.5885,
+      "step": 4377
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.06527225488656e-06,
+      "loss": 3.5805,
+      "step": 4378
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0632639866973265e-06,
+      "loss": 3.5293,
+      "step": 4379
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0612557083003045e-06,
+      "loss": 3.5033,
+      "step": 4380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.059247420019532e-06,
+      "loss": 3.5891,
+      "step": 4381
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.057239122179054e-06,
+      "loss": 3.506,
+      "step": 4382
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.055230815102909e-06,
+      "loss": 3.6437,
+      "step": 4383
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.053222499115147e-06,
+      "loss": 3.5418,
+      "step": 4384
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.05121417453981e-06,
+      "loss": 3.516,
+      "step": 4385
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.04920584170095e-06,
+      "loss": 3.4891,
+      "step": 4386
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.047197500922612e-06,
+      "loss": 3.5811,
+      "step": 4387
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.045189152528851e-06,
+      "loss": 3.5645,
+      "step": 4388
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.043180796843714e-06,
+      "loss": 3.513,
+      "step": 4389
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.041172434191257e-06,
+      "loss": 3.5339,
+      "step": 4390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.039164064895533e-06,
+      "loss": 3.6236,
+      "step": 4391
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.037155689280597e-06,
+      "loss": 3.3723,
+      "step": 4392
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0351473076705046e-06,
+      "loss": 3.526,
+      "step": 4393
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.033138920389313e-06,
+      "loss": 3.6235,
+      "step": 4394
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.031130527761081e-06,
+      "loss": 3.4761,
+      "step": 4395
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0291221301098655e-06,
+      "loss": 3.4926,
+      "step": 4396
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.027113727759729e-06,
+      "loss": 3.5574,
+      "step": 4397
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.025105321034729e-06,
+      "loss": 3.6068,
+      "step": 4398
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.023096910258928e-06,
+      "loss": 3.4974,
+      "step": 4399
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.021088495756388e-06,
+      "loss": 3.5366,
+      "step": 4400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.019080077851169e-06,
+      "loss": 3.5533,
+      "step": 4401
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0170716568673375e-06,
+      "loss": 3.5544,
+      "step": 4402
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.015063233128953e-06,
+      "loss": 3.4905,
+      "step": 4403
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0130548069600814e-06,
+      "loss": 3.5687,
+      "step": 4404
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0110463786847875e-06,
+      "loss": 3.5868,
+      "step": 4405
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.0090379486271346e-06,
+      "loss": 3.6138,
+      "step": 4406
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.007029517111187e-06,
+      "loss": 3.4812,
+      "step": 4407
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.005021084461011e-06,
+      "loss": 3.528,
+      "step": 4408
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.003012651000673e-06,
+      "loss": 3.5848,
+      "step": 4409
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.001004217054235e-06,
+      "loss": 3.5356,
+      "step": 4410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9989957829457666e-06,
+      "loss": 3.6362,
+      "step": 4411
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.996987348999328e-06,
+      "loss": 3.5327,
+      "step": 4412
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.99497891553899e-06,
+      "loss": 3.4836,
+      "step": 4413
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9929704828888134e-06,
+      "loss": 3.6221,
+      "step": 4414
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.990962051372867e-06,
+      "loss": 3.5426,
+      "step": 4415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9889536213152125e-06,
+      "loss": 3.6673,
+      "step": 4416
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.98694519303992e-06,
+      "loss": 3.498,
+      "step": 4417
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.984936766871049e-06,
+      "loss": 3.5418,
+      "step": 4418
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.982928343132665e-06,
+      "loss": 3.4506,
+      "step": 4419
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.980919922148832e-06,
+      "loss": 3.5192,
+      "step": 4420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9789115042436136e-06,
+      "loss": 3.4784,
+      "step": 4421
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.976903089741073e-06,
+      "loss": 3.4301,
+      "step": 4422
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.974894678965273e-06,
+      "loss": 3.626,
+      "step": 4423
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9728862722402715e-06,
+      "loss": 3.5534,
+      "step": 4424
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.970877869890135e-06,
+      "loss": 3.5015,
+      "step": 4425
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.968869472238919e-06,
+      "loss": 3.44,
+      "step": 4426
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.966861079610688e-06,
+      "loss": 3.5007,
+      "step": 4427
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.964852692329495e-06,
+      "loss": 3.5045,
+      "step": 4428
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.962844310719405e-06,
+      "loss": 3.4876,
+      "step": 4429
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9608359351044676e-06,
+      "loss": 3.5888,
+      "step": 4430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.958827565808744e-06,
+      "loss": 3.6657,
+      "step": 4431
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.956819203156286e-06,
+      "loss": 3.5697,
+      "step": 4432
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.954810847471151e-06,
+      "loss": 3.6226,
+      "step": 4433
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9528024990773876e-06,
+      "loss": 3.5425,
+      "step": 4434
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.950794158299052e-06,
+      "loss": 3.4732,
+      "step": 4435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.948785825460191e-06,
+      "loss": 3.5588,
+      "step": 4436
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9467775008848554e-06,
+      "loss": 3.5676,
+      "step": 4437
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.944769184897092e-06,
+      "loss": 3.6006,
+      "step": 4438
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.942760877820948e-06,
+      "loss": 3.4631,
+      "step": 4439
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.940752579980468e-06,
+      "loss": 3.6448,
+      "step": 4440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.938744291699697e-06,
+      "loss": 3.5311,
+      "step": 4441
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9367360133026734e-06,
+      "loss": 3.5128,
+      "step": 4442
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.934727745113442e-06,
+      "loss": 3.6255,
+      "step": 4443
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.932719487456037e-06,
+      "loss": 3.4832,
+      "step": 4444
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.930711240654499e-06,
+      "loss": 3.5773,
+      "step": 4445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.928703005032861e-06,
+      "loss": 3.5895,
+      "step": 4446
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.926694780915159e-06,
+      "loss": 3.566,
+      "step": 4447
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.92468656862542e-06,
+      "loss": 3.474,
+      "step": 4448
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.92267836848768e-06,
+      "loss": 3.4948,
+      "step": 4449
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.9206701808259605e-06,
+      "loss": 3.4979,
+      "step": 4450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.918662005964292e-06,
+      "loss": 3.6475,
+      "step": 4451
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9166538442266945e-06,
+      "loss": 3.5141,
+      "step": 4452
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.914645695937193e-06,
+      "loss": 3.5052,
+      "step": 4453
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.912637561419803e-06,
+      "loss": 3.5735,
+      "step": 4454
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.910629440998545e-06,
+      "loss": 3.6318,
+      "step": 4455
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.908621334997432e-06,
+      "loss": 3.5035,
+      "step": 4456
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.906613243740477e-06,
+      "loss": 3.5685,
+      "step": 4457
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.904605167551691e-06,
+      "loss": 3.5585,
+      "step": 4458
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.9025971067550805e-06,
+      "loss": 3.5535,
+      "step": 4459
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.900589061674649e-06,
+      "loss": 3.5131,
+      "step": 4460
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.898581032634405e-06,
+      "loss": 3.6402,
+      "step": 4461
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.896573019958342e-06,
+      "loss": 3.5367,
+      "step": 4462
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.894565023970462e-06,
+      "loss": 3.6337,
+      "step": 4463
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.892557044994756e-06,
+      "loss": 3.564,
+      "step": 4464
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.890549083355221e-06,
+      "loss": 3.5843,
+      "step": 4465
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.888541139375841e-06,
+      "loss": 3.5698,
+      "step": 4466
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.886533213380607e-06,
+      "loss": 3.6211,
+      "step": 4467
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.884525305693498e-06,
+      "loss": 3.5719,
+      "step": 4468
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8825174166385e-06,
+      "loss": 3.4896,
+      "step": 4469
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.880509546539584e-06,
+      "loss": 3.516,
+      "step": 4470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.87850169572073e-06,
+      "loss": 3.5853,
+      "step": 4471
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.876493864505905e-06,
+      "loss": 3.5593,
+      "step": 4472
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.874486053219082e-06,
+      "loss": 3.5299,
+      "step": 4473
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8724782621842225e-06,
+      "loss": 3.4855,
+      "step": 4474
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.870470491725289e-06,
+      "loss": 3.5526,
+      "step": 4475
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.868462742166239e-06,
+      "loss": 3.6001,
+      "step": 4476
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.866455013831031e-06,
+      "loss": 3.5665,
+      "step": 4477
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8644473070436105e-06,
+      "loss": 3.5728,
+      "step": 4478
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.862439622127932e-06,
+      "loss": 3.5997,
+      "step": 4479
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.860431959407935e-06,
+      "loss": 3.6131,
+      "step": 4480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.858424319207564e-06,
+      "loss": 3.489,
+      "step": 4481
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.856416701850755e-06,
+      "loss": 3.4099,
+      "step": 4482
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.854409107661442e-06,
+      "loss": 3.5897,
+      "step": 4483
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.852401536963553e-06,
+      "loss": 3.6231,
+      "step": 4484
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8503939900810185e-06,
+      "loss": 3.5174,
+      "step": 4485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.848386467337756e-06,
+      "loss": 3.5503,
+      "step": 4486
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.846378969057688e-06,
+      "loss": 3.5973,
+      "step": 4487
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8443714955647255e-06,
+      "loss": 3.4661,
+      "step": 4488
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.842364047182782e-06,
+      "loss": 3.5531,
+      "step": 4489
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.840356624235761e-06,
+      "loss": 3.5526,
+      "step": 4490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8383492270475675e-06,
+      "loss": 3.6,
+      "step": 4491
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8363418559420985e-06,
+      "loss": 3.552,
+      "step": 4492
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8343345112432475e-06,
+      "loss": 3.5357,
+      "step": 4493
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.832327193274904e-06,
+      "loss": 3.5714,
+      "step": 4494
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.830319902360955e-06,
+      "loss": 3.6112,
+      "step": 4495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.828312638825279e-06,
+      "loss": 3.4589,
+      "step": 4496
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.826305402991756e-06,
+      "loss": 3.5866,
+      "step": 4497
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.824298195184254e-06,
+      "loss": 3.5401,
+      "step": 4498
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8222910157266455e-06,
+      "loss": 3.5395,
+      "step": 4499
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.820283864942788e-06,
+      "loss": 3.5095,
+      "step": 4500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.818276743156545e-06,
+      "loss": 3.5224,
+      "step": 4501
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8162696506917665e-06,
+      "loss": 3.5499,
+      "step": 4502
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.814262587872304e-06,
+      "loss": 3.5237,
+      "step": 4503
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.8122555550219995e-06,
+      "loss": 3.53,
+      "step": 4504
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.810248552464695e-06,
+      "loss": 3.4992,
+      "step": 4505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.808241580524221e-06,
+      "loss": 3.6355,
+      "step": 4506
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.806234639524412e-06,
+      "loss": 3.494,
+      "step": 4507
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.804227729789088e-06,
+      "loss": 3.5001,
+      "step": 4508
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.802220851642073e-06,
+      "loss": 3.637,
+      "step": 4509
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.800214005407178e-06,
+      "loss": 3.6263,
+      "step": 4510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7982071914082136e-06,
+      "loss": 3.5868,
+      "step": 4511
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.796200409968983e-06,
+      "loss": 3.5342,
+      "step": 4512
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.794193661413286e-06,
+      "loss": 3.5651,
+      "step": 4513
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.792186946064914e-06,
+      "loss": 3.5638,
+      "step": 4514
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.79018026424766e-06,
+      "loss": 3.5952,
+      "step": 4515
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7881736162853e-06,
+      "loss": 3.5793,
+      "step": 4516
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7861670025016175e-06,
+      "loss": 3.5114,
+      "step": 4517
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.784160423220379e-06,
+      "loss": 3.6302,
+      "step": 4518
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.782153878765356e-06,
+      "loss": 3.5135,
+      "step": 4519
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.780147369460304e-06,
+      "loss": 3.6286,
+      "step": 4520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.778140895628982e-06,
+      "loss": 3.5614,
+      "step": 4521
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.776134457595135e-06,
+      "loss": 3.6312,
+      "step": 4522
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7741280556825124e-06,
+      "loss": 3.5596,
+      "step": 4523
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.772121690214844e-06,
+      "loss": 3.4885,
+      "step": 4524
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.770115361515869e-06,
+      "loss": 3.4349,
+      "step": 4525
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7681090699093076e-06,
+      "loss": 3.596,
+      "step": 4526
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.766102815718883e-06,
+      "loss": 3.6092,
+      "step": 4527
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.764096599268308e-06,
+      "loss": 3.6094,
+      "step": 4528
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.762090420881289e-06,
+      "loss": 3.5092,
+      "step": 4529
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.7600842808815285e-06,
+      "loss": 3.5872,
+      "step": 4530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.758078179592723e-06,
+      "loss": 3.5705,
+      "step": 4531
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.756072117338558e-06,
+      "loss": 3.5444,
+      "step": 4532
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.754066094442721e-06,
+      "loss": 3.4845,
+      "step": 4533
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.752060111228883e-06,
+      "loss": 3.6475,
+      "step": 4534
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.75005416802072e-06,
+      "loss": 3.5179,
+      "step": 4535
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.74804826514189e-06,
+      "loss": 3.4793,
+      "step": 4536
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.746042402916055e-06,
+      "loss": 3.7066,
+      "step": 4537
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.744036581666861e-06,
+      "loss": 3.482,
+      "step": 4538
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.742030801717956e-06,
+      "loss": 3.4902,
+      "step": 4539
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.740025063392973e-06,
+      "loss": 3.6157,
+      "step": 4540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.738019367015546e-06,
+      "loss": 3.5824,
+      "step": 4541
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7360137129092965e-06,
+      "loss": 3.5438,
+      "step": 4542
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.734008101397844e-06,
+      "loss": 3.4657,
+      "step": 4543
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.732002532804795e-06,
+      "loss": 3.7431,
+      "step": 4544
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.729997007453756e-06,
+      "loss": 3.5385,
+      "step": 4545
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.727991525668321e-06,
+      "loss": 3.5938,
+      "step": 4546
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.72598608777208e-06,
+      "loss": 3.5878,
+      "step": 4547
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.723980694088614e-06,
+      "loss": 3.3937,
+      "step": 4548
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.721975344941499e-06,
+      "loss": 3.5423,
+      "step": 4549
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.719970040654301e-06,
+      "loss": 3.5161,
+      "step": 4550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7179647815505846e-06,
+      "loss": 3.5479,
+      "step": 4551
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.715959567953897e-06,
+      "loss": 3.4912,
+      "step": 4552
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.7139544001877904e-06,
+      "loss": 3.6008,
+      "step": 4553
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.711949278575797e-06,
+      "loss": 3.3781,
+      "step": 4554
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.709944203441452e-06,
+      "loss": 3.5314,
+      "step": 4555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.707939175108278e-06,
+      "loss": 3.4733,
+      "step": 4556
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.705934193899791e-06,
+      "loss": 3.447,
+      "step": 4557
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.703929260139497e-06,
+      "loss": 3.5269,
+      "step": 4558
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.701924374150901e-06,
+      "loss": 3.6284,
+      "step": 4559
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6999195362574915e-06,
+      "loss": 3.5209,
+      "step": 4560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.697914746782759e-06,
+      "loss": 3.4882,
+      "step": 4561
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.695910006050174e-06,
+      "loss": 3.4808,
+      "step": 4562
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.693905314383213e-06,
+      "loss": 3.6074,
+      "step": 4563
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.691900672105333e-06,
+      "loss": 3.5808,
+      "step": 4564
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.689896079539988e-06,
+      "loss": 3.4785,
+      "step": 4565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.687891537010626e-06,
+      "loss": 3.5444,
+      "step": 4566
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.685887044840682e-06,
+      "loss": 3.5712,
+      "step": 4567
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.683882603353587e-06,
+      "loss": 3.5198,
+      "step": 4568
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6818782128727634e-06,
+      "loss": 3.4938,
+      "step": 4569
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.679873873721621e-06,
+      "loss": 3.4791,
+      "step": 4570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.677869586223568e-06,
+      "loss": 3.5348,
+      "step": 4571
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.675865350701996e-06,
+      "loss": 3.622,
+      "step": 4572
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.673861167480299e-06,
+      "loss": 3.496,
+      "step": 4573
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.671857036881851e-06,
+      "loss": 3.549,
+      "step": 4574
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.669852959230029e-06,
+      "loss": 3.6508,
+      "step": 4575
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.667848934848189e-06,
+      "loss": 3.5277,
+      "step": 4576
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.665844964059691e-06,
+      "loss": 3.4568,
+      "step": 4577
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.663841047187875e-06,
+      "loss": 3.5798,
+      "step": 4578
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6618371845560816e-06,
+      "loss": 3.588,
+      "step": 4579
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.659833376487636e-06,
+      "loss": 3.5869,
+      "step": 4580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.65782962330586e-06,
+      "loss": 3.4652,
+      "step": 4581
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.655825925334062e-06,
+      "loss": 3.5652,
+      "step": 4582
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.653822282895543e-06,
+      "loss": 3.5698,
+      "step": 4583
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6518186963135955e-06,
+      "loss": 3.5973,
+      "step": 4584
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.649815165911503e-06,
+      "loss": 3.5811,
+      "step": 4585
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.647811692012541e-06,
+      "loss": 3.5166,
+      "step": 4586
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6458082749399734e-06,
+      "loss": 3.585,
+      "step": 4587
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.643804915017055e-06,
+      "loss": 3.5756,
+      "step": 4588
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.641801612567035e-06,
+      "loss": 3.4204,
+      "step": 4589
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.639798367913148e-06,
+      "loss": 3.41,
+      "step": 4590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6377951813786255e-06,
+      "loss": 3.5261,
+      "step": 4591
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.635792053286682e-06,
+      "loss": 3.5253,
+      "step": 4592
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.633788983960533e-06,
+      "loss": 3.5492,
+      "step": 4593
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.631785973723371e-06,
+      "loss": 3.4826,
+      "step": 4594
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.629783022898393e-06,
+      "loss": 3.5653,
+      "step": 4595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.627780131808775e-06,
+      "loss": 3.459,
+      "step": 4596
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6257773007776915e-06,
+      "loss": 3.6493,
+      "step": 4597
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6237745301283e-06,
+      "loss": 3.4518,
+      "step": 4598
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.621771820183757e-06,
+      "loss": 3.4646,
+      "step": 4599
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6197691712672e-06,
+      "loss": 3.5039,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.617766583701764e-06,
+      "loss": 3.4793,
+      "step": 4601
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.615764057810569e-06,
+      "loss": 3.5834,
+      "step": 4602
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.613761593916728e-06,
+      "loss": 3.5339,
+      "step": 4603
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.611759192343344e-06,
+      "loss": 3.5692,
+      "step": 4604
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.609756853413508e-06,
+      "loss": 3.4836,
+      "step": 4605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.607754577450301e-06,
+      "loss": 3.5574,
+      "step": 4606
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.605752364776799e-06,
+      "loss": 3.5912,
+      "step": 4607
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.603750215716057e-06,
+      "loss": 3.4849,
+      "step": 4608
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.6017481305911335e-06,
+      "loss": 3.5559,
+      "step": 4609
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.599746109725064e-06,
+      "loss": 3.6343,
+      "step": 4610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.597744153440884e-06,
+      "loss": 3.6438,
+      "step": 4611
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.595742262061607e-06,
+      "loss": 3.5654,
+      "step": 4612
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.593740435910251e-06,
+      "loss": 3.5184,
+      "step": 4613
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5917386753098085e-06,
+      "loss": 3.6962,
+      "step": 4614
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.589736980583272e-06,
+      "loss": 3.5884,
+      "step": 4615
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.587735352053618e-06,
+      "loss": 3.5943,
+      "step": 4616
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.585733790043815e-06,
+      "loss": 3.625,
+      "step": 4617
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.583732294876818e-06,
+      "loss": 3.405,
+      "step": 4618
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.581730866875575e-06,
+      "loss": 3.5668,
+      "step": 4619
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5797295063630185e-06,
+      "loss": 3.6616,
+      "step": 4620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.577728213662075e-06,
+      "loss": 3.587,
+      "step": 4621
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.575726989095656e-06,
+      "loss": 3.5651,
+      "step": 4622
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.573725832986665e-06,
+      "loss": 3.5943,
+      "step": 4623
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5717247456579905e-06,
+      "loss": 3.5911,
+      "step": 4624
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.569723727432517e-06,
+      "loss": 3.5889,
+      "step": 4625
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.567722778633108e-06,
+      "loss": 3.5963,
+      "step": 4626
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.565721899582627e-06,
+      "loss": 3.548,
+      "step": 4627
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.563721090603915e-06,
+      "loss": 3.5654,
+      "step": 4628
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.561720352019811e-06,
+      "loss": 3.6093,
+      "step": 4629
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.559719684153137e-06,
+      "loss": 3.5715,
+      "step": 4630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.557719087326707e-06,
+      "loss": 3.416,
+      "step": 4631
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.555718561863319e-06,
+      "loss": 3.4792,
+      "step": 4632
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5537181080857665e-06,
+      "loss": 3.6376,
+      "step": 4633
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.551717726316822e-06,
+      "loss": 3.4889,
+      "step": 4634
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.549717416879257e-06,
+      "loss": 3.6655,
+      "step": 4635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.547717180095821e-06,
+      "loss": 3.6622,
+      "step": 4636
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.545717016289262e-06,
+      "loss": 3.4472,
+      "step": 4637
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5437169257823074e-06,
+      "loss": 3.5312,
+      "step": 4638
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.541716908897678e-06,
+      "loss": 3.5136,
+      "step": 4639
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.53971696595808e-06,
+      "loss": 3.5748,
+      "step": 4640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.537717097286211e-06,
+      "loss": 3.5459,
+      "step": 4641
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.535717303204752e-06,
+      "loss": 3.5573,
+      "step": 4642
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.533717584036375e-06,
+      "loss": 3.45,
+      "step": 4643
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.53171794010374e-06,
+      "loss": 3.5444,
+      "step": 4644
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5297183717294926e-06,
+      "loss": 3.4542,
+      "step": 4645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.52771887923627e-06,
+      "loss": 3.4451,
+      "step": 4646
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5257194629466925e-06,
+      "loss": 3.5673,
+      "step": 4647
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.523720123183373e-06,
+      "loss": 3.5112,
+      "step": 4648
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.521720860268907e-06,
+      "loss": 3.4919,
+      "step": 4649
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5197216745258835e-06,
+      "loss": 3.601,
+      "step": 4650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.517722566276871e-06,
+      "loss": 3.57,
+      "step": 4651
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.515723535844436e-06,
+      "loss": 3.5226,
+      "step": 4652
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5137245835511216e-06,
+      "loss": 3.4595,
+      "step": 4653
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.511725709719467e-06,
+      "loss": 3.4876,
+      "step": 4654
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.509726914671991e-06,
+      "loss": 3.5873,
+      "step": 4655
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.507728198731209e-06,
+      "loss": 3.5437,
+      "step": 4656
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.505729562219614e-06,
+      "loss": 3.6326,
+      "step": 4657
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.5037310054596936e-06,
+      "loss": 3.4592,
+      "step": 4658
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.501732528773917e-06,
+      "loss": 3.5797,
+      "step": 4659
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.499734132484745e-06,
+      "loss": 3.5996,
+      "step": 4660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.497735816914622e-06,
+      "loss": 3.552,
+      "step": 4661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.495737582385982e-06,
+      "loss": 3.481,
+      "step": 4662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.493739429221245e-06,
+      "loss": 3.5077,
+      "step": 4663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.491741357742816e-06,
+      "loss": 3.6232,
+      "step": 4664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.489743368273089e-06,
+      "loss": 3.587,
+      "step": 4665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.487745461134446e-06,
+      "loss": 3.4346,
+      "step": 4666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.48574763664925e-06,
+      "loss": 3.5312,
+      "step": 4667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.483749895139859e-06,
+      "loss": 3.6646,
+      "step": 4668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.481752236928609e-06,
+      "loss": 3.6305,
+      "step": 4669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.479754662337832e-06,
+      "loss": 3.4336,
+      "step": 4670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.477757171689836e-06,
+      "loss": 3.5565,
+      "step": 4671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.475759765306924e-06,
+      "loss": 3.5761,
+      "step": 4672
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.47376244351138e-06,
+      "loss": 3.4642,
+      "step": 4673
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.471765206625479e-06,
+      "loss": 3.5184,
+      "step": 4674
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.469768054971476e-06,
+      "loss": 3.5316,
+      "step": 4675
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4677709888716215e-06,
+      "loss": 3.6873,
+      "step": 4676
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.465774008648141e-06,
+      "loss": 3.5354,
+      "step": 4677
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4637771146232555e-06,
+      "loss": 3.557,
+      "step": 4678
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4617803071191666e-06,
+      "loss": 3.6399,
+      "step": 4679
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.459783586458065e-06,
+      "loss": 3.5542,
+      "step": 4680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.457786952962126e-06,
+      "loss": 3.5333,
+      "step": 4681
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.455790406953512e-06,
+      "loss": 3.5654,
+      "step": 4682
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.453793948754365e-06,
+      "loss": 3.5726,
+      "step": 4683
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.451797578686826e-06,
+      "loss": 3.5086,
+      "step": 4684
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.449801297073007e-06,
+      "loss": 3.676,
+      "step": 4685
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4478051042350185e-06,
+      "loss": 3.5615,
+      "step": 4686
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.445809000494945e-06,
+      "loss": 3.559,
+      "step": 4687
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.443812986174868e-06,
+      "loss": 3.5485,
+      "step": 4688
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.441817061596845e-06,
+      "loss": 3.4794,
+      "step": 4689
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.439821227082926e-06,
+      "loss": 3.5472,
+      "step": 4690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.43782548295514e-06,
+      "loss": 3.5053,
+      "step": 4691
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.435829829535508e-06,
+      "loss": 3.5735,
+      "step": 4692
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.433834267146031e-06,
+      "loss": 3.4588,
+      "step": 4693
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.431838796108701e-06,
+      "loss": 3.5809,
+      "step": 4694
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4298434167454875e-06,
+      "loss": 3.6363,
+      "step": 4695
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4278481293783524e-06,
+      "loss": 3.5321,
+      "step": 4696
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.425852934329239e-06,
+      "loss": 3.5496,
+      "step": 4697
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.423857831920077e-06,
+      "loss": 3.4603,
+      "step": 4698
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.42186282247278e-06,
+      "loss": 3.6329,
+      "step": 4699
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.41986790630925e-06,
+      "loss": 3.4685,
+      "step": 4700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.417873083751365e-06,
+      "loss": 3.5523,
+      "step": 4701
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.415878355121003e-06,
+      "loss": 3.5872,
+      "step": 4702
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.41388372074001e-06,
+      "loss": 3.4581,
+      "step": 4703
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.41188918093023e-06,
+      "loss": 3.5044,
+      "step": 4704
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.409894736013483e-06,
+      "loss": 3.539,
+      "step": 4705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.407900386311582e-06,
+      "loss": 3.4424,
+      "step": 4706
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.405906132146315e-06,
+      "loss": 3.5637,
+      "step": 4707
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4039119738394634e-06,
+      "loss": 3.5636,
+      "step": 4708
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.4019179117127854e-06,
+      "loss": 3.3342,
+      "step": 4709
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3999239460880315e-06,
+      "loss": 3.5405,
+      "step": 4710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.397930077286929e-06,
+      "loss": 3.4709,
+      "step": 4711
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.395936305631197e-06,
+      "loss": 3.5501,
+      "step": 4712
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.393942631442531e-06,
+      "loss": 3.6874,
+      "step": 4713
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.391949055042618e-06,
+      "loss": 3.6232,
+      "step": 4714
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.389955576753125e-06,
+      "loss": 3.4124,
+      "step": 4715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.387962196895704e-06,
+      "loss": 3.4409,
+      "step": 4716
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.385968915791993e-06,
+      "loss": 3.4666,
+      "step": 4717
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.383975733763611e-06,
+      "loss": 3.5106,
+      "step": 4718
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.381982651132162e-06,
+      "loss": 3.4831,
+      "step": 4719
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3799896682192365e-06,
+      "loss": 3.4483,
+      "step": 4720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.377996785346404e-06,
+      "loss": 3.5949,
+      "step": 4721
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.376004002835225e-06,
+      "loss": 3.536,
+      "step": 4722
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.374011321007234e-06,
+      "loss": 3.5419,
+      "step": 4723
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.372018740183961e-06,
+      "loss": 3.6373,
+      "step": 4724
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.370026260686908e-06,
+      "loss": 3.5365,
+      "step": 4725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.36803388283757e-06,
+      "loss": 3.5183,
+      "step": 4726
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.366041606957419e-06,
+      "loss": 3.4409,
+      "step": 4727
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3640494333679165e-06,
+      "loss": 3.5293,
+      "step": 4728
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.362057362390499e-06,
+      "loss": 3.3781,
+      "step": 4729
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.360065394346599e-06,
+      "loss": 3.6102,
+      "step": 4730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.358073529557618e-06,
+      "loss": 3.4884,
+      "step": 4731
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.356081768344953e-06,
+      "loss": 3.5215,
+      "step": 4732
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.354090111029977e-06,
+      "loss": 3.5473,
+      "step": 4733
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3520985579340495e-06,
+      "loss": 3.345,
+      "step": 4734
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3501071093785106e-06,
+      "loss": 3.4715,
+      "step": 4735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3481157656846875e-06,
+      "loss": 3.4009,
+      "step": 4736
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.346124527173884e-06,
+      "loss": 3.5364,
+      "step": 4737
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.344133394167397e-06,
+      "loss": 3.5018,
+      "step": 4738
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3421423669864944e-06,
+      "loss": 3.4575,
+      "step": 4739
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.340151445952438e-06,
+      "loss": 3.4983,
+      "step": 4740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3381606313864624e-06,
+      "loss": 3.413,
+      "step": 4741
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.336169923609796e-06,
+      "loss": 3.5481,
+      "step": 4742
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.334179322943639e-06,
+      "loss": 3.5038,
+      "step": 4743
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.332188829709183e-06,
+      "loss": 3.5491,
+      "step": 4744
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3301984442275956e-06,
+      "loss": 3.5972,
+      "step": 4745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.3282081668200334e-06,
+      "loss": 3.6689,
+      "step": 4746
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.326217997807628e-06,
+      "loss": 3.5205,
+      "step": 4747
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.324227937511503e-06,
+      "loss": 3.5619,
+      "step": 4748
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.322237986252753e-06,
+      "loss": 3.4615,
+      "step": 4749
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.320248144352468e-06,
+      "loss": 3.5856,
+      "step": 4750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.318258412131709e-06,
+      "loss": 3.4996,
+      "step": 4751
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.316268789911524e-06,
+      "loss": 3.5756,
+      "step": 4752
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.314279278012945e-06,
+      "loss": 3.5558,
+      "step": 4753
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.312289876756985e-06,
+      "loss": 3.5375,
+      "step": 4754
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.310300586464634e-06,
+      "loss": 3.5517,
+      "step": 4755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.308311407456875e-06,
+      "loss": 3.5775,
+      "step": 4756
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.30632234005466e-06,
+      "loss": 3.5399,
+      "step": 4757
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.304333384578935e-06,
+      "loss": 3.5323,
+      "step": 4758
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.302344541350617e-06,
+      "loss": 3.6174,
+      "step": 4759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.300355810690618e-06,
+      "loss": 3.4322,
+      "step": 4760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.298367192919817e-06,
+      "loss": 3.566,
+      "step": 4761
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.296378688359087e-06,
+      "loss": 3.5782,
+      "step": 4762
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.294390297329274e-06,
+      "loss": 3.4763,
+      "step": 4763
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.292402020151213e-06,
+      "loss": 3.509,
+      "step": 4764
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2904138571457145e-06,
+      "loss": 3.551,
+      "step": 4765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2884258086335755e-06,
+      "loss": 3.3935,
+      "step": 4766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.286437874935568e-06,
+      "loss": 3.6515,
+      "step": 4767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.284450056372456e-06,
+      "loss": 3.5366,
+      "step": 4768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.282462353264973e-06,
+      "loss": 3.5547,
+      "step": 4769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2804747659338426e-06,
+      "loss": 3.5584,
+      "step": 4770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.278487294699765e-06,
+      "loss": 3.5012,
+      "step": 4771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.276499939883425e-06,
+      "loss": 3.5262,
+      "step": 4772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.274512701805483e-06,
+      "loss": 3.5086,
+      "step": 4773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.27252558078659e-06,
+      "loss": 3.6545,
+      "step": 4774
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.270538577147367e-06,
+      "loss": 3.5509,
+      "step": 4775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.268551691208427e-06,
+      "loss": 3.5093,
+      "step": 4776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.266564923290354e-06,
+      "loss": 3.5126,
+      "step": 4777
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.264578273713721e-06,
+      "loss": 3.4028,
+      "step": 4778
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.262591742799076e-06,
+      "loss": 3.5701,
+      "step": 4779
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.260605330866953e-06,
+      "loss": 3.5053,
+      "step": 4780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2586190382378596e-06,
+      "loss": 3.4949,
+      "step": 4781
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.256632865232294e-06,
+      "loss": 3.6138,
+      "step": 4782
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.254646812170725e-06,
+      "loss": 3.5297,
+      "step": 4783
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.252660879373612e-06,
+      "loss": 3.572,
+      "step": 4784
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2506750671613844e-06,
+      "loss": 3.4762,
+      "step": 4785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.248689375854463e-06,
+      "loss": 3.4982,
+      "step": 4786
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.246703805773239e-06,
+      "loss": 3.5248,
+      "step": 4787
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.244718357238091e-06,
+      "loss": 3.6377,
+      "step": 4788
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2427330305693745e-06,
+      "loss": 3.5963,
+      "step": 4789
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2407478260874294e-06,
+      "loss": 3.5041,
+      "step": 4790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.238762744112569e-06,
+      "loss": 3.4743,
+      "step": 4791
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.236777784965094e-06,
+      "loss": 3.5305,
+      "step": 4792
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.234792948965279e-06,
+      "loss": 3.6695,
+      "step": 4793
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.232808236433386e-06,
+      "loss": 3.5136,
+      "step": 4794
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2308236476896484e-06,
+      "loss": 3.5127,
+      "step": 4795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2288391830542885e-06,
+      "loss": 3.5943,
+      "step": 4796
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2268548428475e-06,
+      "loss": 3.5686,
+      "step": 4797
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.224870627389466e-06,
+      "loss": 3.5567,
+      "step": 4798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.222886537000338e-06,
+      "loss": 3.5404,
+      "step": 4799
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.220902572000259e-06,
+      "loss": 3.5079,
+      "step": 4800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.218918732709342e-06,
+      "loss": 3.6193,
+      "step": 4801
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.2169350194476886e-06,
+      "loss": 3.5878,
+      "step": 4802
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.21495143253537e-06,
+      "loss": 3.5174,
+      "step": 4803
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.212967972292449e-06,
+      "loss": 3.5345,
+      "step": 4804
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.210984639038956e-06,
+      "loss": 3.5732,
+      "step": 4805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.209001433094909e-06,
+      "loss": 3.6138,
+      "step": 4806
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.207018354780301e-06,
+      "loss": 3.6411,
+      "step": 4807
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.205035404415109e-06,
+      "loss": 3.6593,
+      "step": 4808
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.203052582319284e-06,
+      "loss": 3.4958,
+      "step": 4809
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.201069888812761e-06,
+      "loss": 3.5615,
+      "step": 4810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.199087324215449e-06,
+      "loss": 3.5644,
+      "step": 4811
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.197104888847245e-06,
+      "loss": 3.4171,
+      "step": 4812
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.195122583028012e-06,
+      "loss": 3.6341,
+      "step": 4813
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.193140407077607e-06,
+      "loss": 3.4878,
+      "step": 4814
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.191158361315853e-06,
+      "loss": 3.6352,
+      "step": 4815
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.189176446062562e-06,
+      "loss": 3.5301,
+      "step": 4816
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1871946616375155e-06,
+      "loss": 3.5636,
+      "step": 4817
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.185213008360486e-06,
+      "loss": 3.5964,
+      "step": 4818
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.18323148655121e-06,
+      "loss": 3.5833,
+      "step": 4819
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.181250096529418e-06,
+      "loss": 3.5268,
+      "step": 4820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.179268838614805e-06,
+      "loss": 3.4939,
+      "step": 4821
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.177287713127057e-06,
+      "loss": 3.523,
+      "step": 4822
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.175306720385831e-06,
+      "loss": 3.5632,
+      "step": 4823
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.173325860710764e-06,
+      "loss": 3.5379,
+      "step": 4824
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.171345134421472e-06,
+      "loss": 3.6,
+      "step": 4825
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.169364541837552e-06,
+      "loss": 3.4898,
+      "step": 4826
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.167384083278576e-06,
+      "loss": 3.5713,
+      "step": 4827
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1654037590640956e-06,
+      "loss": 3.4362,
+      "step": 4828
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.163423569513639e-06,
+      "loss": 3.5797,
+      "step": 4829
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1614435149467175e-06,
+      "loss": 3.5229,
+      "step": 4830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1594635956828135e-06,
+      "loss": 3.5639,
+      "step": 4831
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.157483812041397e-06,
+      "loss": 3.5303,
+      "step": 4832
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1555041643419046e-06,
+      "loss": 3.5383,
+      "step": 4833
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.153524652903762e-06,
+      "loss": 3.5969,
+      "step": 4834
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.151545278046364e-06,
+      "loss": 3.5649,
+      "step": 4835
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.149566040089092e-06,
+      "loss": 3.5963,
+      "step": 4836
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.147586939351294e-06,
+      "loss": 3.5064,
+      "step": 4837
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.145607976152311e-06,
+      "loss": 3.4787,
+      "step": 4838
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.143629150811444e-06,
+      "loss": 3.4739,
+      "step": 4839
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.141650463647988e-06,
+      "loss": 3.6036,
+      "step": 4840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.139671914981205e-06,
+      "loss": 3.5474,
+      "step": 4841
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.13769350513034e-06,
+      "loss": 3.5916,
+      "step": 4842
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.135715234414614e-06,
+      "loss": 3.503,
+      "step": 4843
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.133737103153224e-06,
+      "loss": 3.5638,
+      "step": 4844
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.131759111665349e-06,
+      "loss": 3.5592,
+      "step": 4845
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.12978126027014e-06,
+      "loss": 3.5486,
+      "step": 4846
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.127803549286727e-06,
+      "loss": 3.6392,
+      "step": 4847
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1258259790342225e-06,
+      "loss": 3.6245,
+      "step": 4848
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.123848549831707e-06,
+      "loss": 3.4586,
+      "step": 4849
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.121871261998247e-06,
+      "loss": 3.4909,
+      "step": 4850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.119894115852878e-06,
+      "loss": 3.5063,
+      "step": 4851
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.117917111714624e-06,
+      "loss": 3.5648,
+      "step": 4852
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1159402499024726e-06,
+      "loss": 3.4743,
+      "step": 4853
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1139635307353995e-06,
+      "loss": 3.4497,
+      "step": 4854
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.111986954532349e-06,
+      "loss": 3.6376,
+      "step": 4855
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.11001052161225e-06,
+      "loss": 3.4694,
+      "step": 4856
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.108034232294e-06,
+      "loss": 3.567,
+      "step": 4857
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.106058086896484e-06,
+      "loss": 3.5866,
+      "step": 4858
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.104082085738553e-06,
+      "loss": 3.5507,
+      "step": 4859
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.10210622913904e-06,
+      "loss": 3.5112,
+      "step": 4860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.100130517416754e-06,
+      "loss": 3.6867,
+      "step": 4861
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.098154950890483e-06,
+      "loss": 3.5957,
+      "step": 4862
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.096179529878985e-06,
+      "loss": 3.5694,
+      "step": 4863
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.094204254701004e-06,
+      "loss": 3.5888,
+      "step": 4864
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.09222912567525e-06,
+      "loss": 3.5692,
+      "step": 4865
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.090254143120418e-06,
+      "loss": 3.4999,
+      "step": 4866
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0882793073551735e-06,
+      "loss": 3.4958,
+      "step": 4867
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.086304618698164e-06,
+      "loss": 3.5637,
+      "step": 4868
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.084330077468006e-06,
+      "loss": 3.5322,
+      "step": 4869
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0823556839833014e-06,
+      "loss": 3.5682,
+      "step": 4870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.080381438562618e-06,
+      "loss": 3.4936,
+      "step": 4871
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.078407341524508e-06,
+      "loss": 3.5303,
+      "step": 4872
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.076433393187495e-06,
+      "loss": 3.4543,
+      "step": 4873
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.074459593870082e-06,
+      "loss": 3.5534,
+      "step": 4874
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.072485943890743e-06,
+      "loss": 3.4991,
+      "step": 4875
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.070512443567935e-06,
+      "loss": 3.5265,
+      "step": 4876
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.068539093220083e-06,
+      "loss": 3.514,
+      "step": 4877
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.066565893165594e-06,
+      "loss": 3.5007,
+      "step": 4878
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.064592843722847e-06,
+      "loss": 3.5545,
+      "step": 4879
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.062619945210199e-06,
+      "loss": 3.5756,
+      "step": 4880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.060647197945981e-06,
+      "loss": 3.6181,
+      "step": 4881
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.058674602248502e-06,
+      "loss": 3.4837,
+      "step": 4882
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.056702158436042e-06,
+      "loss": 3.5533,
+      "step": 4883
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.054729866826862e-06,
+      "loss": 3.5816,
+      "step": 4884
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.052757727739193e-06,
+      "loss": 3.55,
+      "step": 4885
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.050785741491248e-06,
+      "loss": 3.5647,
+      "step": 4886
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.048813908401207e-06,
+      "loss": 3.5815,
+      "step": 4887
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.046842228787234e-06,
+      "loss": 3.6415,
+      "step": 4888
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.044870702967461e-06,
+      "loss": 3.4412,
+      "step": 4889
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.042899331260001e-06,
+      "loss": 3.5427,
+      "step": 4890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0409281139829355e-06,
+      "loss": 3.5913,
+      "step": 4891
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.038957051454331e-06,
+      "loss": 3.5382,
+      "step": 4892
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.036986143992215e-06,
+      "loss": 3.4628,
+      "step": 4893
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.035015391914607e-06,
+      "loss": 3.5441,
+      "step": 4894
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.033044795539484e-06,
+      "loss": 3.5709,
+      "step": 4895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.03107435518481e-06,
+      "loss": 3.3794,
+      "step": 4896
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.029104071168522e-06,
+      "loss": 3.5822,
+      "step": 4897
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.027133943808526e-06,
+      "loss": 3.5704,
+      "step": 4898
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.025163973422709e-06,
+      "loss": 3.4819,
+      "step": 4899
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.023194160328931e-06,
+      "loss": 3.5358,
+      "step": 4900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.021224504845022e-06,
+      "loss": 3.5408,
+      "step": 4901
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.019255007288795e-06,
+      "loss": 3.637,
+      "step": 4902
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.017285667978028e-06,
+      "loss": 3.5913,
+      "step": 4903
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.015316487230485e-06,
+      "loss": 3.5719,
+      "step": 4904
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.013347465363891e-06,
+      "loss": 3.6088,
+      "step": 4905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.011378602695959e-06,
+      "loss": 3.4928,
+      "step": 4906
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.009409899544363e-06,
+      "loss": 3.5316,
+      "step": 4907
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.007441356226763e-06,
+      "loss": 3.5011,
+      "step": 4908
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.005472973060783e-06,
+      "loss": 3.4829,
+      "step": 4909
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.003504750364032e-06,
+      "loss": 3.6001,
+      "step": 4910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0015366884540815e-06,
+      "loss": 3.5362,
+      "step": 4911
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.999568787648488e-06,
+      "loss": 3.6266,
+      "step": 4912
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.997601048264772e-06,
+      "loss": 3.5679,
+      "step": 4913
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.995633470620437e-06,
+      "loss": 3.5274,
+      "step": 4914
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.993666055032954e-06,
+      "loss": 3.5278,
+      "step": 4915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.991698801819769e-06,
+      "loss": 3.6429,
+      "step": 4916
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.989731711298306e-06,
+      "loss": 3.4024,
+      "step": 4917
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.987764783785958e-06,
+      "loss": 3.559,
+      "step": 4918
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.98579801960009e-06,
+      "loss": 3.5006,
+      "step": 4919
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.983831419058049e-06,
+      "loss": 3.527,
+      "step": 4920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.981864982477148e-06,
+      "loss": 3.521,
+      "step": 4921
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.979898710174678e-06,
+      "loss": 3.5043,
+      "step": 4922
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9779326024678976e-06,
+      "loss": 3.4628,
+      "step": 4923
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.975966659674048e-06,
+      "loss": 3.5442,
+      "step": 4924
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.974000882110333e-06,
+      "loss": 3.5528,
+      "step": 4925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9720352700939415e-06,
+      "loss": 3.5759,
+      "step": 4926
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.970069823942023e-06,
+      "loss": 3.5297,
+      "step": 4927
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.968104543971712e-06,
+      "loss": 3.5326,
+      "step": 4928
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.966139430500109e-06,
+      "loss": 3.6121,
+      "step": 4929
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.964174483844289e-06,
+      "loss": 3.5391,
+      "step": 4930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.962209704321301e-06,
+      "loss": 3.5555,
+      "step": 4931
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.960245092248168e-06,
+      "loss": 3.5733,
+      "step": 4932
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.958280647941883e-06,
+      "loss": 3.5661,
+      "step": 4933
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.956316371719415e-06,
+      "loss": 3.4671,
+      "step": 4934
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.954352263897702e-06,
+      "loss": 3.579,
+      "step": 4935
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.952388324793662e-06,
+      "loss": 3.5444,
+      "step": 4936
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.950424554724175e-06,
+      "loss": 3.6084,
+      "step": 4937
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.948460954006106e-06,
+      "loss": 3.5534,
+      "step": 4938
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.94649752295628e-06,
+      "loss": 3.5618,
+      "step": 4939
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.944534261891508e-06,
+      "loss": 3.5746,
+      "step": 4940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.94257117112856e-06,
+      "loss": 3.5305,
+      "step": 4941
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.940608250984191e-06,
+      "loss": 3.426,
+      "step": 4942
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.938645501775119e-06,
+      "loss": 3.5944,
+      "step": 4943
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.936682923818041e-06,
+      "loss": 3.4859,
+      "step": 4944
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.934720517429619e-06,
+      "loss": 3.5999,
+      "step": 4945
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9327582829264964e-06,
+      "loss": 3.5366,
+      "step": 4946
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9307962206252805e-06,
+      "loss": 3.6224,
+      "step": 4947
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.928834330842559e-06,
+      "loss": 3.5243,
+      "step": 4948
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.926872613894883e-06,
+      "loss": 3.414,
+      "step": 4949
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9249110700987836e-06,
+      "loss": 3.6164,
+      "step": 4950
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.922949699770758e-06,
+      "loss": 3.504,
+      "step": 4951
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.920988503227277e-06,
+      "loss": 3.3431,
+      "step": 4952
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.919027480784788e-06,
+      "loss": 3.5019,
+      "step": 4953
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.917066632759703e-06,
+      "loss": 3.6152,
+      "step": 4954
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.91510595946841e-06,
+      "loss": 3.5484,
+      "step": 4955
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.91314546122727e-06,
+      "loss": 3.5063,
+      "step": 4956
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9111851383526105e-06,
+      "loss": 3.5301,
+      "step": 4957
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.909224991160739e-06,
+      "loss": 3.5316,
+      "step": 4958
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.907265019967924e-06,
+      "loss": 3.5143,
+      "step": 4959
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.905305225090417e-06,
+      "loss": 3.5031,
+      "step": 4960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.90334560684443e-06,
+      "loss": 3.5238,
+      "step": 4961
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.901386165546157e-06,
+      "loss": 3.5415,
+      "step": 4962
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.899426901511754e-06,
+      "loss": 3.582,
+      "step": 4963
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.897467815057356e-06,
+      "loss": 3.4956,
+      "step": 4964
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8955089064990635e-06,
+      "loss": 3.4654,
+      "step": 4965
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.893550176152955e-06,
+      "loss": 3.4946,
+      "step": 4966
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8915916243350695e-06,
+      "loss": 3.5914,
+      "step": 4967
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.889633251361432e-06,
+      "loss": 3.5566,
+      "step": 4968
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.887675057548024e-06,
+      "loss": 3.5145,
+      "step": 4969
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.885717043210807e-06,
+      "loss": 3.5584,
+      "step": 4970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8837592086657105e-06,
+      "loss": 3.6632,
+      "step": 4971
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.881801554228638e-06,
+      "loss": 3.4917,
+      "step": 4972
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.879844080215457e-06,
+      "loss": 3.4924,
+      "step": 4973
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.877886786942016e-06,
+      "loss": 3.4874,
+      "step": 4974
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.875929674724123e-06,
+      "loss": 3.5337,
+      "step": 4975
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.873972743877567e-06,
+      "loss": 3.5979,
+      "step": 4976
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8720159947181e-06,
+      "loss": 3.6067,
+      "step": 4977
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.870059427561452e-06,
+      "loss": 3.5022,
+      "step": 4978
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.868103042723315e-06,
+      "loss": 3.6174,
+      "step": 4979
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8661468405193614e-06,
+      "loss": 3.5435,
+      "step": 4980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.864190821265224e-06,
+      "loss": 3.6045,
+      "step": 4981
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.862234985276514e-06,
+      "loss": 3.6583,
+      "step": 4982
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.860279332868807e-06,
+      "loss": 3.5996,
+      "step": 4983
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.858323864357657e-06,
+      "loss": 3.5836,
+      "step": 4984
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.856368580058578e-06,
+      "loss": 3.5734,
+      "step": 4985
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.854413480287064e-06,
+      "loss": 3.6036,
+      "step": 4986
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.852458565358572e-06,
+      "loss": 3.5992,
+      "step": 4987
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.850503835588533e-06,
+      "loss": 3.5474,
+      "step": 4988
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.848549291292347e-06,
+      "loss": 3.5209,
+      "step": 4989
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8465949327853854e-06,
+      "loss": 3.5789,
+      "step": 4990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.844640760382985e-06,
+      "loss": 3.3162,
+      "step": 4991
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.842686774400461e-06,
+      "loss": 3.4461,
+      "step": 4992
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8407329751530896e-06,
+      "loss": 3.6252,
+      "step": 4993
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.838779362956123e-06,
+      "loss": 3.533,
+      "step": 4994
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.836825938124778e-06,
+      "loss": 3.4908,
+      "step": 4995
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.834872700974249e-06,
+      "loss": 3.4708,
+      "step": 4996
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8329196518196895e-06,
+      "loss": 3.5636,
+      "step": 4997
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8309667909762345e-06,
+      "loss": 3.5426,
+      "step": 4998
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.829014118758979e-06,
+      "loss": 3.4328,
+      "step": 4999
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.827061635482992e-06,
+      "loss": 3.5547,
+      "step": 5000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.82510934146331e-06,
+      "loss": 3.4662,
+      "step": 5001
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.823157237014942e-06,
+      "loss": 3.5678,
+      "step": 5002
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.821205322452863e-06,
+      "loss": 3.5634,
+      "step": 5003
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.819253598092022e-06,
+      "loss": 3.6215,
+      "step": 5004
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.81730206424733e-06,
+      "loss": 3.7004,
+      "step": 5005
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.815350721233674e-06,
+      "loss": 3.4241,
+      "step": 5006
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8133995693659066e-06,
+      "loss": 3.508,
+      "step": 5007
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.811448608958852e-06,
+      "loss": 3.4293,
+      "step": 5008
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8094978403272994e-06,
+      "loss": 3.4896,
+      "step": 5009
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8075472637860135e-06,
+      "loss": 3.4838,
+      "step": 5010
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8055968796497207e-06,
+      "loss": 3.455,
+      "step": 5011
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8036466882331236e-06,
+      "loss": 3.5018,
+      "step": 5012
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.8016966898508862e-06,
+      "loss": 3.5264,
+      "step": 5013
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.799746884817649e-06,
+      "loss": 3.4557,
+      "step": 5014
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.797797273448014e-06,
+      "loss": 3.5503,
+      "step": 5015
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.79584785605656e-06,
+      "loss": 3.5952,
+      "step": 5016
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7938986329578243e-06,
+      "loss": 3.5332,
+      "step": 5017
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.791949604466324e-06,
+      "loss": 3.5187,
+      "step": 5018
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7900007708965346e-06,
+      "loss": 3.4949,
+      "step": 5019
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7880521325629087e-06,
+      "loss": 3.5673,
+      "step": 5020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.786103689779861e-06,
+      "loss": 3.5001,
+      "step": 5021
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7841554428617794e-06,
+      "loss": 3.5103,
+      "step": 5022
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7822073921230164e-06,
+      "loss": 3.5411,
+      "step": 5023
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7802595378778943e-06,
+      "loss": 3.6802,
+      "step": 5024
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7783118804407047e-06,
+      "loss": 3.4597,
+      "step": 5025
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7763644201257065e-06,
+      "loss": 3.5705,
+      "step": 5026
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.774417157247127e-06,
+      "loss": 3.5976,
+      "step": 5027
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7724700921191625e-06,
+      "loss": 3.5078,
+      "step": 5028
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7705232250559724e-06,
+      "loss": 3.4599,
+      "step": 5029
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.768576556371694e-06,
+      "loss": 3.5798,
+      "step": 5030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7666300863804217e-06,
+      "loss": 3.5623,
+      "step": 5031
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7646838153962267e-06,
+      "loss": 3.5618,
+      "step": 5032
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7627377437331407e-06,
+      "loss": 3.5117,
+      "step": 5033
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.760791871705171e-06,
+      "loss": 3.4638,
+      "step": 5034
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.758846199626284e-06,
+      "loss": 3.5935,
+      "step": 5035
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7569007278104235e-06,
+      "loss": 3.5349,
+      "step": 5036
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.754955456571491e-06,
+      "loss": 3.56,
+      "step": 5037
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7530103862233635e-06,
+      "loss": 3.5258,
+      "step": 5038
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7510655170798804e-06,
+      "loss": 3.5397,
+      "step": 5039
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.749120849454855e-06,
+      "loss": 3.5833,
+      "step": 5040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.747176383662058e-06,
+      "loss": 3.5474,
+      "step": 5041
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7452321200152375e-06,
+      "loss": 3.6516,
+      "step": 5042
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7432880588281038e-06,
+      "loss": 3.461,
+      "step": 5043
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.741344200414336e-06,
+      "loss": 3.4415,
+      "step": 5044
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7394005450875793e-06,
+      "loss": 3.6245,
+      "step": 5045
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7374570931614486e-06,
+      "loss": 3.5056,
+      "step": 5046
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.735513844949521e-06,
+      "loss": 3.5525,
+      "step": 5047
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.733570800765348e-06,
+      "loss": 3.5808,
+      "step": 5048
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7316279609224415e-06,
+      "loss": 3.5597,
+      "step": 5049
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7296853257342857e-06,
+      "loss": 3.5143,
+      "step": 5050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7277428955143258e-06,
+      "loss": 3.5774,
+      "step": 5051
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7258006705759808e-06,
+      "loss": 3.4836,
+      "step": 5052
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.72385865123263e-06,
+      "loss": 3.596,
+      "step": 5053
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.721916837797627e-06,
+      "loss": 3.5489,
+      "step": 5054
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.719975230584283e-06,
+      "loss": 3.5675,
+      "step": 5055
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.718033829905885e-06,
+      "loss": 3.3999,
+      "step": 5056
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.716092636075679e-06,
+      "loss": 3.515,
+      "step": 5057
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.714151649406884e-06,
+      "loss": 3.6151,
+      "step": 5058
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7122108702126807e-06,
+      "loss": 3.5863,
+      "step": 5059
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.710270298806219e-06,
+      "loss": 3.5753,
+      "step": 5060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7083299355006142e-06,
+      "loss": 3.5956,
+      "step": 5061
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.706389780608949e-06,
+      "loss": 3.4714,
+      "step": 5062
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7044498344442707e-06,
+      "loss": 3.5762,
+      "step": 5063
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.702510097319597e-06,
+      "loss": 3.4847,
+      "step": 5064
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.7005705695479046e-06,
+      "loss": 3.4662,
+      "step": 5065
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.698631251442144e-06,
+      "loss": 3.4739,
+      "step": 5066
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6966921433152264e-06,
+      "loss": 3.5956,
+      "step": 5067
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.694753245480035e-06,
+      "loss": 3.4628,
+      "step": 5068
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6928145582494102e-06,
+      "loss": 3.6314,
+      "step": 5069
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6908760819361692e-06,
+      "loss": 3.5373,
+      "step": 5070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.688937816853085e-06,
+      "loss": 3.6315,
+      "step": 5071
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.686999763312905e-06,
+      "loss": 3.725,
+      "step": 5072
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6850619216283345e-06,
+      "loss": 3.4094,
+      "step": 5073
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6831242921120523e-06,
+      "loss": 3.4408,
+      "step": 5074
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6811868750766967e-06,
+      "loss": 3.4766,
+      "step": 5075
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.679249670834878e-06,
+      "loss": 3.5919,
+      "step": 5076
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.677312679699164e-06,
+      "loss": 3.492,
+      "step": 5077
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6753759019820963e-06,
+      "loss": 3.6832,
+      "step": 5078
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6734393379961765e-06,
+      "loss": 3.5398,
+      "step": 5079
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.671502988053874e-06,
+      "loss": 3.434,
+      "step": 5080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.669566852467623e-06,
+      "loss": 3.4827,
+      "step": 5081
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.667630931549826e-06,
+      "loss": 3.462,
+      "step": 5082
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6656952256128443e-06,
+      "loss": 3.5691,
+      "step": 5083
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.663759734969012e-06,
+      "loss": 3.5817,
+      "step": 5084
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6618244599306213e-06,
+      "loss": 3.5499,
+      "step": 5085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.659889400809938e-06,
+      "loss": 3.4874,
+      "step": 5086
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6579545579191834e-06,
+      "loss": 3.5626,
+      "step": 5087
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.656019931570553e-06,
+      "loss": 3.5689,
+      "step": 5088
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6540855220761996e-06,
+      "loss": 3.4478,
+      "step": 5089
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6521513297482493e-06,
+      "loss": 3.5384,
+      "step": 5090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6502173548987823e-06,
+      "loss": 3.6044,
+      "step": 5091
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.648283597839856e-06,
+      "loss": 3.5609,
+      "step": 5092
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6463500588834823e-06,
+      "loss": 3.5588,
+      "step": 5093
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.644416738341645e-06,
+      "loss": 3.5273,
+      "step": 5094
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.642483636526286e-06,
+      "loss": 3.5538,
+      "step": 5095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6405507537493212e-06,
+      "loss": 3.5313,
+      "step": 5096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6386180903226217e-06,
+      "loss": 3.5538,
+      "step": 5097
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.636685646558028e-06,
+      "loss": 3.5197,
+      "step": 5098
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6347534227673442e-06,
+      "loss": 3.5481,
+      "step": 5099
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6328214192623414e-06,
+      "loss": 3.5724,
+      "step": 5100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6308896363547488e-06,
+      "loss": 3.5044,
+      "step": 5101
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.628958074356266e-06,
+      "loss": 3.5021,
+      "step": 5102
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.627026733578557e-06,
+      "loss": 3.6596,
+      "step": 5103
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.625095614333243e-06,
+      "loss": 3.5322,
+      "step": 5104
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6231647169319207e-06,
+      "loss": 3.4105,
+      "step": 5105
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6212340416861393e-06,
+      "loss": 3.5212,
+      "step": 5106
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.619303588907422e-06,
+      "loss": 3.6262,
+      "step": 5107
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.617373358907248e-06,
+      "loss": 3.6922,
+      "step": 5108
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6154433519970683e-06,
+      "loss": 3.5421,
+      "step": 5109
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6135135684882907e-06,
+      "loss": 3.5819,
+      "step": 5110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6115840086922926e-06,
+      "loss": 3.5822,
+      "step": 5111
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.60965467292041e-06,
+      "loss": 3.5082,
+      "step": 5112
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6077255614839502e-06,
+      "loss": 3.482,
+      "step": 5113
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6057966746941743e-06,
+      "loss": 3.6092,
+      "step": 5114
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6038680128623184e-06,
+      "loss": 3.616,
+      "step": 5115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6019395762995703e-06,
+      "loss": 3.5492,
+      "step": 5116
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.600011365317094e-06,
+      "loss": 3.4563,
+      "step": 5117
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5980833802260055e-06,
+      "loss": 3.5591,
+      "step": 5118
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5961556213373925e-06,
+      "loss": 3.5136,
+      "step": 5119
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5942280889623028e-06,
+      "loss": 3.4732,
+      "step": 5120
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.592300783411748e-06,
+      "loss": 3.5933,
+      "step": 5121
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5903737049967035e-06,
+      "loss": 3.5227,
+      "step": 5122
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.588446854028108e-06,
+      "loss": 3.6207,
+      "step": 5123
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.586520230816862e-06,
+      "loss": 3.5706,
+      "step": 5124
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.584593835673833e-06,
+      "loss": 3.4669,
+      "step": 5125
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.582667668909846e-06,
+      "loss": 3.6541,
+      "step": 5126
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.580741730835696e-06,
+      "loss": 3.5433,
+      "step": 5127
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5788160217621343e-06,
+      "loss": 3.6084,
+      "step": 5128
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5768905419998825e-06,
+      "loss": 3.5389,
+      "step": 5129
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5749652918596157e-06,
+      "loss": 3.4635,
+      "step": 5130
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.573040271651983e-06,
+      "loss": 3.5548,
+      "step": 5131
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.571115481687586e-06,
+      "loss": 3.4846,
+      "step": 5132
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5691909222769992e-06,
+      "loss": 3.4721,
+      "step": 5133
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5672665937307493e-06,
+      "loss": 3.5724,
+      "step": 5134
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.565342496359336e-06,
+      "loss": 3.5304,
+      "step": 5135
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5634186304732132e-06,
+      "loss": 3.5186,
+      "step": 5136
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5614949963828027e-06,
+      "loss": 3.592,
+      "step": 5137
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5595715943984865e-06,
+      "loss": 3.5073,
+      "step": 5138
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.557648424830611e-06,
+      "loss": 3.6485,
+      "step": 5139
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.555725487989483e-06,
+      "loss": 3.4544,
+      "step": 5140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.553802784185375e-06,
+      "loss": 3.6805,
+      "step": 5141
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.551880313728515e-06,
+      "loss": 3.5542,
+      "step": 5142
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5499580769291044e-06,
+      "loss": 3.6275,
+      "step": 5143
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.548036074097294e-06,
+      "loss": 3.5175,
+      "step": 5144
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.546114305543209e-06,
+      "loss": 3.607,
+      "step": 5145
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5441927715769264e-06,
+      "loss": 3.5152,
+      "step": 5146
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5422714725084944e-06,
+      "loss": 3.7177,
+      "step": 5147
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.540350408647915e-06,
+      "loss": 3.5199,
+      "step": 5148
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5384295803051606e-06,
+      "loss": 3.5108,
+      "step": 5149
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5365089877901573e-06,
+      "loss": 3.6195,
+      "step": 5150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5345886314128e-06,
+      "loss": 3.6106,
+      "step": 5151
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.53266851148294e-06,
+      "loss": 3.4666,
+      "step": 5152
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5307486283103966e-06,
+      "loss": 3.5617,
+      "step": 5153
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5288289822049437e-06,
+      "loss": 3.5562,
+      "step": 5154
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.526909573476323e-06,
+      "loss": 3.5386,
+      "step": 5155
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5249904024342342e-06,
+      "loss": 3.5346,
+      "step": 5156
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5230714693883412e-06,
+      "loss": 3.5693,
+      "step": 5157
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5211527746482663e-06,
+      "loss": 3.5583,
+      "step": 5158
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.519234318523599e-06,
+      "loss": 3.4953,
+      "step": 5159
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5173161013238817e-06,
+      "loss": 3.5445,
+      "step": 5160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5153981233586277e-06,
+      "loss": 3.6327,
+      "step": 5161
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5134803849373024e-06,
+      "loss": 3.4534,
+      "step": 5162
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.511562886369343e-06,
+      "loss": 3.6126,
+      "step": 5163
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.509645627964137e-06,
+      "loss": 3.6191,
+      "step": 5164
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5077286100310426e-06,
+      "loss": 3.6736,
+      "step": 5165
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.505811832879371e-06,
+      "loss": 3.5361,
+      "step": 5166
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5038952968184036e-06,
+      "loss": 3.5871,
+      "step": 5167
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.501979002157373e-06,
+      "loss": 3.5823,
+      "step": 5168
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.5000629492054816e-06,
+      "loss": 3.6012,
+      "step": 5169
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.498147138271885e-06,
+      "loss": 3.5649,
+      "step": 5170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.496231569665709e-06,
+      "loss": 3.5044,
+      "step": 5171
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4943162436960297e-06,
+      "loss": 3.557,
+      "step": 5172
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4924011606718926e-06,
+      "loss": 3.4841,
+      "step": 5173
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4904863209023e-06,
+      "loss": 3.5816,
+      "step": 5174
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4885717246962148e-06,
+      "loss": 3.5315,
+      "step": 5175
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4866573723625622e-06,
+      "loss": 3.503,
+      "step": 5176
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.484743264210229e-06,
+      "loss": 3.5647,
+      "step": 5177
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.482829400548057e-06,
+      "loss": 3.5249,
+      "step": 5178
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.480915781684857e-06,
+      "loss": 3.5436,
+      "step": 5179
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4790024079293917e-06,
+      "loss": 3.6125,
+      "step": 5180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.477089279590392e-06,
+      "loss": 3.6664,
+      "step": 5181
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.475176396976542e-06,
+      "loss": 3.6328,
+      "step": 5182
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4732637603964937e-06,
+      "loss": 3.5709,
+      "step": 5183
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4713513701588508e-06,
+      "loss": 3.6479,
+      "step": 5184
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4694392265721872e-06,
+      "loss": 3.5094,
+      "step": 5185
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4675273299450264e-06,
+      "loss": 3.5189,
+      "step": 5186
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4656156805858622e-06,
+      "loss": 3.6393,
+      "step": 5187
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4637042788031385e-06,
+      "loss": 3.5464,
+      "step": 5188
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4617931249052695e-06,
+      "loss": 3.5431,
+      "step": 5189
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4598822192006203e-06,
+      "loss": 3.4519,
+      "step": 5190
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4579715619975217e-06,
+      "loss": 3.5146,
+      "step": 5191
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4560611536042634e-06,
+      "loss": 3.6038,
+      "step": 5192
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.454150994329092e-06,
+      "loss": 3.4445,
+      "step": 5193
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.452241084480218e-06,
+      "loss": 3.4914,
+      "step": 5194
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4503314243658105e-06,
+      "loss": 3.6343,
+      "step": 5195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4484220142939928e-06,
+      "loss": 3.4872,
+      "step": 5196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4465128545728588e-06,
+      "loss": 3.4927,
+      "step": 5197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4446039455104507e-06,
+      "loss": 3.5748,
+      "step": 5198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4426952874147802e-06,
+      "loss": 3.4404,
+      "step": 5199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4407868805938082e-06,
+      "loss": 3.6908,
+      "step": 5200
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4388787253554667e-06,
+      "loss": 3.4904,
+      "step": 5201
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.436970822007635e-06,
+      "loss": 3.575,
+      "step": 5202
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.435063170858162e-06,
+      "loss": 3.5589,
+      "step": 5203
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4331557722148483e-06,
+      "loss": 3.5643,
+      "step": 5204
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4312486263854613e-06,
+      "loss": 3.5657,
+      "step": 5205
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.429341733677718e-06,
+      "loss": 3.5302,
+      "step": 5206
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.427435094399305e-06,
+      "loss": 3.3638,
+      "step": 5207
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4255287088578593e-06,
+      "loss": 3.5676,
+      "step": 5208
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.423622577360983e-06,
+      "loss": 3.5965,
+      "step": 5209
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4217167002162333e-06,
+      "loss": 3.6491,
+      "step": 5210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.419811077731129e-06,
+      "loss": 3.465,
+      "step": 5211
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4179057102131453e-06,
+      "loss": 3.6186,
+      "step": 5212
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4160005979697192e-06,
+      "loss": 3.5362,
+      "step": 5213
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4140957413082422e-06,
+      "loss": 3.447,
+      "step": 5214
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.412191140536072e-06,
+      "loss": 3.4753,
+      "step": 5215
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4102867959605144e-06,
+      "loss": 3.5501,
+      "step": 5216
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4083827078888458e-06,
+      "loss": 3.4348,
+      "step": 5217
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4064788766282904e-06,
+      "loss": 3.5316,
+      "step": 5218
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4045753024860393e-06,
+      "loss": 3.5718,
+      "step": 5219
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4026719857692346e-06,
+      "loss": 3.4739,
+      "step": 5220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.4007689267849864e-06,
+      "loss": 3.5025,
+      "step": 5221
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.398866125840351e-06,
+      "loss": 3.5687,
+      "step": 5222
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3969635832423564e-06,
+      "loss": 3.5239,
+      "step": 5223
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.395061299297976e-06,
+      "loss": 3.4207,
+      "step": 5224
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3931592743141533e-06,
+      "loss": 3.5364,
+      "step": 5225
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3912575085977796e-06,
+      "loss": 3.5377,
+      "step": 5226
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.389356002455714e-06,
+      "loss": 3.5658,
+      "step": 5227
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3874547561947646e-06,
+      "loss": 3.496,
+      "step": 5228
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3855537701217034e-06,
+      "loss": 3.5581,
+      "step": 5229
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3836530445432592e-06,
+      "loss": 3.4186,
+      "step": 5230
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3817525797661193e-06,
+      "loss": 3.7178,
+      "step": 5231
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3798523760969248e-06,
+      "loss": 3.5304,
+      "step": 5232
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.377952433842283e-06,
+      "loss": 3.4516,
+      "step": 5233
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3760527533087484e-06,
+      "loss": 3.6547,
+      "step": 5234
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3741533348028434e-06,
+      "loss": 3.5012,
+      "step": 5235
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3722541786310403e-06,
+      "loss": 3.5589,
+      "step": 5236
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3703552850997757e-06,
+      "loss": 3.4809,
+      "step": 5237
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3684566545154358e-06,
+      "loss": 3.4923,
+      "step": 5238
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3665582871843737e-06,
+      "loss": 3.5392,
+      "step": 5239
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3646601834128924e-06,
+      "loss": 3.519,
+      "step": 5240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3627623435072572e-06,
+      "loss": 3.468,
+      "step": 5241
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.360864767773686e-06,
+      "loss": 3.5683,
+      "step": 5242
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3589674565183618e-06,
+      "loss": 3.5741,
+      "step": 5243
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3570704100474144e-06,
+      "loss": 3.4934,
+      "step": 5244
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3551736286669424e-06,
+      "loss": 3.6245,
+      "step": 5245
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.353277112682991e-06,
+      "loss": 3.5212,
+      "step": 5246
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3513808624015703e-06,
+      "loss": 3.4735,
+      "step": 5247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.349484878128644e-06,
+      "loss": 3.5903,
+      "step": 5248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3475891601701333e-06,
+      "loss": 3.4553,
+      "step": 5249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.345693708831915e-06,
+      "loss": 3.6369,
+      "step": 5250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3437985244198283e-06,
+      "loss": 3.4733,
+      "step": 5251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3419036072396614e-06,
+      "loss": 3.5324,
+      "step": 5252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.340008957597168e-06,
+      "loss": 3.451,
+      "step": 5253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3381145757980493e-06,
+      "loss": 3.5042,
+      "step": 5254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3362204621479727e-06,
+      "loss": 3.5315,
+      "step": 5255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3343266169525533e-06,
+      "loss": 3.4281,
+      "step": 5256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3324330405173733e-06,
+      "loss": 3.5691,
+      "step": 5257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.330539733147959e-06,
+      "loss": 3.5618,
+      "step": 5258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3286466951498053e-06,
+      "loss": 3.6389,
+      "step": 5259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3267539268283538e-06,
+      "loss": 3.5048,
+      "step": 5260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.324861428489011e-06,
+      "loss": 3.5849,
+      "step": 5261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.322969200437133e-06,
+      "loss": 3.5991,
+      "step": 5262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.321077242978038e-06,
+      "loss": 3.6117,
+      "step": 5263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.319185556416995e-06,
+      "loss": 3.5725,
+      "step": 5264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3172941410592324e-06,
+      "loss": 3.4965,
+      "step": 5265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.315402997209935e-06,
+      "loss": 3.5514,
+      "step": 5266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.313512125174244e-06,
+      "loss": 3.5071,
+      "step": 5267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3116215252572535e-06,
+      "loss": 3.6226,
+      "step": 5268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.30973119776402e-06,
+      "loss": 3.4911,
+      "step": 5269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.307841142999548e-06,
+      "loss": 3.3912,
+      "step": 5270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3059513612688053e-06,
+      "loss": 3.6037,
+      "step": 5271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.30406185287671e-06,
+      "loss": 3.6099,
+      "step": 5272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.302172618128142e-06,
+      "loss": 3.5548,
+      "step": 5273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3002836573279295e-06,
+      "loss": 3.5278,
+      "step": 5274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2983949707808655e-06,
+      "loss": 3.5505,
+      "step": 5275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.296506558791689e-06,
+      "loss": 3.4182,
+      "step": 5276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2946184216651034e-06,
+      "loss": 3.5914,
+      "step": 5277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.292730559705761e-06,
+      "loss": 3.5801,
+      "step": 5278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2908429732182756e-06,
+      "loss": 3.4668,
+      "step": 5279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2889556625072106e-06,
+      "loss": 3.5098,
+      "step": 5280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2870686278770912e-06,
+      "loss": 3.4991,
+      "step": 5281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2851818696323924e-06,
+      "loss": 3.568,
+      "step": 5282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2832953880775477e-06,
+      "loss": 3.5507,
+      "step": 5283
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2814091835169455e-06,
+      "loss": 3.5311,
+      "step": 5284
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2795232562549296e-06,
+      "loss": 3.4791,
+      "step": 5285
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2776376065957958e-06,
+      "loss": 3.5419,
+      "step": 5286
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.275752234843803e-06,
+      "loss": 3.5689,
+      "step": 5287
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2738671413031555e-06,
+      "loss": 3.497,
+      "step": 5288
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.27198232627802e-06,
+      "loss": 3.4791,
+      "step": 5289
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.270097790072515e-06,
+      "loss": 3.5202,
+      "step": 5290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.268213532990717e-06,
+      "loss": 3.6426,
+      "step": 5291
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2663295553366504e-06,
+      "loss": 3.5313,
+      "step": 5292
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2644458574143052e-06,
+      "loss": 3.5729,
+      "step": 5293
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2625624395276144e-06,
+      "loss": 3.4737,
+      "step": 5294
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2606793019804772e-06,
+      "loss": 3.6007,
+      "step": 5295
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2587964450767384e-06,
+      "loss": 3.6137,
+      "step": 5296
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.256913869120204e-06,
+      "loss": 3.5109,
+      "step": 5297
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2550315744146287e-06,
+      "loss": 3.628,
+      "step": 5298
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.25314956126373e-06,
+      "loss": 3.542,
+      "step": 5299
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2512678299711696e-06,
+      "loss": 3.619,
+      "step": 5300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2493863808405725e-06,
+      "loss": 3.5582,
+      "step": 5301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2475052141755136e-06,
+      "loss": 3.4475,
+      "step": 5302
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.245624330279524e-06,
+      "loss": 3.5008,
+      "step": 5303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2437437294560892e-06,
+      "loss": 3.5706,
+      "step": 5304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.241863412008649e-06,
+      "loss": 3.5662,
+      "step": 5305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2399833782405933e-06,
+      "loss": 3.5808,
+      "step": 5306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2381036284552734e-06,
+      "loss": 3.6238,
+      "step": 5307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.236224162955989e-06,
+      "loss": 3.5569,
+      "step": 5308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.234344982046e-06,
+      "loss": 3.5543,
+      "step": 5309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.232466086028511e-06,
+      "loss": 3.6023,
+      "step": 5310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2305874752066917e-06,
+      "loss": 3.6081,
+      "step": 5311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2287091498836554e-06,
+      "loss": 3.4929,
+      "step": 5312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.226831110362479e-06,
+      "loss": 3.5654,
+      "step": 5313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2249533569461835e-06,
+      "loss": 3.5535,
+      "step": 5314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2230758899377533e-06,
+      "loss": 3.7098,
+      "step": 5315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2211987096401185e-06,
+      "loss": 3.654,
+      "step": 5316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2193218163561703e-06,
+      "loss": 3.5319,
+      "step": 5317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2174452103887455e-06,
+      "loss": 3.5875,
+      "step": 5318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2155688920406415e-06,
+      "loss": 3.5796,
+      "step": 5319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2136928616146055e-06,
+      "loss": 3.6199,
+      "step": 5320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2118171194133396e-06,
+      "loss": 3.5223,
+      "step": 5321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.209941665739499e-06,
+      "loss": 3.5093,
+      "step": 5322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.208066500895694e-06,
+      "loss": 3.4645,
+      "step": 5323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2061916251844833e-06,
+      "loss": 3.6406,
+      "step": 5324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2043170389083867e-06,
+      "loss": 3.4823,
+      "step": 5325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2024427423698682e-06,
+      "loss": 3.4405,
+      "step": 5326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.2005687358713556e-06,
+      "loss": 3.4952,
+      "step": 5327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.198695019715218e-06,
+      "loss": 3.4855,
+      "step": 5328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1968215942037896e-06,
+      "loss": 3.5257,
+      "step": 5329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1949484596393466e-06,
+      "loss": 3.5586,
+      "step": 5330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.193075616324128e-06,
+      "loss": 3.6083,
+      "step": 5331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.191203064560318e-06,
+      "loss": 3.4396,
+      "step": 5332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.189330804650061e-06,
+      "loss": 3.6353,
+      "step": 5333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1874588368954457e-06,
+      "loss": 3.5106,
+      "step": 5334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1855871615985227e-06,
+      "loss": 3.5043,
+      "step": 5335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.183715779061288e-06,
+      "loss": 3.5665,
+      "step": 5336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1818446895856957e-06,
+      "loss": 3.5461,
+      "step": 5337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.179973893473649e-06,
+      "loss": 3.5113,
+      "step": 5338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1781033910270053e-06,
+      "loss": 3.605,
+      "step": 5339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1762331825475755e-06,
+      "loss": 3.5429,
+      "step": 5340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.174363268337122e-06,
+      "loss": 3.5296,
+      "step": 5341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1724936486973567e-06,
+      "loss": 3.5572,
+      "step": 5342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1706243239299516e-06,
+      "loss": 3.565,
+      "step": 5343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.168755294336523e-06,
+      "loss": 3.4825,
+      "step": 5344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.166886560218646e-06,
+      "loss": 3.5389,
+      "step": 5345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1650181218778414e-06,
+      "loss": 3.5284,
+      "step": 5346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1631499796155905e-06,
+      "loss": 3.4837,
+      "step": 5347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1612821337333184e-06,
+      "loss": 3.488,
+      "step": 5348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.15941458453241e-06,
+      "loss": 3.5547,
+      "step": 5349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1575473323141946e-06,
+      "loss": 3.4687,
+      "step": 5350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1556803773799616e-06,
+      "loss": 3.7074,
+      "step": 5351
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.153813720030945e-06,
+      "loss": 3.5385,
+      "step": 5352
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1519473605683376e-06,
+      "loss": 3.5689,
+      "step": 5353
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.150081299293278e-06,
+      "loss": 3.5219,
+      "step": 5354
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1482155365068614e-06,
+      "loss": 3.6178,
+      "step": 5355
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1463500725101313e-06,
+      "loss": 3.5593,
+      "step": 5356
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1444849076040862e-06,
+      "loss": 3.4705,
+      "step": 5357
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1426200420896735e-06,
+      "loss": 3.5967,
+      "step": 5358
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.140755476267795e-06,
+      "loss": 3.5021,
+      "step": 5359
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1388912104393e-06,
+      "loss": 3.5278,
+      "step": 5360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1370272449049953e-06,
+      "loss": 3.5434,
+      "step": 5361
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1351635799656334e-06,
+      "loss": 3.5422,
+      "step": 5362
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.133300215921924e-06,
+      "loss": 3.4842,
+      "step": 5363
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.13143715307452e-06,
+      "loss": 3.5484,
+      "step": 5364
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1295743917240374e-06,
+      "loss": 3.5402,
+      "step": 5365
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1277119321710314e-06,
+      "loss": 3.5841,
+      "step": 5366
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1258497747160187e-06,
+      "loss": 3.5533,
+      "step": 5367
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.123987919659458e-06,
+      "loss": 3.6018,
+      "step": 5368
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.122126367301769e-06,
+      "loss": 3.5907,
+      "step": 5369
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.120265117943312e-06,
+      "loss": 3.5013,
+      "step": 5370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.118404171884409e-06,
+      "loss": 3.4316,
+      "step": 5371
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1165435294253244e-06,
+      "loss": 3.6329,
+      "step": 5372
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.11468319086628e-06,
+      "loss": 3.4939,
+      "step": 5373
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.112823156507443e-06,
+      "loss": 3.4716,
+      "step": 5374
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1109634266489353e-06,
+      "loss": 3.5949,
+      "step": 5375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.109104001590829e-06,
+      "loss": 3.3891,
+      "step": 5376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1072448816331467e-06,
+      "loss": 3.5631,
+      "step": 5377
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.1053860670758593e-06,
+      "loss": 3.4821,
+      "step": 5378
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.103527558218895e-06,
+      "loss": 3.5277,
+      "step": 5379
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.101669355362124e-06,
+      "loss": 3.547,
+      "step": 5380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.099811458805376e-06,
+      "loss": 3.4293,
+      "step": 5381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0979538688484227e-06,
+      "loss": 3.6067,
+      "step": 5382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0960965857909946e-06,
+      "loss": 3.4992,
+      "step": 5383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0942396099327645e-06,
+      "loss": 3.5305,
+      "step": 5384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0923829415733643e-06,
+      "loss": 3.4411,
+      "step": 5385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0905265810123663e-06,
+      "loss": 3.4686,
+      "step": 5386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0886705285493034e-06,
+      "loss": 3.4703,
+      "step": 5387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0868147844836504e-06,
+      "loss": 3.4695,
+      "step": 5388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.08495934911484e-06,
+      "loss": 3.5044,
+      "step": 5389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0831042227422457e-06,
+      "loss": 3.4951,
+      "step": 5390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0812494056652014e-06,
+      "loss": 3.4665,
+      "step": 5391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.079394898182983e-06,
+      "loss": 3.4962,
+      "step": 5392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.07754070059482e-06,
+      "loss": 3.5226,
+      "step": 5393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0756868131998917e-06,
+      "loss": 3.5233,
+      "step": 5394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.073833236297329e-06,
+      "loss": 3.5219,
+      "step": 5395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.071979970186206e-06,
+      "loss": 3.4849,
+      "step": 5396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0701270151655576e-06,
+      "loss": 3.4936,
+      "step": 5397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0682743715343565e-06,
+      "loss": 3.6911,
+      "step": 5398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.066422039591537e-06,
+      "loss": 3.4483,
+      "step": 5399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0645700196359706e-06,
+      "loss": 3.6193,
+      "step": 5400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0627183119664904e-06,
+      "loss": 3.6059,
+      "step": 5401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.06086691688187e-06,
+      "loss": 3.5088,
+      "step": 5402
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0590158346808395e-06,
+      "loss": 3.542,
+      "step": 5403
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0571650656620722e-06,
+      "loss": 3.4581,
+      "step": 5404
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.055314610124197e-06,
+      "loss": 3.6244,
+      "step": 5405
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0534644683657856e-06,
+      "loss": 3.448,
+      "step": 5406
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0516146406853664e-06,
+      "loss": 3.6583,
+      "step": 5407
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.04976512738141e-06,
+      "loss": 3.628,
+      "step": 5408
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0479159287523425e-06,
+      "loss": 3.5704,
+      "step": 5409
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.046067045096534e-06,
+      "loss": 3.4783,
+      "step": 5410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0442184767123074e-06,
+      "loss": 3.5849,
+      "step": 5411
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0423702238979337e-06,
+      "loss": 3.395,
+      "step": 5412
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0405222869516327e-06,
+      "loss": 3.5782,
+      "step": 5413
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0386746661715704e-06,
+      "loss": 3.5224,
+      "step": 5414
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0368273618558695e-06,
+      "loss": 3.5616,
+      "step": 5415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.034980374302593e-06,
+      "loss": 3.6174,
+      "step": 5416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.03313370380976e-06,
+      "loss": 3.5256,
+      "step": 5417
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.03128735067533e-06,
+      "loss": 3.4304,
+      "step": 5418
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0294413151972217e-06,
+      "loss": 3.5426,
+      "step": 5419
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.027595597673293e-06,
+      "loss": 3.496,
+      "step": 5420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.025750198401359e-06,
+      "loss": 3.5185,
+      "step": 5421
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.023905117679174e-06,
+      "loss": 3.5134,
+      "step": 5422
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.02206035580445e-06,
+      "loss": 3.5527,
+      "step": 5423
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.020215913074842e-06,
+      "loss": 3.5729,
+      "step": 5424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.018371789787956e-06,
+      "loss": 3.6179,
+      "step": 5425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0165279862413433e-06,
+      "loss": 3.4398,
+      "step": 5426
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.01468450273251e-06,
+      "loss": 3.5801,
+      "step": 5427
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.012841339558902e-06,
+      "loss": 3.5549,
+      "step": 5428
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0109984970179194e-06,
+      "loss": 3.4679,
+      "step": 5429
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.00915597540691e-06,
+      "loss": 3.6002,
+      "step": 5430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0073137750231694e-06,
+      "loss": 3.598,
+      "step": 5431
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0054718961639377e-06,
+      "loss": 3.6375,
+      "step": 5432
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.003630339126411e-06,
+      "loss": 3.5593,
+      "step": 5433
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.001789104207723e-06,
+      "loss": 3.6041,
+      "step": 5434
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.999948191704967e-06,
+      "loss": 3.6305,
+      "step": 5435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.998107601915174e-06,
+      "loss": 3.44,
+      "step": 5436
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.996267335135331e-06,
+      "loss": 3.6262,
+      "step": 5437
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.994427391662366e-06,
+      "loss": 3.6707,
+      "step": 5438
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9925877717931605e-06,
+      "loss": 3.4394,
+      "step": 5439
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9907484758245393e-06,
+      "loss": 3.5868,
+      "step": 5440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.98890950405328e-06,
+      "loss": 3.5118,
+      "step": 5441
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9870708567761015e-06,
+      "loss": 3.6284,
+      "step": 5442
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.985232534289677e-06,
+      "loss": 3.5406,
+      "step": 5443
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9833945368906214e-06,
+      "loss": 3.4879,
+      "step": 5444
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9815568648755023e-06,
+      "loss": 3.6096,
+      "step": 5445
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.97971951854083e-06,
+      "loss": 3.5124,
+      "step": 5446
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9778824981830656e-06,
+      "loss": 3.5534,
+      "step": 5447
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9760458040986163e-06,
+      "loss": 3.5306,
+      "step": 5448
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.974209436583838e-06,
+      "loss": 3.6327,
+      "step": 5449
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.972373395935031e-06,
+      "loss": 3.5163,
+      "step": 5450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9705376824484466e-06,
+      "loss": 3.6012,
+      "step": 5451
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.968702296420279e-06,
+      "loss": 3.5284,
+      "step": 5452
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9668672381466746e-06,
+      "loss": 3.53,
+      "step": 5453
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.965032507923722e-06,
+      "loss": 3.4676,
+      "step": 5454
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9631981060474613e-06,
+      "loss": 3.4623,
+      "step": 5455
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9613640328138747e-06,
+      "loss": 3.4091,
+      "step": 5456
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9595302885188987e-06,
+      "loss": 3.5323,
+      "step": 5457
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9576968734584067e-06,
+      "loss": 3.6113,
+      "step": 5458
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9558637879282287e-06,
+      "loss": 3.4454,
+      "step": 5459
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9540310322241337e-06,
+      "loss": 3.5983,
+      "step": 5460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9521986066418453e-06,
+      "loss": 3.4923,
+      "step": 5461
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9503665114770247e-06,
+      "loss": 3.5194,
+      "step": 5462
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9485347470252896e-06,
+      "loss": 3.5803,
+      "step": 5463
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.946703313582196e-06,
+      "loss": 3.4109,
+      "step": 5464
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9448722114432503e-06,
+      "loss": 3.5111,
+      "step": 5465
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9430414409039066e-06,
+      "loss": 3.4905,
+      "step": 5466
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9412110022595637e-06,
+      "loss": 3.4565,
+      "step": 5467
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9393808958055643e-06,
+      "loss": 3.5741,
+      "step": 5468
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9375511218372045e-06,
+      "loss": 3.5034,
+      "step": 5469
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.935721680649718e-06,
+      "loss": 3.4781,
+      "step": 5470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9338925725382942e-06,
+      "loss": 3.5274,
+      "step": 5471
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9320637977980593e-06,
+      "loss": 3.5004,
+      "step": 5472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.930235356724095e-06,
+      "loss": 3.6293,
+      "step": 5473
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.92840724961142e-06,
+      "loss": 3.5546,
+      "step": 5474
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.926579476755007e-06,
+      "loss": 3.5695,
+      "step": 5475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.924752038449768e-06,
+      "loss": 3.5931,
+      "step": 5476
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9229249349905686e-06,
+      "loss": 3.4449,
+      "step": 5477
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9210981666722115e-06,
+      "loss": 3.5944,
+      "step": 5478
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.919271733789455e-06,
+      "loss": 3.48,
+      "step": 5479
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.917445636636994e-06,
+      "loss": 3.5905,
+      "step": 5480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.915619875509476e-06,
+      "loss": 3.5358,
+      "step": 5481
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9137944507014893e-06,
+      "loss": 3.6057,
+      "step": 5482
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.911969362507574e-06,
+      "loss": 3.5319,
+      "step": 5483
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9101446112222077e-06,
+      "loss": 3.6498,
+      "step": 5484
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9083201971398228e-06,
+      "loss": 3.577,
+      "step": 5485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.906496120554789e-06,
+      "loss": 3.5527,
+      "step": 5486
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.904672381761429e-06,
+      "loss": 3.4843,
+      "step": 5487
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9028489810540027e-06,
+      "loss": 3.6169,
+      "step": 5488
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9010259187267242e-06,
+      "loss": 3.4793,
+      "step": 5489
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.899203195073746e-06,
+      "loss": 3.5932,
+      "step": 5490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.897380810389171e-06,
+      "loss": 3.5425,
+      "step": 5491
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8955587649670425e-06,
+      "loss": 3.5787,
+      "step": 5492
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8937370591013553e-06,
+      "loss": 3.5466,
+      "step": 5493
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8919156930860415e-06,
+      "loss": 3.5725,
+      "step": 5494
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8900946672149866e-06,
+      "loss": 3.4671,
+      "step": 5495
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.888273981782014e-06,
+      "loss": 3.5651,
+      "step": 5496
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8864536370808993e-06,
+      "loss": 3.5596,
+      "step": 5497
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.884633633405355e-06,
+      "loss": 3.4821,
+      "step": 5498
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8828139710490476e-06,
+      "loss": 3.5474,
+      "step": 5499
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8809946503055785e-06,
+      "loss": 3.4012,
+      "step": 5500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8791756714685043e-06,
+      "loss": 3.552,
+      "step": 5501
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8773570348313175e-06,
+      "loss": 3.5898,
+      "step": 5502
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8755387406874623e-06,
+      "loss": 3.634,
+      "step": 5503
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.873720789330322e-06,
+      "loss": 3.5295,
+      "step": 5504
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8719031810532305e-06,
+      "loss": 3.548,
+      "step": 5505
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8700859161494585e-06,
+      "loss": 3.4805,
+      "step": 5506
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8682689949122313e-06,
+      "loss": 3.4758,
+      "step": 5507
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8664524176347085e-06,
+      "loss": 3.489,
+      "step": 5508
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8646361846100022e-06,
+      "loss": 3.5252,
+      "step": 5509
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.862820296131166e-06,
+      "loss": 3.5137,
+      "step": 5510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8610047524911932e-06,
+      "loss": 3.5811,
+      "step": 5511
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8591895539830315e-06,
+      "loss": 3.4301,
+      "step": 5512
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.857374700899565e-06,
+      "loss": 3.5089,
+      "step": 5513
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8555601935336224e-06,
+      "loss": 3.5259,
+      "step": 5514
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8537460321779832e-06,
+      "loss": 3.5022,
+      "step": 5515
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8519322171253605e-06,
+      "loss": 3.6056,
+      "step": 5516
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8501187486684245e-06,
+      "loss": 3.5313,
+      "step": 5517
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.848305627099776e-06,
+      "loss": 3.5721,
+      "step": 5518
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8464928527119715e-06,
+      "loss": 3.5921,
+      "step": 5519
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8446804257975024e-06,
+      "loss": 3.631,
+      "step": 5520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8428683466488116e-06,
+      "loss": 3.4772,
+      "step": 5521
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8410566155582786e-06,
+      "loss": 3.6203,
+      "step": 5522
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8392452328182334e-06,
+      "loss": 3.5821,
+      "step": 5523
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8374341987209444e-06,
+      "loss": 3.5359,
+      "step": 5524
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8356235135586285e-06,
+      "loss": 3.4537,
+      "step": 5525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8338131776234414e-06,
+      "loss": 3.5998,
+      "step": 5526
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8320031912074875e-06,
+      "loss": 3.5202,
+      "step": 5527
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8301935546028094e-06,
+      "loss": 3.4548,
+      "step": 5528
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.828384268101399e-06,
+      "loss": 3.4567,
+      "step": 5529
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.826575331995186e-06,
+      "loss": 3.5775,
+      "step": 5530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8247667465760505e-06,
+      "loss": 3.5144,
+      "step": 5531
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.822958512135807e-06,
+      "loss": 3.5167,
+      "step": 5532
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8211506289662226e-06,
+      "loss": 3.4876,
+      "step": 5533
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.819343097359e-06,
+      "loss": 3.5671,
+      "step": 5534
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8175359176057915e-06,
+      "loss": 3.5493,
+      "step": 5535
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.815729089998186e-06,
+      "loss": 3.4953,
+      "step": 5536
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8139226148277243e-06,
+      "loss": 3.4653,
+      "step": 5537
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.81211649238588e-06,
+      "loss": 3.4127,
+      "step": 5538
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8103107229640803e-06,
+      "loss": 3.6077,
+      "step": 5539
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8085053068536862e-06,
+      "loss": 3.4913,
+      "step": 5540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8067002443460077e-06,
+      "loss": 3.4899,
+      "step": 5541
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8048955357322947e-06,
+      "loss": 3.5023,
+      "step": 5542
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.803091181303743e-06,
+      "loss": 3.6182,
+      "step": 5543
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.8012871813514862e-06,
+      "loss": 3.5085,
+      "step": 5544
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7994835361666085e-06,
+      "loss": 3.4218,
+      "step": 5545
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7976802460401287e-06,
+      "loss": 3.5591,
+      "step": 5546
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7958773112630107e-06,
+      "loss": 3.4995,
+      "step": 5547
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.794074732126165e-06,
+      "loss": 3.5339,
+      "step": 5548
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.792272508920443e-06,
+      "loss": 3.6383,
+      "step": 5549
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7904706419366316e-06,
+      "loss": 3.5898,
+      "step": 5550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.788669131465473e-06,
+      "loss": 3.496,
+      "step": 5551
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7868679777976392e-06,
+      "loss": 3.5405,
+      "step": 5552
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.785067181223756e-06,
+      "loss": 3.5252,
+      "step": 5553
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.783266742034382e-06,
+      "loss": 3.4716,
+      "step": 5554
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7814666605200254e-06,
+      "loss": 3.567,
+      "step": 5555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7796669369711294e-06,
+      "loss": 3.5538,
+      "step": 5556
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7778675716780883e-06,
+      "loss": 3.6943,
+      "step": 5557
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.776068564931229e-06,
+      "loss": 3.5444,
+      "step": 5558
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.774269917020831e-06,
+      "loss": 3.6147,
+      "step": 5559
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7724716282371037e-06,
+      "loss": 3.4585,
+      "step": 5560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.770673698870211e-06,
+      "loss": 3.5641,
+      "step": 5561
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7688761292102505e-06,
+      "loss": 3.506,
+      "step": 5562
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.767078919547263e-06,
+      "loss": 3.7058,
+      "step": 5563
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7652820701712345e-06,
+      "loss": 3.4939,
+      "step": 5564
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.763485581372089e-06,
+      "loss": 3.5132,
+      "step": 5565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7616894534396964e-06,
+      "loss": 3.6479,
+      "step": 5566
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7598936866638627e-06,
+      "loss": 3.6095,
+      "step": 5567
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7580982813343425e-06,
+      "loss": 3.4816,
+      "step": 5568
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.756303237740826e-06,
+      "loss": 3.7614,
+      "step": 5569
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.75450855617295e-06,
+      "loss": 3.5084,
+      "step": 5570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.752714236920288e-06,
+      "loss": 3.4722,
+      "step": 5571
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.75092028027236e-06,
+      "loss": 3.5914,
+      "step": 5572
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7491266865186227e-06,
+      "loss": 3.4727,
+      "step": 5573
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.747333455948479e-06,
+      "loss": 3.476,
+      "step": 5574
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7455405888512678e-06,
+      "loss": 3.5332,
+      "step": 5575
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.743748085516275e-06,
+      "loss": 3.4539,
+      "step": 5576
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.741955946232723e-06,
+      "loss": 3.4185,
+      "step": 5577
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7401641712897803e-06,
+      "loss": 3.5278,
+      "step": 5578
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.738372760976551e-06,
+      "loss": 3.6055,
+      "step": 5579
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.736581715582086e-06,
+      "loss": 3.5614,
+      "step": 5580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.734791035395371e-06,
+      "loss": 3.5199,
+      "step": 5581
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7330007207053413e-06,
+      "loss": 3.6073,
+      "step": 5582
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.731210771800863e-06,
+      "loss": 3.5453,
+      "step": 5583
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7294211889707534e-06,
+      "loss": 3.5609,
+      "step": 5584
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7276319725037613e-06,
+      "loss": 3.5317,
+      "step": 5585
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.725843122688585e-06,
+      "loss": 3.4873,
+      "step": 5586
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.724054639813858e-06,
+      "loss": 3.507,
+      "step": 5587
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7222665241681533e-06,
+      "loss": 3.532,
+      "step": 5588
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7204787760399924e-06,
+      "loss": 3.5273,
+      "step": 5589
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7186913957178296e-06,
+      "loss": 3.5942,
+      "step": 5590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.716904383490062e-06,
+      "loss": 3.6103,
+      "step": 5591
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7151177396450306e-06,
+      "loss": 3.5499,
+      "step": 5592
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.713331464471012e-06,
+      "loss": 3.5577,
+      "step": 5593
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.711545558256229e-06,
+      "loss": 3.4899,
+      "step": 5594
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7097600212888374e-06,
+      "loss": 3.5329,
+      "step": 5595
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7079748538569427e-06,
+      "loss": 3.5251,
+      "step": 5596
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.706190056248581e-06,
+      "loss": 3.4977,
+      "step": 5597
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.704405628751738e-06,
+      "loss": 3.5397,
+      "step": 5598
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.70262157165433e-06,
+      "loss": 3.583,
+      "step": 5599
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.700837885244225e-06,
+      "loss": 3.5816,
+      "step": 5600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6990545698092184e-06,
+      "loss": 3.6499,
+      "step": 5601
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6972716256370584e-06,
+      "loss": 3.6609,
+      "step": 5602
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6954890530154214e-06,
+      "loss": 3.6039,
+      "step": 5603
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.693706852231935e-06,
+      "loss": 3.5287,
+      "step": 5604
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.691925023574156e-06,
+      "loss": 3.6645,
+      "step": 5605
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6901435673295916e-06,
+      "loss": 3.5799,
+      "step": 5606
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6883624837856797e-06,
+      "loss": 3.5659,
+      "step": 5607
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.686581773229806e-06,
+      "loss": 3.4647,
+      "step": 5608
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6848014359492884e-06,
+      "loss": 3.4982,
+      "step": 5609
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6830214722313924e-06,
+      "loss": 3.5253,
+      "step": 5610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6812418823633152e-06,
+      "loss": 3.481,
+      "step": 5611
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6794626666322015e-06,
+      "loss": 3.496,
+      "step": 5612
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.677683825325128e-06,
+      "loss": 3.4836,
+      "step": 5613
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6759053587291194e-06,
+      "loss": 3.4951,
+      "step": 5614
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.674127267131131e-06,
+      "loss": 3.589,
+      "step": 5615
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.672349550818065e-06,
+      "loss": 3.4646,
+      "step": 5616
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6705722100767582e-06,
+      "loss": 3.5453,
+      "step": 5617
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6687952451939902e-06,
+      "loss": 3.5617,
+      "step": 5618
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6670186564564758e-06,
+      "loss": 3.5674,
+      "step": 5619
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.665242444150876e-06,
+      "loss": 3.5078,
+      "step": 5620
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6634666085637817e-06,
+      "loss": 3.5307,
+      "step": 5621
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.661691149981732e-06,
+      "loss": 3.5083,
+      "step": 5622
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6599160686912005e-06,
+      "loss": 3.5857,
+      "step": 5623
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.658141364978598e-06,
+      "loss": 3.5353,
+      "step": 5624
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6563670391302805e-06,
+      "loss": 3.5056,
+      "step": 5625
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.654593091432538e-06,
+      "loss": 3.6106,
+      "step": 5626
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6528195221715993e-06,
+      "loss": 3.5237,
+      "step": 5627
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6510463316336367e-06,
+      "loss": 3.5278,
+      "step": 5628
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.649273520104756e-06,
+      "loss": 3.5898,
+      "step": 5629
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6475010878710085e-06,
+      "loss": 3.504,
+      "step": 5630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6457290352183747e-06,
+      "loss": 3.4986,
+      "step": 5631
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.643957362432785e-06,
+      "loss": 3.6165,
+      "step": 5632
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6421860698000976e-06,
+      "loss": 3.4551,
+      "step": 5633
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6404151576061194e-06,
+      "loss": 3.532,
+      "step": 5634
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6386446261365874e-06,
+      "loss": 3.5335,
+      "step": 5635
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6368744756771837e-06,
+      "loss": 3.522,
+      "step": 5636
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6351047065135238e-06,
+      "loss": 3.5766,
+      "step": 5637
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6333353189311673e-06,
+      "loss": 3.6021,
+      "step": 5638
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.631566313215604e-06,
+      "loss": 3.5926,
+      "step": 5639
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6297976896522727e-06,
+      "loss": 3.467,
+      "step": 5640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6280294485265396e-06,
+      "loss": 3.4307,
+      "step": 5641
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.626261590123719e-06,
+      "loss": 3.574,
+      "step": 5642
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.624494114729055e-06,
+      "loss": 3.4648,
+      "step": 5643
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.622727022627738e-06,
+      "loss": 3.5995,
+      "step": 5644
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.620960314104888e-06,
+      "loss": 3.4653,
+      "step": 5645
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.619193989445571e-06,
+      "loss": 3.5703,
+      "step": 5646
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.617428048934785e-06,
+      "loss": 3.5946,
+      "step": 5647
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.615662492857471e-06,
+      "loss": 3.6143,
+      "step": 5648
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6138973214985026e-06,
+      "loss": 3.5301,
+      "step": 5649
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6121325351426974e-06,
+      "loss": 3.5294,
+      "step": 5650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.610368134074805e-06,
+      "loss": 3.5357,
+      "step": 5651
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6086041185795186e-06,
+      "loss": 3.5646,
+      "step": 5652
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6068404889414624e-06,
+      "loss": 3.5841,
+      "step": 5653
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6050772454452054e-06,
+      "loss": 3.5062,
+      "step": 5654
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.603314388375249e-06,
+      "loss": 3.5323,
+      "step": 5655
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.601551918016036e-06,
+      "loss": 3.5504,
+      "step": 5656
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5997898346519425e-06,
+      "loss": 3.4607,
+      "step": 5657
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5980281385672873e-06,
+      "loss": 3.4239,
+      "step": 5658
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5962668300463236e-06,
+      "loss": 3.5195,
+      "step": 5659
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5945059093732405e-06,
+      "loss": 3.5728,
+      "step": 5660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5927453768321692e-06,
+      "loss": 3.423,
+      "step": 5661
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5909852327071747e-06,
+      "loss": 3.4564,
+      "step": 5662
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.589225477282259e-06,
+      "loss": 3.4315,
+      "step": 5663
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5874661108413658e-06,
+      "loss": 3.4896,
+      "step": 5664
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5857071336683683e-06,
+      "loss": 3.4545,
+      "step": 5665
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5839485460470866e-06,
+      "loss": 3.4878,
+      "step": 5666
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5821903482612683e-06,
+      "loss": 3.5758,
+      "step": 5667
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5804325405946067e-06,
+      "loss": 3.4878,
+      "step": 5668
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.578675123330725e-06,
+      "loss": 3.5624,
+      "step": 5669
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.576918096753189e-06,
+      "loss": 3.621,
+      "step": 5670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5751614611454963e-06,
+      "loss": 3.6174,
+      "step": 5671
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5734052167910873e-06,
+      "loss": 3.5109,
+      "step": 5672
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5716493639733324e-06,
+      "loss": 3.6209,
+      "step": 5673
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5698939029755463e-06,
+      "loss": 3.5874,
+      "step": 5674
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.568138834080973e-06,
+      "loss": 3.487,
+      "step": 5675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.566384157572801e-06,
+      "loss": 3.4594,
+      "step": 5676
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.564629873734147e-06,
+      "loss": 3.4792,
+      "step": 5677
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5628759828480722e-06,
+      "loss": 3.5974,
+      "step": 5678
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.561122485197568e-06,
+      "loss": 3.5016,
+      "step": 5679
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5593693810655695e-06,
+      "loss": 3.5589,
+      "step": 5680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5576166707349387e-06,
+      "loss": 3.4625,
+      "step": 5681
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5558643544884844e-06,
+      "loss": 3.5242,
+      "step": 5682
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.554112432608943e-06,
+      "loss": 3.4109,
+      "step": 5683
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5523609053789946e-06,
+      "loss": 3.4854,
+      "step": 5684
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.550609773081248e-06,
+      "loss": 3.4786,
+      "step": 5685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5488590359982567e-06,
+      "loss": 3.4957,
+      "step": 5686
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.547108694412503e-06,
+      "loss": 3.5062,
+      "step": 5687
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5453587486064113e-06,
+      "loss": 3.6044,
+      "step": 5688
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5436091988623357e-06,
+      "loss": 3.53,
+      "step": 5689
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.541860045462574e-06,
+      "loss": 3.5162,
+      "step": 5690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.540111288689353e-06,
+      "loss": 3.5673,
+      "step": 5691
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5383629288248423e-06,
+      "loss": 3.5271,
+      "step": 5692
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5366149661511385e-06,
+      "loss": 3.5192,
+      "step": 5693
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5348674009502854e-06,
+      "loss": 3.5022,
+      "step": 5694
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.533120233504253e-06,
+      "loss": 3.6211,
+      "step": 5695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.53137346409495e-06,
+      "loss": 3.4682,
+      "step": 5696
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5296270930042244e-06,
+      "loss": 3.5768,
+      "step": 5697
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5278811205138547e-06,
+      "loss": 3.5262,
+      "step": 5698
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.52613554690556e-06,
+      "loss": 3.5776,
+      "step": 5699
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.524390372460992e-06,
+      "loss": 3.4659,
+      "step": 5700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5226455974617358e-06,
+      "loss": 3.4739,
+      "step": 5701
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5209012221893198e-06,
+      "loss": 3.5264,
+      "step": 5702
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.519157246925198e-06,
+      "loss": 3.5265,
+      "step": 5703
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5174136719507686e-06,
+      "loss": 3.4799,
+      "step": 5704
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5156704975473577e-06,
+      "loss": 3.5083,
+      "step": 5705
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5139277239962357e-06,
+      "loss": 3.6015,
+      "step": 5706
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.512185351578598e-06,
+      "loss": 3.6043,
+      "step": 5707
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5104433805755834e-06,
+      "loss": 3.5672,
+      "step": 5708
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5087018112682614e-06,
+      "loss": 3.4594,
+      "step": 5709
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5069606439376404e-06,
+      "loss": 3.4942,
+      "step": 5710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5052198788646585e-06,
+      "loss": 3.5423,
+      "step": 5711
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5034795163301955e-06,
+      "loss": 3.5124,
+      "step": 5712
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.50173955661506e-06,
+      "loss": 3.4444,
+      "step": 5713
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.5000000000000015e-06,
+      "loss": 3.5814,
+      "step": 5714
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.498260846765698e-06,
+      "loss": 3.5887,
+      "step": 5715
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4965220971927688e-06,
+      "loss": 3.4282,
+      "step": 5716
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4947837515617627e-06,
+      "loss": 3.4997,
+      "step": 5717
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.493045810153169e-06,
+      "loss": 3.4928,
+      "step": 5718
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.491308273247405e-06,
+      "loss": 3.4977,
+      "step": 5719
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.48957114112483e-06,
+      "loss": 3.5229,
+      "step": 5720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4878344140657296e-06,
+      "loss": 3.4459,
+      "step": 5721
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4860980923503337e-06,
+      "loss": 3.5995,
+      "step": 5722
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4843621762587973e-06,
+      "loss": 3.5776,
+      "step": 5723
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4826266660712182e-06,
+      "loss": 3.4207,
+      "step": 5724
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.480891562067622e-06,
+      "loss": 3.5972,
+      "step": 5725
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4791568645279744e-06,
+      "loss": 3.5638,
+      "step": 5726
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.47742257373217e-06,
+      "loss": 3.6216,
+      "step": 5727
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.475688689960044e-06,
+      "loss": 3.5938,
+      "step": 5728
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.473955213491359e-06,
+      "loss": 3.6659,
+      "step": 5729
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.472222144605819e-06,
+      "loss": 3.6709,
+      "step": 5730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4704894835830564e-06,
+      "loss": 3.6398,
+      "step": 5731
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.468757230702638e-06,
+      "loss": 3.6012,
+      "step": 5732
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.467025386244072e-06,
+      "loss": 3.4641,
+      "step": 5733
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.46529395048679e-06,
+      "loss": 3.6201,
+      "step": 5734
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4635629237101684e-06,
+      "loss": 3.5237,
+      "step": 5735
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4618323061935097e-06,
+      "loss": 3.5336,
+      "step": 5736
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4601020982160507e-06,
+      "loss": 3.5006,
+      "step": 5737
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4583723000569693e-06,
+      "loss": 3.556,
+      "step": 5738
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4566429119953673e-06,
+      "loss": 3.5684,
+      "step": 5739
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4549139343102905e-06,
+      "loss": 3.5632,
+      "step": 5740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4531853672807093e-06,
+      "loss": 3.5312,
+      "step": 5741
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.451457211185535e-06,
+      "loss": 3.5296,
+      "step": 5742
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.449729466303607e-06,
+      "loss": 3.4917,
+      "step": 5743
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.448002132913704e-06,
+      "loss": 3.4986,
+      "step": 5744
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.446275211294531e-06,
+      "loss": 3.5028,
+      "step": 5745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4445487017247354e-06,
+      "loss": 3.4612,
+      "step": 5746
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4428226044828896e-06,
+      "loss": 3.5445,
+      "step": 5747
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4410969198475066e-06,
+      "loss": 3.5051,
+      "step": 5748
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4393716480970263e-06,
+      "loss": 3.5507,
+      "step": 5749
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4376467895098287e-06,
+      "loss": 3.4877,
+      "step": 5750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.43592234436422e-06,
+      "loss": 3.5148,
+      "step": 5751
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.434198312938448e-06,
+      "loss": 3.5701,
+      "step": 5752
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4324746955106836e-06,
+      "loss": 3.506,
+      "step": 5753
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.430751492359042e-06,
+      "loss": 3.6099,
+      "step": 5754
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4290287037615614e-06,
+      "loss": 3.4075,
+      "step": 5755
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4273063299962213e-06,
+      "loss": 3.6404,
+      "step": 5756
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4255843713409277e-06,
+      "loss": 3.5167,
+      "step": 5757
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4238628280735255e-06,
+      "loss": 3.5388,
+      "step": 5758
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.422141700471786e-06,
+      "loss": 3.5515,
+      "step": 5759
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.420420988813421e-06,
+      "loss": 3.5507,
+      "step": 5760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4187006933760683e-06,
+      "loss": 3.4496,
+      "step": 5761
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.416980814437304e-06,
+      "loss": 3.5857,
+      "step": 5762
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.415261352274632e-06,
+      "loss": 3.3979,
+      "step": 5763
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.413542307165494e-06,
+      "loss": 3.586,
+      "step": 5764
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.411823679387259e-06,
+      "loss": 3.5695,
+      "step": 5765
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4101054692172354e-06,
+      "loss": 3.5277,
+      "step": 5766
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.408387676932656e-06,
+      "loss": 3.5867,
+      "step": 5767
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4066703028106946e-06,
+      "loss": 3.6861,
+      "step": 5768
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4049533471284514e-06,
+      "loss": 3.5435,
+      "step": 5769
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4032368101629594e-06,
+      "loss": 3.655,
+      "step": 5770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.40152069219119e-06,
+      "loss": 3.5472,
+      "step": 5771
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3998049934900404e-06,
+      "loss": 3.5015,
+      "step": 5772
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3980897143363404e-06,
+      "loss": 3.5478,
+      "step": 5773
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3963748550068584e-06,
+      "loss": 3.6377,
+      "step": 5774
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.394660415778287e-06,
+      "loss": 3.4925,
+      "step": 5775
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3929463969272594e-06,
+      "loss": 3.5269,
+      "step": 5776
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.391232798730331e-06,
+      "loss": 3.4147,
+      "step": 5777
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.389519621464001e-06,
+      "loss": 3.4705,
+      "step": 5778
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3878068654046883e-06,
+      "loss": 3.4258,
+      "step": 5779
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3860945308287554e-06,
+      "loss": 3.4561,
+      "step": 5780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3843826180124872e-06,
+      "loss": 3.5755,
+      "step": 5781
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3826711272321087e-06,
+      "loss": 3.5007,
+      "step": 5782
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3809600587637682e-06,
+      "loss": 3.5219,
+      "step": 5783
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.379249412883556e-06,
+      "loss": 3.6544,
+      "step": 5784
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.377539189867483e-06,
+      "loss": 3.5093,
+      "step": 5785
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.375829389991503e-06,
+      "loss": 3.5293,
+      "step": 5786
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.374120013531492e-06,
+      "loss": 3.5218,
+      "step": 5787
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.372411060763265e-06,
+      "loss": 3.5519,
+      "step": 5788
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.370702531962562e-06,
+      "loss": 3.6641,
+      "step": 5789
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.368994427405062e-06,
+      "loss": 3.5579,
+      "step": 5790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3672867473663672e-06,
+      "loss": 3.5775,
+      "step": 5791
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.365579492122021e-06,
+      "loss": 3.5256,
+      "step": 5792
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.363872661947488e-06,
+      "loss": 3.5873,
+      "step": 5793
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3621662571181724e-06,
+      "loss": 3.5823,
+      "step": 5794
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3604602779094045e-06,
+      "loss": 3.6442,
+      "step": 5795
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.35875472459645e-06,
+      "loss": 3.487,
+      "step": 5796
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3570495974545012e-06,
+      "loss": 3.5593,
+      "step": 5797
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3553448967586878e-06,
+      "loss": 3.7,
+      "step": 5798
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3536406227840632e-06,
+      "loss": 3.5412,
+      "step": 5799
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3519367758056194e-06,
+      "loss": 3.56,
+      "step": 5800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.350233356098273e-06,
+      "loss": 3.5537,
+      "step": 5801
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3485303639368783e-06,
+      "loss": 3.4467,
+      "step": 5802
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3468277995962136e-06,
+      "loss": 3.5312,
+      "step": 5803
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3451256633509938e-06,
+      "loss": 3.4417,
+      "step": 5804
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3434239554758627e-06,
+      "loss": 3.6043,
+      "step": 5805
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.341722676245391e-06,
+      "loss": 3.4532,
+      "step": 5806
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.340021825934089e-06,
+      "loss": 3.5505,
+      "step": 5807
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3383214048163905e-06,
+      "loss": 3.5761,
+      "step": 5808
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3366214131666604e-06,
+      "loss": 3.6067,
+      "step": 5809
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3349218512592e-06,
+      "loss": 3.4272,
+      "step": 5810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3332227193682348e-06,
+      "loss": 3.3979,
+      "step": 5811
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3315240177679262e-06,
+      "loss": 3.5396,
+      "step": 5812
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3298257467323605e-06,
+      "loss": 3.5012,
+      "step": 5813
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3281279065355607e-06,
+      "loss": 3.4801,
+      "step": 5814
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3264304974514735e-06,
+      "loss": 3.4591,
+      "step": 5815
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3247335197539844e-06,
+      "loss": 3.424,
+      "step": 5816
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3230369737169005e-06,
+      "loss": 3.5928,
+      "step": 5817
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.321340859613967e-06,
+      "loss": 3.4794,
+      "step": 5818
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3196451777188515e-06,
+      "loss": 3.5243,
+      "step": 5819
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3179499283051615e-06,
+      "loss": 3.4302,
+      "step": 5820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3162551116464243e-06,
+      "loss": 3.5579,
+      "step": 5821
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3145607280161064e-06,
+      "loss": 3.5865,
+      "step": 5822
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.312866777687597e-06,
+      "loss": 3.6665,
+      "step": 5823
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.311173260934223e-06,
+      "loss": 3.519,
+      "step": 5824
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.309480178029233e-06,
+      "loss": 3.4335,
+      "step": 5825
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3077875292458133e-06,
+      "loss": 3.5144,
+      "step": 5826
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.306095314857074e-06,
+      "loss": 3.5756,
+      "step": 5827
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.3044035351360607e-06,
+      "loss": 3.5124,
+      "step": 5828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.302712190355742e-06,
+      "loss": 3.5888,
+      "step": 5829
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.301021280789024e-06,
+      "loss": 3.6577,
+      "step": 5830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.299330806708736e-06,
+      "loss": 3.6139,
+      "step": 5831
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2976407683876433e-06,
+      "loss": 3.6472,
+      "step": 5832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.295951166098433e-06,
+      "loss": 3.538,
+      "step": 5833
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.294262000113731e-06,
+      "loss": 3.5502,
+      "step": 5834
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.292573270706084e-06,
+      "loss": 3.5679,
+      "step": 5835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2908849781479757e-06,
+      "loss": 3.5713,
+      "step": 5836
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.289197122711813e-06,
+      "loss": 3.5342,
+      "step": 5837
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2875097046699384e-06,
+      "loss": 3.5017,
+      "step": 5838
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.285822724294617e-06,
+      "loss": 3.6169,
+      "step": 5839
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2841361818580517e-06,
+      "loss": 3.46,
+      "step": 5840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2824500776323667e-06,
+      "loss": 3.565,
+      "step": 5841
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2807644118896175e-06,
+      "loss": 3.6058,
+      "step": 5842
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2790791849017947e-06,
+      "loss": 3.5329,
+      "step": 5843
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.277394396940808e-06,
+      "loss": 3.5486,
+      "step": 5844
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.275710048278507e-06,
+      "loss": 3.5672,
+      "step": 5845
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2740261391866634e-06,
+      "loss": 3.4577,
+      "step": 5846
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2723426699369776e-06,
+      "loss": 3.4827,
+      "step": 5847
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2706596408010846e-06,
+      "loss": 3.5529,
+      "step": 5848
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2689770520505416e-06,
+      "loss": 3.4088,
+      "step": 5849
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.267294903956842e-06,
+      "loss": 3.553,
+      "step": 5850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2656131967914002e-06,
+      "loss": 3.5567,
+      "step": 5851
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2639319308255675e-06,
+      "loss": 3.6481,
+      "step": 5852
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2622511063306167e-06,
+      "loss": 3.4594,
+      "step": 5853
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.260570723577756e-06,
+      "loss": 3.5854,
+      "step": 5854
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2588907828381144e-06,
+      "loss": 3.6287,
+      "step": 5855
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.25721128438276e-06,
+      "loss": 3.5222,
+      "step": 5856
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2555322284826786e-06,
+      "loss": 3.5825,
+      "step": 5857
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2538536154087935e-06,
+      "loss": 3.5356,
+      "step": 5858
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2521754454319495e-06,
+      "loss": 3.6092,
+      "step": 5859
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2504977188229267e-06,
+      "loss": 3.5212,
+      "step": 5860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.248820435852426e-06,
+      "loss": 3.6702,
+      "step": 5861
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.247143596791086e-06,
+      "loss": 3.544,
+      "step": 5862
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.245467201909463e-06,
+      "loss": 3.4233,
+      "step": 5863
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.243791251478052e-06,
+      "loss": 3.5269,
+      "step": 5864
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2421157457672678e-06,
+      "loss": 3.4819,
+      "step": 5865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2404406850474604e-06,
+      "loss": 3.5217,
+      "step": 5866
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2387660695889006e-06,
+      "loss": 3.6498,
+      "step": 5867
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2370918996617962e-06,
+      "loss": 3.5191,
+      "step": 5868
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.235418175536274e-06,
+      "loss": 3.449,
+      "step": 5869
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.233744897482397e-06,
+      "loss": 3.5648,
+      "step": 5870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.232072065770149e-06,
+      "loss": 3.5447,
+      "step": 5871
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.230399680669449e-06,
+      "loss": 3.429,
+      "step": 5872
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2287277424501357e-06,
+      "loss": 3.5168,
+      "step": 5873
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.227056251381985e-06,
+      "loss": 3.5821,
+      "step": 5874
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2253852077346915e-06,
+      "loss": 3.523,
+      "step": 5875
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2237146117778856e-06,
+      "loss": 3.5106,
+      "step": 5876
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2220444637811208e-06,
+      "loss": 3.5282,
+      "step": 5877
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2203747640138756e-06,
+      "loss": 3.5821,
+      "step": 5878
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2187055127455653e-06,
+      "loss": 3.4542,
+      "step": 5879
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2170367102455236e-06,
+      "loss": 3.6137,
+      "step": 5880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.215368356783019e-06,
+      "loss": 3.4826,
+      "step": 5881
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2137004526272415e-06,
+      "loss": 3.5304,
+      "step": 5882
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2120329980473104e-06,
+      "loss": 3.5654,
+      "step": 5883
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.210365993312277e-06,
+      "loss": 3.5488,
+      "step": 5884
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2086994386911124e-06,
+      "loss": 3.4338,
+      "step": 5885
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2070333344527232e-06,
+      "loss": 3.4916,
+      "step": 5886
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2053676808659347e-06,
+      "loss": 3.5439,
+      "step": 5887
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.203702478199508e-06,
+      "loss": 3.6114,
+      "step": 5888
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2020377267221245e-06,
+      "loss": 3.5541,
+      "step": 5889
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.2003734267023975e-06,
+      "loss": 3.5441,
+      "step": 5890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1987095784088637e-06,
+      "loss": 3.4243,
+      "step": 5891
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1970461821099914e-06,
+      "loss": 3.4931,
+      "step": 5892
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.19538323807417e-06,
+      "loss": 3.5977,
+      "step": 5893
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1937207465697236e-06,
+      "loss": 3.5309,
+      "step": 5894
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.192058707864895e-06,
+      "loss": 3.4795,
+      "step": 5895
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1903971222278614e-06,
+      "loss": 3.665,
+      "step": 5896
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.18873598992672e-06,
+      "loss": 3.4566,
+      "step": 5897
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.187075311229503e-06,
+      "loss": 3.6093,
+      "step": 5898
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.185415086404159e-06,
+      "loss": 3.5052,
+      "step": 5899
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1837553157185744e-06,
+      "loss": 3.507,
+      "step": 5900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1820959994405522e-06,
+      "loss": 3.475,
+      "step": 5901
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.180437137837832e-06,
+      "loss": 3.5771,
+      "step": 5902
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1787787311780694e-06,
+      "loss": 3.4507,
+      "step": 5903
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.177120779728858e-06,
+      "loss": 3.5503,
+      "step": 5904
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1754632837577066e-06,
+      "loss": 3.4431,
+      "step": 5905
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.17380624353206e-06,
+      "loss": 3.5118,
+      "step": 5906
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1721496593192833e-06,
+      "loss": 3.663,
+      "step": 5907
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1704935313866715e-06,
+      "loss": 3.5269,
+      "step": 5908
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1688378600014425e-06,
+      "loss": 3.5888,
+      "step": 5909
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.167182645430746e-06,
+      "loss": 3.5738,
+      "step": 5910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1655278879416515e-06,
+      "loss": 3.4945,
+      "step": 5911
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1638735878011603e-06,
+      "loss": 3.5323,
+      "step": 5912
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.162219745276197e-06,
+      "loss": 3.5335,
+      "step": 5913
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1605663606336104e-06,
+      "loss": 3.5609,
+      "step": 5914
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.158913434140182e-06,
+      "loss": 3.4947,
+      "step": 5915
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1572609660626116e-06,
+      "loss": 3.5432,
+      "step": 5916
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1556089566675315e-06,
+      "loss": 3.5302,
+      "step": 5917
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1539574062214965e-06,
+      "loss": 3.5311,
+      "step": 5918
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1523063149909857e-06,
+      "loss": 3.5459,
+      "step": 5919
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.15065568324241e-06,
+      "loss": 3.5633,
+      "step": 5920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1490055112421e-06,
+      "loss": 3.5797,
+      "step": 5921
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.147355799256317e-06,
+      "loss": 3.4699,
+      "step": 5922
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1457065475512433e-06,
+      "loss": 3.601,
+      "step": 5923
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.144057756392992e-06,
+      "loss": 3.5012,
+      "step": 5924
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.142409426047597e-06,
+      "loss": 3.6486,
+      "step": 5925
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1407615567810226e-06,
+      "loss": 3.6289,
+      "step": 5926
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1391141488591536e-06,
+      "loss": 3.4883,
+      "step": 5927
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.137467202547806e-06,
+      "loss": 3.5604,
+      "step": 5928
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1358207181127154e-06,
+      "loss": 3.4418,
+      "step": 5929
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1341746958195493e-06,
+      "loss": 3.4886,
+      "step": 5930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.132529135933893e-06,
+      "loss": 3.6005,
+      "step": 5931
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.130884038721265e-06,
+      "loss": 3.4757,
+      "step": 5932
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.129239404447103e-06,
+      "loss": 3.5079,
+      "step": 5933
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1275952333767756e-06,
+      "loss": 3.6428,
+      "step": 5934
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.12595152577557e-06,
+      "loss": 3.5646,
+      "step": 5935
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1243082819087046e-06,
+      "loss": 3.5082,
+      "step": 5936
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.122665502041319e-06,
+      "loss": 3.4828,
+      "step": 5937
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1210231864384816e-06,
+      "loss": 3.4497,
+      "step": 5938
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.119381335365181e-06,
+      "loss": 3.5499,
+      "step": 5939
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1177399490863373e-06,
+      "loss": 3.4865,
+      "step": 5940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1160990278667874e-06,
+      "loss": 3.5425,
+      "step": 5941
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1144585719713024e-06,
+      "loss": 3.5278,
+      "step": 5942
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1128185816645696e-06,
+      "loss": 3.5901,
+      "step": 5943
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.111179057211209e-06,
+      "loss": 3.5124,
+      "step": 5944
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1095399988757574e-06,
+      "loss": 3.5077,
+      "step": 5945
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.107901406922685e-06,
+      "loss": 3.5792,
+      "step": 5946
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.106263281616378e-06,
+      "loss": 3.5095,
+      "step": 5947
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1046256232211565e-06,
+      "loss": 3.5686,
+      "step": 5948
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1029884320012557e-06,
+      "loss": 3.4819,
+      "step": 5949
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1013517082208444e-06,
+      "loss": 3.6615,
+      "step": 5950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.09971545214401e-06,
+      "loss": 3.6482,
+      "step": 5951
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.098079664034764e-06,
+      "loss": 3.5403,
+      "step": 5952
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0964443441570483e-06,
+      "loss": 3.5594,
+      "step": 5953
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.094809492774723e-06,
+      "loss": 3.6173,
+      "step": 5954
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.093175110151575e-06,
+      "loss": 3.4587,
+      "step": 5955
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0915411965513182e-06,
+      "loss": 3.548,
+      "step": 5956
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.089907752237585e-06,
+      "loss": 3.4637,
+      "step": 5957
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0882747774739385e-06,
+      "loss": 3.524,
+      "step": 5958
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.08664227252386e-06,
+      "loss": 3.4919,
+      "step": 5959
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0850102376507616e-06,
+      "loss": 3.5142,
+      "step": 5960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0833786731179712e-06,
+      "loss": 3.4699,
+      "step": 5961
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0817475791887505e-06,
+      "loss": 3.5592,
+      "step": 5962
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0801169561262756e-06,
+      "loss": 3.6012,
+      "step": 5963
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0784868041936555e-06,
+      "loss": 3.6898,
+      "step": 5964
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0768571236539153e-06,
+      "loss": 3.5584,
+      "step": 5965
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0752279147700113e-06,
+      "loss": 3.413,
+      "step": 5966
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0735991778048158e-06,
+      "loss": 3.5244,
+      "step": 5967
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.071970913021134e-06,
+      "loss": 3.5134,
+      "step": 5968
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.070343120681686e-06,
+      "loss": 3.5664,
+      "step": 5969
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.068715801049123e-06,
+      "loss": 3.4679,
+      "step": 5970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0670889543860145e-06,
+      "loss": 3.6199,
+      "step": 5971
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.065462580954858e-06,
+      "loss": 3.4578,
+      "step": 5972
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0638366810180698e-06,
+      "loss": 3.5502,
+      "step": 5973
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0622112548379963e-06,
+      "loss": 3.5743,
+      "step": 5974
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0605863026768995e-06,
+      "loss": 3.5273,
+      "step": 5975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.058961824796974e-06,
+      "loss": 3.6347,
+      "step": 5976
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0573378214603277e-06,
+      "loss": 3.486,
+      "step": 5977
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0557142929290027e-06,
+      "loss": 3.6174,
+      "step": 5978
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0540912394649542e-06,
+      "loss": 3.4628,
+      "step": 5979
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0524686613300692e-06,
+      "loss": 3.5646,
+      "step": 5980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.050846558786152e-06,
+      "loss": 3.5213,
+      "step": 5981
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0492249320949355e-06,
+      "loss": 3.5273,
+      "step": 5982
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0476037815180684e-06,
+      "loss": 3.4805,
+      "step": 5983
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0459831073171323e-06,
+      "loss": 3.5371,
+      "step": 5984
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0443629097536216e-06,
+      "loss": 3.6364,
+      "step": 5985
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0427431890889637e-06,
+      "loss": 3.4325,
+      "step": 5986
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0411239455845016e-06,
+      "loss": 3.4314,
+      "step": 5987
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.039505179501503e-06,
+      "loss": 3.5249,
+      "step": 5988
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0378868911011624e-06,
+      "loss": 3.438,
+      "step": 5989
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.036269080644593e-06,
+      "loss": 3.4754,
+      "step": 5990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0346517483928302e-06,
+      "loss": 3.4562,
+      "step": 5991
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0330348946068377e-06,
+      "loss": 3.5154,
+      "step": 5992
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0314185195474955e-06,
+      "loss": 3.5577,
+      "step": 5993
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.029802623475613e-06,
+      "loss": 3.5589,
+      "step": 5994
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.028187206651915e-06,
+      "loss": 3.5169,
+      "step": 5995
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.026572269337056e-06,
+      "loss": 3.4926,
+      "step": 5996
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.024957811791606e-06,
+      "loss": 3.5221,
+      "step": 5997
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0233438342760664e-06,
+      "loss": 3.5505,
+      "step": 5998
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.021730337050852e-06,
+      "loss": 3.4945,
+      "step": 5999
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0201173203763068e-06,
+      "loss": 3.5333,
+      "step": 6000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0185047845126928e-06,
+      "loss": 3.5371,
+      "step": 6001
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0168927297201986e-06,
+      "loss": 3.5314,
+      "step": 6002
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0152811562589298e-06,
+      "loss": 3.5742,
+      "step": 6003
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0136700643889212e-06,
+      "loss": 3.5201,
+      "step": 6004
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.012059454370123e-06,
+      "loss": 3.6536,
+      "step": 6005
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.010449326462414e-06,
+      "loss": 3.471,
+      "step": 6006
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.008839680925588e-06,
+      "loss": 3.4438,
+      "step": 6007
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0072305180193695e-06,
+      "loss": 3.6345,
+      "step": 6008
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.005621838003397e-06,
+      "loss": 3.5419,
+      "step": 6009
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0040136411372376e-06,
+      "loss": 3.4944,
+      "step": 6010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0024059276803742e-06,
+      "loss": 3.5054,
+      "step": 6011
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.000798697892219e-06,
+      "loss": 3.523,
+      "step": 6012
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9991919520320978e-06,
+      "loss": 3.517,
+      "step": 6013
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.997585690359268e-06,
+      "loss": 3.5201,
+      "step": 6014
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.995979913132898e-06,
+      "loss": 3.5362,
+      "step": 6015
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.994374620612089e-06,
+      "loss": 3.4927,
+      "step": 6016
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.992769813055854e-06,
+      "loss": 3.4537,
+      "step": 6017
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9911654907231367e-06,
+      "loss": 3.4735,
+      "step": 6018
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9895616538727934e-06,
+      "loss": 3.4348,
+      "step": 6019
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9879583027636116e-06,
+      "loss": 3.6487,
+      "step": 6020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9863554376542937e-06,
+      "loss": 3.4651,
+      "step": 6021
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9847530588034635e-06,
+      "loss": 3.4287,
+      "step": 6022
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.983151166469672e-06,
+      "loss": 3.6236,
+      "step": 6023
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.981549760911386e-06,
+      "loss": 3.5196,
+      "step": 6024
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9799488423869977e-06,
+      "loss": 3.4499,
+      "step": 6025
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.978348411154816e-06,
+      "loss": 3.6623,
+      "step": 6026
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9767484674730787e-06,
+      "loss": 3.5123,
+      "step": 6027
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9751490115999377e-06,
+      "loss": 3.5667,
+      "step": 6028
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.973550043793467e-06,
+      "loss": 3.4291,
+      "step": 6029
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.971951564311668e-06,
+      "loss": 3.5269,
+      "step": 6030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.970353573412456e-06,
+      "loss": 3.5957,
+      "step": 6031
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9687560713536707e-06,
+      "loss": 3.5314,
+      "step": 6032
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9671590583930744e-06,
+      "loss": 3.5315,
+      "step": 6033
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.965562534788347e-06,
+      "loss": 3.504,
+      "step": 6034
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9639665007970925e-06,
+      "loss": 3.5541,
+      "step": 6035
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9623709566768333e-06,
+      "loss": 3.4906,
+      "step": 6036
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.960775902685017e-06,
+      "loss": 3.5608,
+      "step": 6037
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.959181339079005e-06,
+      "loss": 3.4921,
+      "step": 6038
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9575872661160883e-06,
+      "loss": 3.5317,
+      "step": 6039
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9559936840534698e-06,
+      "loss": 3.5626,
+      "step": 6040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.954400593148282e-06,
+      "loss": 3.6249,
+      "step": 6041
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9528079936575693e-06,
+      "loss": 3.5688,
+      "step": 6042
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9512158858383064e-06,
+      "loss": 3.4749,
+      "step": 6043
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.949624269947378e-06,
+      "loss": 3.4963,
+      "step": 6044
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.948033146241601e-06,
+      "loss": 3.5839,
+      "step": 6045
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9464425149777005e-06,
+      "loss": 3.5296,
+      "step": 6046
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9448523764123346e-06,
+      "loss": 3.4945,
+      "step": 6047
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.94326273080207e-06,
+      "loss": 3.4057,
+      "step": 6048
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.941673578403405e-06,
+      "loss": 3.4816,
+      "step": 6049
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.940084919472749e-06,
+      "loss": 3.5102,
+      "step": 6050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.938496754266439e-06,
+      "loss": 3.5864,
+      "step": 6051
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9369090830407254e-06,
+      "loss": 3.565,
+      "step": 6052
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.935321906051787e-06,
+      "loss": 3.5593,
+      "step": 6053
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9337352235557144e-06,
+      "loss": 3.5586,
+      "step": 6054
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9321490358085253e-06,
+      "loss": 3.5648,
+      "step": 6055
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9305633430661523e-06,
+      "loss": 3.578,
+      "step": 6056
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9289781455844543e-06,
+      "loss": 3.5471,
+      "step": 6057
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9273934436192014e-06,
+      "loss": 3.566,
+      "step": 6058
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9258092374260946e-06,
+      "loss": 3.4748,
+      "step": 6059
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.924225527260744e-06,
+      "loss": 3.4755,
+      "step": 6060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9226423133786886e-06,
+      "loss": 3.5249,
+      "step": 6061
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.921059596035381e-06,
+      "loss": 3.5194,
+      "step": 6062
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9194773754861997e-06,
+      "loss": 3.5203,
+      "step": 6063
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9178956519864367e-06,
+      "loss": 3.5702,
+      "step": 6064
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9163144257913057e-06,
+      "loss": 3.5245,
+      "step": 6065
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.914733697155945e-06,
+      "loss": 3.4202,
+      "step": 6066
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.913153466335407e-06,
+      "loss": 3.4681,
+      "step": 6067
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9115737335846636e-06,
+      "loss": 3.5033,
+      "step": 6068
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.909994499158611e-06,
+      "loss": 3.5075,
+      "step": 6069
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9084157633120604e-06,
+      "loss": 3.5005,
+      "step": 6070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9068375262997475e-06,
+      "loss": 3.5276,
+      "step": 6071
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9052597883763196e-06,
+      "loss": 3.5446,
+      "step": 6072
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9036825497963541e-06,
+      "loss": 3.5177,
+      "step": 6073
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9021058108143365e-06,
+      "loss": 3.4555,
+      "step": 6074
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9005295716846822e-06,
+      "loss": 3.5012,
+      "step": 6075
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.898953832661717e-06,
+      "loss": 3.5913,
+      "step": 6076
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.897378593999693e-06,
+      "loss": 3.5644,
+      "step": 6077
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8958038559527757e-06,
+      "loss": 3.5447,
+      "step": 6078
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8942296187750564e-06,
+      "loss": 3.5433,
+      "step": 6079
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8926558827205376e-06,
+      "loss": 3.5541,
+      "step": 6080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.89108264804315e-06,
+      "loss": 3.5572,
+      "step": 6081
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8895099149967338e-06,
+      "loss": 3.6091,
+      "step": 6082
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8879376838350571e-06,
+      "loss": 3.5072,
+      "step": 6083
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8863659548118002e-06,
+      "loss": 3.5175,
+      "step": 6084
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8847947281805674e-06,
+      "loss": 3.5178,
+      "step": 6085
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8832240041948774e-06,
+      "loss": 3.6513,
+      "step": 6086
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8816537831081728e-06,
+      "loss": 3.5434,
+      "step": 6087
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8800840651738096e-06,
+      "loss": 3.616,
+      "step": 6088
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8785148506450679e-06,
+      "loss": 3.592,
+      "step": 6089
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8769461397751414e-06,
+      "loss": 3.4396,
+      "step": 6090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.875377932817149e-06,
+      "loss": 3.4392,
+      "step": 6091
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8738102300241201e-06,
+      "loss": 3.5059,
+      "step": 6092
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8722430316490114e-06,
+      "loss": 3.5195,
+      "step": 6093
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8706763379446897e-06,
+      "loss": 3.4784,
+      "step": 6094
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.869110149163949e-06,
+      "loss": 3.6119,
+      "step": 6095
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8675444655594932e-06,
+      "loss": 3.5309,
+      "step": 6096
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8659792873839532e-06,
+      "loss": 3.5133,
+      "step": 6097
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8644146148898701e-06,
+      "loss": 3.5802,
+      "step": 6098
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.862850448329711e-06,
+      "loss": 3.4906,
+      "step": 6099
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.861286787955856e-06,
+      "loss": 3.4869,
+      "step": 6100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8597236340206038e-06,
+      "loss": 3.5627,
+      "step": 6101
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8581609867761757e-06,
+      "loss": 3.4827,
+      "step": 6102
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8565988464747075e-06,
+      "loss": 3.6146,
+      "step": 6103
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.855037213368252e-06,
+      "loss": 3.6047,
+      "step": 6104
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.853476087708786e-06,
+      "loss": 3.5228,
+      "step": 6105
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8519154697481966e-06,
+      "loss": 3.4957,
+      "step": 6106
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8503553597382973e-06,
+      "loss": 3.5009,
+      "step": 6107
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8487957579308119e-06,
+      "loss": 3.5736,
+      "step": 6108
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8472366645773892e-06,
+      "loss": 3.4499,
+      "step": 6109
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8456780799295888e-06,
+      "loss": 3.5502,
+      "step": 6110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.844120004238895e-06,
+      "loss": 3.4101,
+      "step": 6111
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8425624377567042e-06,
+      "loss": 3.6003,
+      "step": 6112
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8410053807343363e-06,
+      "loss": 3.5182,
+      "step": 6113
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8394488334230232e-06,
+      "loss": 3.5571,
+      "step": 6114
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.83789279607392e-06,
+      "loss": 3.6546,
+      "step": 6115
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8363372689380943e-06,
+      "loss": 3.53,
+      "step": 6116
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8347822522665371e-06,
+      "loss": 3.5313,
+      "step": 6117
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.83322774631015e-06,
+      "loss": 3.4976,
+      "step": 6118
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8316737513197602e-06,
+      "loss": 3.5334,
+      "step": 6119
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8301202675461044e-06,
+      "loss": 3.6448,
+      "step": 6120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8285672952398448e-06,
+      "loss": 3.5376,
+      "step": 6121
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.827014834651553e-06,
+      "loss": 3.457,
+      "step": 6122
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8254628860317258e-06,
+      "loss": 3.563,
+      "step": 6123
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8239114496307708e-06,
+      "loss": 3.4613,
+      "step": 6124
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8223605256990184e-06,
+      "loss": 3.5426,
+      "step": 6125
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8208101144867107e-06,
+      "loss": 3.4448,
+      "step": 6126
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8192602162440137e-06,
+      "loss": 3.6315,
+      "step": 6127
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8177108312210034e-06,
+      "loss": 3.5742,
+      "step": 6128
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.81616195966768e-06,
+      "loss": 3.5753,
+      "step": 6129
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8146136018339544e-06,
+      "loss": 3.501,
+      "step": 6130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.813065757969661e-06,
+      "loss": 3.4961,
+      "step": 6131
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8115184283245441e-06,
+      "loss": 3.5161,
+      "step": 6132
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8099716131482726e-06,
+      "loss": 3.4848,
+      "step": 6133
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8084253126904255e-06,
+      "loss": 3.4482,
+      "step": 6134
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8068795272005058e-06,
+      "loss": 3.5822,
+      "step": 6135
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8053342569279264e-06,
+      "loss": 3.4468,
+      "step": 6136
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8037895021220198e-06,
+      "loss": 3.5241,
+      "step": 6137
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8022452630320387e-06,
+      "loss": 3.4674,
+      "step": 6138
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8007015399071464e-06,
+      "loss": 3.5379,
+      "step": 6139
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7991583329964291e-06,
+      "loss": 3.6035,
+      "step": 6140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7976156425488855e-06,
+      "loss": 3.3823,
+      "step": 6141
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7960734688134307e-06,
+      "loss": 3.584,
+      "step": 6142
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.794531812038901e-06,
+      "loss": 3.6222,
+      "step": 6143
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7929906724740431e-06,
+      "loss": 3.571,
+      "step": 6144
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.791450050367527e-06,
+      "loss": 3.5262,
+      "step": 6145
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7899099459679314e-06,
+      "loss": 3.4726,
+      "step": 6146
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7883703595237601e-06,
+      "loss": 3.5886,
+      "step": 6147
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7868312912834245e-06,
+      "loss": 3.6942,
+      "step": 6148
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7852927414952613e-06,
+      "loss": 3.4993,
+      "step": 6149
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7837547104075154e-06,
+      "loss": 3.5273,
+      "step": 6150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7822171982683544e-06,
+      "loss": 3.3869,
+      "step": 6151
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7806802053258566e-06,
+      "loss": 3.4683,
+      "step": 6152
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7791437318280219e-06,
+      "loss": 3.562,
+      "step": 6153
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7776077780227619e-06,
+      "loss": 3.5563,
+      "step": 6154
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7760723441579086e-06,
+      "loss": 3.5862,
+      "step": 6155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7745374304812047e-06,
+      "loss": 3.5073,
+      "step": 6156
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7730030372403156e-06,
+      "loss": 3.6011,
+      "step": 6157
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7714691646828153e-06,
+      "loss": 3.5384,
+      "step": 6158
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7699358130562017e-06,
+      "loss": 3.5089,
+      "step": 6159
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7684029826078807e-06,
+      "loss": 3.5062,
+      "step": 6160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7668706735851814e-06,
+      "loss": 3.5817,
+      "step": 6161
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7653388862353427e-06,
+      "loss": 3.4305,
+      "step": 6162
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763807620805525e-06,
+      "loss": 3.5282,
+      "step": 6163
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.762276877542798e-06,
+      "loss": 3.5096,
+      "step": 6164
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7607466566941534e-06,
+      "loss": 3.5505,
+      "step": 6165
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7592169585064939e-06,
+      "loss": 3.6804,
+      "step": 6166
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.757687783226642e-06,
+      "loss": 3.6134,
+      "step": 6167
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7561591311013309e-06,
+      "loss": 3.6245,
+      "step": 6168
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7546310023772157e-06,
+      "loss": 3.6349,
+      "step": 6169
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7531033973008593e-06,
+      "loss": 3.4794,
+      "step": 6170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7515763161187482e-06,
+      "loss": 3.5139,
+      "step": 6171
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7500497590772786e-06,
+      "loss": 3.5072,
+      "step": 6172
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7485237264227628e-06,
+      "loss": 3.4666,
+      "step": 6173
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7469982184014323e-06,
+      "loss": 3.5361,
+      "step": 6174
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7454732352594289e-06,
+      "loss": 3.458,
+      "step": 6175
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7439487772428142e-06,
+      "loss": 3.5488,
+      "step": 6176
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7424248445975628e-06,
+      "loss": 3.6047,
+      "step": 6177
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7409014375695616e-06,
+      "loss": 3.478,
+      "step": 6178
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7393785564046206e-06,
+      "loss": 3.4991,
+      "step": 6179
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7378562013484558e-06,
+      "loss": 3.6186,
+      "step": 6180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7363343726467064e-06,
+      "loss": 3.5128,
+      "step": 6181
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7348130705449195e-06,
+      "loss": 3.5403,
+      "step": 6182
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7332922952885645e-06,
+      "loss": 3.6475,
+      "step": 6183
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7317720471230182e-06,
+      "loss": 3.3959,
+      "step": 6184
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7302523262935793e-06,
+      "loss": 3.5992,
+      "step": 6185
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7287331330454555e-06,
+      "loss": 3.5026,
+      "step": 6186
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.727214467623775e-06,
+      "loss": 3.433,
+      "step": 6187
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7256963302735752e-06,
+      "loss": 3.535,
+      "step": 6188
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7241787212398136e-06,
+      "loss": 3.4914,
+      "step": 6189
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7226616407673568e-06,
+      "loss": 3.6261,
+      "step": 6190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.721145089100993e-06,
+      "loss": 3.4521,
+      "step": 6191
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7196290664854176e-06,
+      "loss": 3.4289,
+      "step": 6192
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7181135731652482e-06,
+      "loss": 3.4931,
+      "step": 6193
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.716598609385009e-06,
+      "loss": 3.619,
+      "step": 6194
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7150841753891468e-06,
+      "loss": 3.5202,
+      "step": 6195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7135702714220154e-06,
+      "loss": 3.3841,
+      "step": 6196
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7120568977278896e-06,
+      "loss": 3.5206,
+      "step": 6197
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7105440545509533e-06,
+      "loss": 3.5564,
+      "step": 6198
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7090317421353104e-06,
+      "loss": 3.4989,
+      "step": 6199
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.707519960724972e-06,
+      "loss": 3.393,
+      "step": 6200
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7060087105638718e-06,
+      "loss": 3.6019,
+      "step": 6201
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7044979918958493e-06,
+      "loss": 3.5537,
+      "step": 6202
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7029878049646658e-06,
+      "loss": 3.4589,
+      "step": 6203
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.701478150013991e-06,
+      "loss": 3.5191,
+      "step": 6204
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6999690272874135e-06,
+      "loss": 3.508,
+      "step": 6205
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6984604370284314e-06,
+      "loss": 3.6037,
+      "step": 6206
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6969523794804616e-06,
+      "loss": 3.4738,
+      "step": 6207
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6954448548868297e-06,
+      "loss": 3.5417,
+      "step": 6208
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6939378634907815e-06,
+      "loss": 3.4732,
+      "step": 6209
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6924314055354724e-06,
+      "loss": 3.5366,
+      "step": 6210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6909254812639702e-06,
+      "loss": 3.6409,
+      "step": 6211
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6894200909192632e-06,
+      "loss": 3.5519,
+      "step": 6212
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6879152347442484e-06,
+      "loss": 3.4727,
+      "step": 6213
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6864109129817352e-06,
+      "loss": 3.5224,
+      "step": 6214
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6849071258744532e-06,
+      "loss": 3.5123,
+      "step": 6215
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6834038736650388e-06,
+      "loss": 3.5302,
+      "step": 6216
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.681901156596048e-06,
+      "loss": 3.5592,
+      "step": 6217
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6803989749099441e-06,
+      "loss": 3.5601,
+      "step": 6218
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6788973288491122e-06,
+      "loss": 3.5269,
+      "step": 6219
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6773962186558418e-06,
+      "loss": 3.6443,
+      "step": 6220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6758956445723446e-06,
+      "loss": 3.5713,
+      "step": 6221
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6743956068407373e-06,
+      "loss": 3.5703,
+      "step": 6222
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.672896105703059e-06,
+      "loss": 3.5683,
+      "step": 6223
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6713971414012536e-06,
+      "loss": 3.5748,
+      "step": 6224
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6698987141771856e-06,
+      "loss": 3.5468,
+      "step": 6225
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6684008242726273e-06,
+      "loss": 3.4721,
+      "step": 6226
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6669034719292693e-06,
+      "loss": 3.5032,
+      "step": 6227
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.665406657388709e-06,
+      "loss": 3.4386,
+      "step": 6228
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6639103808924655e-06,
+      "loss": 3.4883,
+      "step": 6229
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.662414642681962e-06,
+      "loss": 3.5236,
+      "step": 6230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6609194429985437e-06,
+      "loss": 3.6259,
+      "step": 6231
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6594247820834603e-06,
+      "loss": 3.4523,
+      "step": 6232
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.657930660177882e-06,
+      "loss": 3.5475,
+      "step": 6233
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6564370775228866e-06,
+      "loss": 3.4725,
+      "step": 6234
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6549440343594692e-06,
+      "loss": 3.4988,
+      "step": 6235
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6534515309285333e-06,
+      "loss": 3.5475,
+      "step": 6236
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6519595674709006e-06,
+      "loss": 3.5079,
+      "step": 6237
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6504681442272996e-06,
+      "loss": 3.5891,
+      "step": 6238
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.648977261438378e-06,
+      "loss": 3.5557,
+      "step": 6239
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.64748691934469e-06,
+      "loss": 3.5501,
+      "step": 6240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6459971181867097e-06,
+      "loss": 3.5593,
+      "step": 6241
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6445078582048158e-06,
+      "loss": 3.4442,
+      "step": 6242
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6430191396393069e-06,
+      "loss": 3.4037,
+      "step": 6243
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6415309627303888e-06,
+      "loss": 3.6043,
+      "step": 6244
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6400433277181854e-06,
+      "loss": 3.5442,
+      "step": 6245
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.638556234842728e-06,
+      "loss": 3.5191,
+      "step": 6246
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6370696843439605e-06,
+      "loss": 3.5855,
+      "step": 6247
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6355836764617455e-06,
+      "loss": 3.5006,
+      "step": 6248
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.634098211435851e-06,
+      "loss": 3.5989,
+      "step": 6249
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.63261328950596e-06,
+      "loss": 3.52,
+      "step": 6250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6311289109116702e-06,
+      "loss": 3.5269,
+      "step": 6251
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6296450758924864e-06,
+      "loss": 3.6733,
+      "step": 6252
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6281617846878323e-06,
+      "loss": 3.4899,
+      "step": 6253
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6266790375370368e-06,
+      "loss": 3.5391,
+      "step": 6254
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6251968346793485e-06,
+      "loss": 3.4654,
+      "step": 6255
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.62371517635392e-06,
+      "loss": 3.3592,
+      "step": 6256
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6222340627998246e-06,
+      "loss": 3.5979,
+      "step": 6257
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6207534942560393e-06,
+      "loss": 3.4316,
+      "step": 6258
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.619273470961461e-06,
+      "loss": 3.4103,
+      "step": 6259
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.617793993154891e-06,
+      "loss": 3.5539,
+      "step": 6260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6163150610750505e-06,
+      "loss": 3.6389,
+      "step": 6261
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6148366749605648e-06,
+      "loss": 3.5377,
+      "step": 6262
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6133588350499785e-06,
+      "loss": 3.5141,
+      "step": 6263
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6118815415817408e-06,
+      "loss": 3.5129,
+      "step": 6264
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6104047947942197e-06,
+      "loss": 3.5798,
+      "step": 6265
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6089285949256888e-06,
+      "loss": 3.5922,
+      "step": 6266
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6074529422143398e-06,
+      "loss": 3.5338,
+      "step": 6267
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6059778368982687e-06,
+      "loss": 3.5014,
+      "step": 6268
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6045032792154907e-06,
+      "loss": 3.5522,
+      "step": 6269
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6030292694039262e-06,
+      "loss": 3.6063,
+      "step": 6270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6015558077014127e-06,
+      "loss": 3.6174,
+      "step": 6271
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6000828943456936e-06,
+      "loss": 3.6224,
+      "step": 6272
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5986105295744302e-06,
+      "loss": 3.426,
+      "step": 6273
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5971387136251892e-06,
+      "loss": 3.5041,
+      "step": 6274
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5956674467354538e-06,
+      "loss": 3.442,
+      "step": 6275
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5941967291426137e-06,
+      "loss": 3.4975,
+      "step": 6276
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5927265610839754e-06,
+      "loss": 3.5383,
+      "step": 6277
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5912569427967505e-06,
+      "loss": 3.5091,
+      "step": 6278
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5897878745180696e-06,
+      "loss": 3.549,
+      "step": 6279
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5883193564849658e-06,
+      "loss": 3.6265,
+      "step": 6280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5868513889343917e-06,
+      "loss": 3.4383,
+      "step": 6281
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5853839721032054e-06,
+      "loss": 3.5572,
+      "step": 6282
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5839171062281771e-06,
+      "loss": 3.5697,
+      "step": 6283
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5824507915459909e-06,
+      "loss": 3.4792,
+      "step": 6284
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.58098502829324e-06,
+      "loss": 3.5544,
+      "step": 6285
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5795198167064252e-06,
+      "loss": 3.3386,
+      "step": 6286
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5780551570219666e-06,
+      "loss": 3.4798,
+      "step": 6287
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5765910494761866e-06,
+      "loss": 3.5741,
+      "step": 6288
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5751274943053253e-06,
+      "loss": 3.6149,
+      "step": 6289
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.573664491745528e-06,
+      "loss": 3.6539,
+      "step": 6290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.572202042032856e-06,
+      "loss": 3.5367,
+      "step": 6291
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5707401454032767e-06,
+      "loss": 3.5107,
+      "step": 6292
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.569278802092673e-06,
+      "loss": 3.502,
+      "step": 6293
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5678180123368326e-06,
+      "loss": 3.5113,
+      "step": 6294
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.566357776371461e-06,
+      "loss": 3.4997,
+      "step": 6295
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.564898094432168e-06,
+      "loss": 3.5245,
+      "step": 6296
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5634389667544785e-06,
+      "loss": 3.5488,
+      "step": 6297
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5619803935738237e-06,
+      "loss": 3.6023,
+      "step": 6298
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5605223751255511e-06,
+      "loss": 3.5711,
+      "step": 6299
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5590649116449113e-06,
+      "loss": 3.6189,
+      "step": 6300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5576080033670732e-06,
+      "loss": 3.5708,
+      "step": 6301
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5561516505271096e-06,
+      "loss": 3.5275,
+      "step": 6302
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.554695853360008e-06,
+      "loss": 3.5892,
+      "step": 6303
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5532406121006632e-06,
+      "loss": 3.5427,
+      "step": 6304
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5517859269838841e-06,
+      "loss": 3.4715,
+      "step": 6305
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5503317982443844e-06,
+      "loss": 3.5605,
+      "step": 6306
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5488782261167945e-06,
+      "loss": 3.4891,
+      "step": 6307
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5474252108356475e-06,
+      "loss": 3.5594,
+      "step": 6308
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5459727526353951e-06,
+      "loss": 3.5169,
+      "step": 6309
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5445208517503913e-06,
+      "loss": 3.543,
+      "step": 6310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5430695084149062e-06,
+      "loss": 3.4779,
+      "step": 6311
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5416187228631153e-06,
+      "loss": 3.5605,
+      "step": 6312
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.540168495329108e-06,
+      "loss": 3.5915,
+      "step": 6313
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.53871882604688e-06,
+      "loss": 3.5012,
+      "step": 6314
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5372697152503408e-06,
+      "loss": 3.4227,
+      "step": 6315
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5358211631733055e-06,
+      "loss": 3.4497,
+      "step": 6316
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5343731700495046e-06,
+      "loss": 3.484,
+      "step": 6317
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5329257361125727e-06,
+      "loss": 3.5834,
+      "step": 6318
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5314788615960551e-06,
+      "loss": 3.6003,
+      "step": 6319
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.530032546733412e-06,
+      "loss": 3.5312,
+      "step": 6320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5285867917580065e-06,
+      "loss": 3.5011,
+      "step": 6321
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5271415969031172e-06,
+      "loss": 3.5453,
+      "step": 6322
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5256969624019286e-06,
+      "loss": 3.4313,
+      "step": 6323
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5242528884875335e-06,
+      "loss": 3.5716,
+      "step": 6324
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5228093753929401e-06,
+      "loss": 3.7152,
+      "step": 6325
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5213664233510594e-06,
+      "loss": 3.6317,
+      "step": 6326
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5199240325947183e-06,
+      "loss": 3.5016,
+      "step": 6327
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5184822033566471e-06,
+      "loss": 3.5468,
+      "step": 6328
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5170409358694903e-06,
+      "loss": 3.5329,
+      "step": 6329
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5156002303657975e-06,
+      "loss": 3.6492,
+      "step": 6330
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5141600870780327e-06,
+      "loss": 3.6094,
+      "step": 6331
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.512720506238563e-06,
+      "loss": 3.4625,
+      "step": 6332
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5112814880796717e-06,
+      "loss": 3.4879,
+      "step": 6333
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.509843032833544e-06,
+      "loss": 3.5375,
+      "step": 6334
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5084051407322814e-06,
+      "loss": 3.4552,
+      "step": 6335
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5069678120078878e-06,
+      "loss": 3.5422,
+      "step": 6336
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5055310468922836e-06,
+      "loss": 3.5672,
+      "step": 6337
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5040948456172894e-06,
+      "loss": 3.5012,
+      "step": 6338
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.502659208414644e-06,
+      "loss": 3.5241,
+      "step": 6339
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5012241355159868e-06,
+      "loss": 3.5444,
+      "step": 6340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.499789627152874e-06,
+      "loss": 3.5416,
+      "step": 6341
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4983556835567626e-06,
+      "loss": 3.5869,
+      "step": 6342
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.496922304959027e-06,
+      "loss": 3.5192,
+      "step": 6343
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.495489491590943e-06,
+      "loss": 3.5085,
+      "step": 6344
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4940572436837002e-06,
+      "loss": 3.5635,
+      "step": 6345
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4926255614683931e-06,
+      "loss": 3.5866,
+      "step": 6346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4911944451760296e-06,
+      "loss": 3.6381,
+      "step": 6347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.489763895037521e-06,
+      "loss": 3.6079,
+      "step": 6348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4883339112836926e-06,
+      "loss": 3.5575,
+      "step": 6349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.486904494145272e-06,
+      "loss": 3.6675,
+      "step": 6350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4854756438529027e-06,
+      "loss": 3.5467,
+      "step": 6351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4840473606371297e-06,
+      "loss": 3.4426,
+      "step": 6352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.482619644728413e-06,
+      "loss": 3.5225,
+      "step": 6353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.481192496357116e-06,
+      "loss": 3.6153,
+      "step": 6354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4797659157535111e-06,
+      "loss": 3.52,
+      "step": 6355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4783399031477834e-06,
+      "loss": 3.4358,
+      "step": 6356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4769144587700202e-06,
+      "loss": 3.5271,
+      "step": 6357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.475489582850223e-06,
+      "loss": 3.555,
+      "step": 6358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4740652756182982e-06,
+      "loss": 3.416,
+      "step": 6359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.472641537304058e-06,
+      "loss": 3.6055,
+      "step": 6360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4712183681372305e-06,
+      "loss": 3.5105,
+      "step": 6361
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4697957683474434e-06,
+      "loss": 3.6071,
+      "step": 6362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.46837373816424e-06,
+      "loss": 3.5206,
+      "step": 6363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4669522778170642e-06,
+      "loss": 3.6696,
+      "step": 6364
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4655313875352761e-06,
+      "loss": 3.6072,
+      "step": 6365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4641110675481357e-06,
+      "loss": 3.5935,
+      "step": 6366
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.462691318084818e-06,
+      "loss": 3.508,
+      "step": 6367
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4612721393744006e-06,
+      "loss": 3.565,
+      "step": 6368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4598535316458733e-06,
+      "loss": 3.4699,
+      "step": 6369
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4584354951281294e-06,
+      "loss": 3.5973,
+      "step": 6370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4570180300499747e-06,
+      "loss": 3.5289,
+      "step": 6371
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.455601136640118e-06,
+      "loss": 3.5313,
+      "step": 6372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4541848151271814e-06,
+      "loss": 3.5028,
+      "step": 6373
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.452769065739688e-06,
+      "loss": 3.5232,
+      "step": 6374
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4513538887060763e-06,
+      "loss": 3.4796,
+      "step": 6375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4499392842546845e-06,
+      "loss": 3.5287,
+      "step": 6376
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.448525252613765e-06,
+      "loss": 3.5597,
+      "step": 6377
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4471117940114732e-06,
+      "loss": 3.5034,
+      "step": 6378
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4456989086758754e-06,
+      "loss": 3.5228,
+      "step": 6379
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4442865968349424e-06,
+      "loss": 3.4671,
+      "step": 6380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4428748587165565e-06,
+      "loss": 3.5051,
+      "step": 6381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4414636945485012e-06,
+      "loss": 3.4453,
+      "step": 6382
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4400531045584748e-06,
+      "loss": 3.3746,
+      "step": 6383
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.438643088974076e-06,
+      "loss": 3.5029,
+      "step": 6384
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4372336480228172e-06,
+      "loss": 3.4794,
+      "step": 6385
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.435824781932112e-06,
+      "loss": 3.4721,
+      "step": 6386
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4344164909292867e-06,
+      "loss": 3.4754,
+      "step": 6387
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.43300877524157e-06,
+      "loss": 3.5335,
+      "step": 6388
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4316016350961025e-06,
+      "loss": 3.467,
+      "step": 6389
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.430195070719927e-06,
+      "loss": 3.5148,
+      "step": 6390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.428789082339998e-06,
+      "loss": 3.4529,
+      "step": 6391
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4273836701831744e-06,
+      "loss": 3.5673,
+      "step": 6392
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.425978834476221e-06,
+      "loss": 3.5297,
+      "step": 6393
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.424574575445814e-06,
+      "loss": 3.547,
+      "step": 6394
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4231708933185328e-06,
+      "loss": 3.6053,
+      "step": 6395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4217677883208626e-06,
+      "loss": 3.5922,
+      "step": 6396
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4203652606792007e-06,
+      "loss": 3.5228,
+      "step": 6397
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4189633106198458e-06,
+      "loss": 3.5682,
+      "step": 6398
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4175619383690082e-06,
+      "loss": 3.5137,
+      "step": 6399
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4161611441527994e-06,
+      "loss": 3.5752,
+      "step": 6400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4147609281972447e-06,
+      "loss": 3.5599,
+      "step": 6401
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.413361290728268e-06,
+      "loss": 3.5524,
+      "step": 6402
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.411962231971708e-06,
+      "loss": 3.6138,
+      "step": 6403
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4105637521533023e-06,
+      "loss": 3.4807,
+      "step": 6404
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4091658514987022e-06,
+      "loss": 3.5843,
+      "step": 6405
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4077685302334593e-06,
+      "loss": 3.6727,
+      "step": 6406
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4063717885830375e-06,
+      "loss": 3.5118,
+      "step": 6407
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4049756267728009e-06,
+      "loss": 3.4864,
+      "step": 6408
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4035800450280268e-06,
+      "loss": 3.6004,
+      "step": 6409
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4021850435738931e-06,
+      "loss": 3.599,
+      "step": 6410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4007906226354884e-06,
+      "loss": 3.5669,
+      "step": 6411
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.399396782437803e-06,
+      "loss": 3.4319,
+      "step": 6412
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3980035232057398e-06,
+      "loss": 3.404,
+      "step": 6413
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3966108451641014e-06,
+      "loss": 3.5795,
+      "step": 6414
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3952187485376017e-06,
+      "loss": 3.5676,
+      "step": 6415
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3938272335508563e-06,
+      "loss": 3.4784,
+      "step": 6416
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3924363004283926e-06,
+      "loss": 3.4824,
+      "step": 6417
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3910459493946371e-06,
+      "loss": 3.5744,
+      "step": 6418
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3896561806739296e-06,
+      "loss": 3.5485,
+      "step": 6419
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3882669944905098e-06,
+      "loss": 3.5353,
+      "step": 6420
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3868783910685284e-06,
+      "loss": 3.6013,
+      "step": 6421
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3854903706320373e-06,
+      "loss": 3.5566,
+      "step": 6422
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3841029334049999e-06,
+      "loss": 3.5677,
+      "step": 6423
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3827160796112788e-06,
+      "loss": 3.6213,
+      "step": 6424
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3813298094746491e-06,
+      "loss": 3.5978,
+      "step": 6425
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.379944123218786e-06,
+      "loss": 3.6541,
+      "step": 6426
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3785590210672768e-06,
+      "loss": 3.5696,
+      "step": 6427
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3771745032436078e-06,
+      "loss": 3.4663,
+      "step": 6428
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3757905699711743e-06,
+      "loss": 3.5134,
+      "step": 6429
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.374407221473279e-06,
+      "loss": 3.4964,
+      "step": 6430
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3730244579731273e-06,
+      "loss": 3.5533,
+      "step": 6431
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.37164227969383e-06,
+      "loss": 3.6253,
+      "step": 6432
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3702606868584073e-06,
+      "loss": 3.4433,
+      "step": 6433
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3688796796897792e-06,
+      "loss": 3.5078,
+      "step": 6434
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3674992584107783e-06,
+      "loss": 3.5402,
+      "step": 6435
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.366119423244135e-06,
+      "loss": 3.5681,
+      "step": 6436
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3647401744124916e-06,
+      "loss": 3.5518,
+      "step": 6437
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3633615121383915e-06,
+      "loss": 3.4918,
+      "step": 6438
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3619834366442864e-06,
+      "loss": 3.4924,
+      "step": 6439
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3606059481525296e-06,
+      "loss": 3.5622,
+      "step": 6440
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3592290468853853e-06,
+      "loss": 3.504,
+      "step": 6441
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3578527330650165e-06,
+      "loss": 3.5348,
+      "step": 6442
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3564770069134976e-06,
+      "loss": 3.569,
+      "step": 6443
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3551018686528017e-06,
+      "loss": 3.5269,
+      "step": 6444
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3537273185048143e-06,
+      "loss": 3.6054,
+      "step": 6445
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3523533566913194e-06,
+      "loss": 3.4151,
+      "step": 6446
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.350979983434011e-06,
+      "loss": 3.4813,
+      "step": 6447
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3496071989544835e-06,
+      "loss": 3.5866,
+      "step": 6448
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3482350034742413e-06,
+      "loss": 3.5029,
+      "step": 6449
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3468633972146895e-06,
+      "loss": 3.6454,
+      "step": 6450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.345492380397142e-06,
+      "loss": 3.7221,
+      "step": 6451
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3441219532428125e-06,
+      "loss": 3.5589,
+      "step": 6452
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3427521159728264e-06,
+      "loss": 3.5522,
+      "step": 6453
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.341382868808206e-06,
+      "loss": 3.4334,
+      "step": 6454
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.340014211969886e-06,
+      "loss": 3.583,
+      "step": 6455
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3386461456786993e-06,
+      "loss": 3.5974,
+      "step": 6456
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3372786701553897e-06,
+      "loss": 3.484,
+      "step": 6457
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3359117856205995e-06,
+      "loss": 3.5343,
+      "step": 6458
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3345454922948813e-06,
+      "loss": 3.4976,
+      "step": 6459
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3331797903986876e-06,
+      "loss": 3.5416,
+      "step": 6460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.331814680152379e-06,
+      "loss": 3.5111,
+      "step": 6461
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3304501617762178e-06,
+      "loss": 3.6168,
+      "step": 6462
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3290862354903743e-06,
+      "loss": 3.4401,
+      "step": 6463
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3277229015149196e-06,
+      "loss": 3.5097,
+      "step": 6464
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3263601600698294e-06,
+      "loss": 3.4733,
+      "step": 6465
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3249980113749884e-06,
+      "loss": 3.417,
+      "step": 6466
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3236364556501807e-06,
+      "loss": 3.5721,
+      "step": 6467
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.322275493115095e-06,
+      "loss": 3.6014,
+      "step": 6468
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3209151239893292e-06,
+      "loss": 3.572,
+      "step": 6469
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3195553484923778e-06,
+      "loss": 3.616,
+      "step": 6470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3181961668436482e-06,
+      "loss": 3.4943,
+      "step": 6471
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3168375792624433e-06,
+      "loss": 3.4985,
+      "step": 6472
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3154795859679781e-06,
+      "loss": 3.5127,
+      "step": 6473
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.314122187179364e-06,
+      "loss": 3.5632,
+      "step": 6474
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3127653831156245e-06,
+      "loss": 3.5048,
+      "step": 6475
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3114091739956796e-06,
+      "loss": 3.5068,
+      "step": 6476
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.31005356003836e-06,
+      "loss": 3.5605,
+      "step": 6477
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3086985414623938e-06,
+      "loss": 3.5825,
+      "step": 6478
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3073441184864194e-06,
+      "loss": 3.5589,
+      "step": 6479
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3059902913289729e-06,
+      "loss": 3.5414,
+      "step": 6480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3046370602085007e-06,
+      "loss": 3.6321,
+      "step": 6481
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3032844253433486e-06,
+      "loss": 3.6008,
+      "step": 6482
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3019323869517653e-06,
+      "loss": 3.6249,
+      "step": 6483
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3005809452519086e-06,
+      "loss": 3.5151,
+      "step": 6484
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2992301004618336e-06,
+      "loss": 3.5888,
+      "step": 6485
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2978798527995052e-06,
+      "loss": 3.6187,
+      "step": 6486
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2965302024827858e-06,
+      "loss": 3.5935,
+      "step": 6487
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.295181149729448e-06,
+      "loss": 3.5756,
+      "step": 6488
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.293832694757161e-06,
+      "loss": 3.6383,
+      "step": 6489
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2924848377835047e-06,
+      "loss": 3.4718,
+      "step": 6490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2911375790259561e-06,
+      "loss": 3.5749,
+      "step": 6491
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2897909187019008e-06,
+      "loss": 3.5813,
+      "step": 6492
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.288444857028623e-06,
+      "loss": 3.5126,
+      "step": 6493
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2870993942233156e-06,
+      "loss": 3.5942,
+      "step": 6494
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2857545305030688e-06,
+      "loss": 3.5472,
+      "step": 6495
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2844102660848834e-06,
+      "loss": 3.5963,
+      "step": 6496
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2830666011856558e-06,
+      "loss": 3.5639,
+      "step": 6497
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.281723536022193e-06,
+      "loss": 3.446,
+      "step": 6498
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.280381070811198e-06,
+      "loss": 3.6779,
+      "step": 6499
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2790392057692842e-06,
+      "loss": 3.406,
+      "step": 6500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2776979411129609e-06,
+      "loss": 3.5465,
+      "step": 6501
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.276357277058648e-06,
+      "loss": 3.581,
+      "step": 6502
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2750172138226618e-06,
+      "loss": 3.5362,
+      "step": 6503
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2736777516212267e-06,
+      "loss": 3.5709,
+      "step": 6504
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2723388906704676e-06,
+      "loss": 3.5098,
+      "step": 6505
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2710006311864104e-06,
+      "loss": 3.5708,
+      "step": 6506
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2696629733849897e-06,
+      "loss": 3.6189,
+      "step": 6507
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2683259174820383e-06,
+      "loss": 3.6332,
+      "step": 6508
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2669894636932917e-06,
+      "loss": 3.493,
+      "step": 6509
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2656536122343927e-06,
+      "loss": 3.5694,
+      "step": 6510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.264318363320881e-06,
+      "loss": 3.5437,
+      "step": 6511
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2629837171682052e-06,
+      "loss": 3.5226,
+      "step": 6512
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.26164967399171e-06,
+      "loss": 3.4869,
+      "step": 6513
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2603162340066505e-06,
+      "loss": 3.4318,
+      "step": 6514
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2589833974281762e-06,
+      "loss": 3.494,
+      "step": 6515
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2576511644713462e-06,
+      "loss": 3.4068,
+      "step": 6516
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2563195353511176e-06,
+      "loss": 3.447,
+      "step": 6517
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2549885102823545e-06,
+      "loss": 3.5752,
+      "step": 6518
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2536580894798167e-06,
+      "loss": 3.5385,
+      "step": 6519
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2523282731581747e-06,
+      "loss": 3.5339,
+      "step": 6520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2509990615319946e-06,
+      "loss": 3.6245,
+      "step": 6521
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2496704548157495e-06,
+      "loss": 3.4606,
+      "step": 6522
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2483424532238114e-06,
+      "loss": 3.5528,
+      "step": 6523
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2470150569704593e-06,
+      "loss": 3.5666,
+      "step": 6524
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2456882662698677e-06,
+      "loss": 3.5328,
+      "step": 6525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2443620813361212e-06,
+      "loss": 3.4511,
+      "step": 6526
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2430365023832003e-06,
+      "loss": 3.5885,
+      "step": 6527
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2417115296249916e-06,
+      "loss": 3.5698,
+      "step": 6528
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2403871632752806e-06,
+      "loss": 3.4566,
+      "step": 6529
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.23906340354776e-06,
+      "loss": 3.5257,
+      "step": 6530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2377402506560176e-06,
+      "loss": 3.5208,
+      "step": 6531
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2364177048135512e-06,
+      "loss": 3.5357,
+      "step": 6532
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2350957662337526e-06,
+      "loss": 3.5508,
+      "step": 6533
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2337744351299235e-06,
+      "loss": 3.4987,
+      "step": 6534
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2324537117152602e-06,
+      "loss": 3.4652,
+      "step": 6535
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.231133596202868e-06,
+      "loss": 3.4584,
+      "step": 6536
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.229814088805747e-06,
+      "loss": 3.5439,
+      "step": 6537
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2284951897368058e-06,
+      "loss": 3.581,
+      "step": 6538
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.227176899208849e-06,
+      "loss": 3.6246,
+      "step": 6539
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.225859217434589e-06,
+      "loss": 3.4415,
+      "step": 6540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2245421446266348e-06,
+      "loss": 3.4655,
+      "step": 6541
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.223225680997498e-06,
+      "loss": 3.4856,
+      "step": 6542
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2219098267595958e-06,
+      "loss": 3.6313,
+      "step": 6543
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.220594582125243e-06,
+      "loss": 3.4649,
+      "step": 6544
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2192799473066557e-06,
+      "loss": 3.5928,
+      "step": 6545
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2179659225159569e-06,
+      "loss": 3.654,
+      "step": 6546
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2166525079651637e-06,
+      "loss": 3.5541,
+      "step": 6547
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2153397038662013e-06,
+      "loss": 3.5648,
+      "step": 6548
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2140275104308918e-06,
+      "loss": 3.6217,
+      "step": 6549
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2127159278709628e-06,
+      "loss": 3.5427,
+      "step": 6550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2114049563980385e-06,
+      "loss": 3.5436,
+      "step": 6551
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2100945962236505e-06,
+      "loss": 3.6419,
+      "step": 6552
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2087848475592244e-06,
+      "loss": 3.532,
+      "step": 6553
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.207475710616095e-06,
+      "loss": 3.528,
+      "step": 6554
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2061671856054912e-06,
+      "loss": 3.6106,
+      "step": 6555
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2048592727385494e-06,
+      "loss": 3.4505,
+      "step": 6556
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2035519722263023e-06,
+      "loss": 3.48,
+      "step": 6557
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2022452842796872e-06,
+      "loss": 3.5421,
+      "step": 6558
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.20093920910954e-06,
+      "loss": 3.625,
+      "step": 6559
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1996337469266006e-06,
+      "loss": 3.4698,
+      "step": 6560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1983288979415065e-06,
+      "loss": 3.4113,
+      "step": 6561
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1970246623647995e-06,
+      "loss": 3.5103,
+      "step": 6562
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1957210404069192e-06,
+      "loss": 3.5429,
+      "step": 6563
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1944180322782113e-06,
+      "loss": 3.5182,
+      "step": 6564
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1931156381889146e-06,
+      "loss": 3.513,
+      "step": 6565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1918138583491784e-06,
+      "loss": 3.4944,
+      "step": 6566
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1905126929690431e-06,
+      "loss": 3.4394,
+      "step": 6567
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.189212142258459e-06,
+      "loss": 3.6417,
+      "step": 6568
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1879122064272692e-06,
+      "loss": 3.5015,
+      "step": 6569
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1866128856852243e-06,
+      "loss": 3.672,
+      "step": 6570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1853141802419705e-06,
+      "loss": 3.5095,
+      "step": 6571
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1840160903070591e-06,
+      "loss": 3.5282,
+      "step": 6572
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1827186160899372e-06,
+      "loss": 3.5131,
+      "step": 6573
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1814217577999581e-06,
+      "loss": 3.538,
+      "step": 6574
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1801255156463703e-06,
+      "loss": 3.554,
+      "step": 6575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1788298898383281e-06,
+      "loss": 3.5504,
+      "step": 6576
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1775348805848824e-06,
+      "loss": 3.3898,
+      "step": 6577
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1762404880949841e-06,
+      "loss": 3.4857,
+      "step": 6578
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1749467125774893e-06,
+      "loss": 3.4756,
+      "step": 6579
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1736535542411509e-06,
+      "loss": 3.5326,
+      "step": 6580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1723610132946206e-06,
+      "loss": 3.4359,
+      "step": 6581
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1710690899464567e-06,
+      "loss": 3.5397,
+      "step": 6582
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1697777844051105e-06,
+      "loss": 3.5189,
+      "step": 6583
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1684870968789402e-06,
+      "loss": 3.4034,
+      "step": 6584
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1671970275761985e-06,
+      "loss": 3.4665,
+      "step": 6585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1659075767050432e-06,
+      "loss": 3.4795,
+      "step": 6586
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.164618744473528e-06,
+      "loss": 3.5936,
+      "step": 6587
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1633305310896121e-06,
+      "loss": 3.5376,
+      "step": 6588
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.162042936761148e-06,
+      "loss": 3.5206,
+      "step": 6589
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1607559616958952e-06,
+      "loss": 3.6193,
+      "step": 6590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.159469606101507e-06,
+      "loss": 3.5351,
+      "step": 6591
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1581838701855435e-06,
+      "loss": 3.5359,
+      "step": 6592
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1568987541554577e-06,
+      "loss": 3.487,
+      "step": 6593
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1556142582186087e-06,
+      "loss": 3.5605,
+      "step": 6594
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.15433038258225e-06,
+      "loss": 3.4423,
+      "step": 6595
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1530471274535414e-06,
+      "loss": 3.6051,
+      "step": 6596
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.151764493039536e-06,
+      "loss": 3.5454,
+      "step": 6597
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1504824795471919e-06,
+      "loss": 3.4812,
+      "step": 6598
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1492010871833631e-06,
+      "loss": 3.5179,
+      "step": 6599
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1479203161548074e-06,
+      "loss": 3.5105,
+      "step": 6600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.146640166668178e-06,
+      "loss": 3.5461,
+      "step": 6601
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.145360638930032e-06,
+      "loss": 3.5091,
+      "step": 6602
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1440817331468218e-06,
+      "loss": 3.4984,
+      "step": 6603
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.142803449524904e-06,
+      "loss": 3.4303,
+      "step": 6604
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.141525788270531e-06,
+      "loss": 3.6078,
+      "step": 6605
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1402487495898579e-06,
+      "loss": 3.5651,
+      "step": 6606
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1389723336889359e-06,
+      "loss": 3.5776,
+      "step": 6607
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.13769654077372e-06,
+      "loss": 3.626,
+      "step": 6608
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1364213710500594e-06,
+      "loss": 3.6249,
+      "step": 6609
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1351468247237091e-06,
+      "loss": 3.4206,
+      "step": 6610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.133872902000317e-06,
+      "loss": 3.4707,
+      "step": 6611
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1325996030854359e-06,
+      "loss": 3.3866,
+      "step": 6612
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1313269281845146e-06,
+      "loss": 3.4402,
+      "step": 6613
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1300548775029003e-06,
+      "loss": 3.5827,
+      "step": 6614
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1287834512458446e-06,
+      "loss": 3.4969,
+      "step": 6615
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1275126496184919e-06,
+      "loss": 3.5285,
+      "step": 6616
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1262424728258915e-06,
+      "loss": 3.4796,
+      "step": 6617
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1249729210729888e-06,
+      "loss": 3.4622,
+      "step": 6618
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1237039945646267e-06,
+      "loss": 3.4873,
+      "step": 6619
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1224356935055525e-06,
+      "loss": 3.5153,
+      "step": 6620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.121168018100407e-06,
+      "loss": 3.5969,
+      "step": 6621
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1199009685537349e-06,
+      "loss": 3.5373,
+      "step": 6622
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.118634545069975e-06,
+      "loss": 3.6316,
+      "step": 6623
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.11736874785347e-06,
+      "loss": 3.5515,
+      "step": 6624
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1161035771084566e-06,
+      "loss": 3.6189,
+      "step": 6625
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1148390330390767e-06,
+      "loss": 3.4468,
+      "step": 6626
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1135751158493634e-06,
+      "loss": 3.5106,
+      "step": 6627
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1123118257432559e-06,
+      "loss": 3.5579,
+      "step": 6628
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1110491629245863e-06,
+      "loss": 3.5259,
+      "step": 6629
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1097871275970913e-06,
+      "loss": 3.4643,
+      "step": 6630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.108525719964399e-06,
+      "loss": 3.5128,
+      "step": 6631
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.107264940230045e-06,
+      "loss": 3.4805,
+      "step": 6632
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.106004788597455e-06,
+      "loss": 3.4589,
+      "step": 6633
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1047452652699604e-06,
+      "loss": 3.6364,
+      "step": 6634
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1034863704507859e-06,
+      "loss": 3.6427,
+      "step": 6635
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1022281043430595e-06,
+      "loss": 3.5157,
+      "step": 6636
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1009704671498023e-06,
+      "loss": 3.6815,
+      "step": 6637
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.09971345907394e-06,
+      "loss": 3.6046,
+      "step": 6638
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.098457080318291e-06,
+      "loss": 3.5817,
+      "step": 6639
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0972013310855778e-06,
+      "loss": 3.5015,
+      "step": 6640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0959462115784153e-06,
+      "loss": 3.3683,
+      "step": 6641
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0946917219993231e-06,
+      "loss": 3.4354,
+      "step": 6642
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.093437862550713e-06,
+      "loss": 3.6194,
+      "step": 6643
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0921846334349008e-06,
+      "loss": 3.5948,
+      "step": 6644
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0909320348540953e-06,
+      "loss": 3.4982,
+      "step": 6645
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0896800670104085e-06,
+      "loss": 3.4935,
+      "step": 6646
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0884287301058466e-06,
+      "loss": 3.5038,
+      "step": 6647
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0871780243423174e-06,
+      "loss": 3.5661,
+      "step": 6648
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.085927949921624e-06,
+      "loss": 3.5339,
+      "step": 6649
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0846785070454675e-06,
+      "loss": 3.4531,
+      "step": 6650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.083429695915451e-06,
+      "loss": 3.5434,
+      "step": 6651
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0821815167330707e-06,
+      "loss": 3.5293,
+      "step": 6652
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0809339696997263e-06,
+      "loss": 3.5441,
+      "step": 6653
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0796870550167094e-06,
+      "loss": 3.5491,
+      "step": 6654
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0784407728852125e-06,
+      "loss": 3.4279,
+      "step": 6655
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0771951235063288e-06,
+      "loss": 3.5821,
+      "step": 6656
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0759501070810436e-06,
+      "loss": 3.5697,
+      "step": 6657
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.074705723810246e-06,
+      "loss": 3.4799,
+      "step": 6658
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.073461973894717e-06,
+      "loss": 3.5177,
+      "step": 6659
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0722188575351423e-06,
+      "loss": 3.4231,
+      "step": 6660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0709763749320973e-06,
+      "loss": 3.4976,
+      "step": 6661
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0697345262860638e-06,
+      "loss": 3.436,
+      "step": 6662
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0684933117974127e-06,
+      "loss": 3.556,
+      "step": 6663
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.067252731666421e-06,
+      "loss": 3.5471,
+      "step": 6664
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.066012786093255e-06,
+      "loss": 3.4666,
+      "step": 6665
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.064773475277987e-06,
+      "loss": 3.577,
+      "step": 6666
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0635347994205785e-06,
+      "loss": 3.6648,
+      "step": 6667
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0622967587208965e-06,
+      "loss": 3.515,
+      "step": 6668
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0610593533786984e-06,
+      "loss": 3.5577,
+      "step": 6669
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0598225835936454e-06,
+      "loss": 3.6316,
+      "step": 6670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0585864495652899e-06,
+      "loss": 3.5439,
+      "step": 6671
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0573509514930885e-06,
+      "loss": 3.5379,
+      "step": 6672
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0561160895763883e-06,
+      "loss": 3.6115,
+      "step": 6673
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0548818640144403e-06,
+      "loss": 3.6224,
+      "step": 6674
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0536482750063865e-06,
+      "loss": 3.5348,
+      "step": 6675
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0524153227512722e-06,
+      "loss": 3.5035,
+      "step": 6676
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0511830074480346e-06,
+      "loss": 3.6097,
+      "step": 6677
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0499513292955133e-06,
+      "loss": 3.4378,
+      "step": 6678
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.048720288492439e-06,
+      "loss": 3.6045,
+      "step": 6679
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0474898852374466e-06,
+      "loss": 3.6094,
+      "step": 6680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.046260119729061e-06,
+      "loss": 3.5058,
+      "step": 6681
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0450309921657115e-06,
+      "loss": 3.4797,
+      "step": 6682
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0438025027457166e-06,
+      "loss": 3.5954,
+      "step": 6683
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0425746516672996e-06,
+      "loss": 3.5644,
+      "step": 6684
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0413474391285738e-06,
+      "loss": 3.5055,
+      "step": 6685
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0401208653275557e-06,
+      "loss": 3.6079,
+      "step": 6686
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0388949304621549e-06,
+      "loss": 3.6621,
+      "step": 6687
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0376696347301767e-06,
+      "loss": 3.3937,
+      "step": 6688
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0364449783293284e-06,
+      "loss": 3.5677,
+      "step": 6689
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0352209614572101e-06,
+      "loss": 3.5461,
+      "step": 6690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0339975843113182e-06,
+      "loss": 3.4323,
+      "step": 6691
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0327748470890496e-06,
+      "loss": 3.4128,
+      "step": 6692
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0315527499876938e-06,
+      "loss": 3.499,
+      "step": 6693
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0303312932044412e-06,
+      "loss": 3.5032,
+      "step": 6694
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.029110476936374e-06,
+      "loss": 3.4094,
+      "step": 6695
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.027890301380477e-06,
+      "loss": 3.5369,
+      "step": 6696
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.026670766733625e-06,
+      "loss": 3.5465,
+      "step": 6697
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0254518731925956e-06,
+      "loss": 3.5346,
+      "step": 6698
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.024233620954057e-06,
+      "loss": 3.5495,
+      "step": 6699
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0230160102145808e-06,
+      "loss": 3.525,
+      "step": 6700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0217990411706275e-06,
+      "loss": 3.4955,
+      "step": 6701
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0205827140185608e-06,
+      "loss": 3.6044,
+      "step": 6702
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0193670289546353e-06,
+      "loss": 3.4605,
+      "step": 6703
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0181519861750078e-06,
+      "loss": 3.5722,
+      "step": 6704
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0169375858757247e-06,
+      "loss": 3.5421,
+      "step": 6705
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0157238282527355e-06,
+      "loss": 3.4452,
+      "step": 6706
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.01451071350188e-06,
+      "loss": 3.537,
+      "step": 6707
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0132982418188998e-06,
+      "loss": 3.5498,
+      "step": 6708
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0120864133994268e-06,
+      "loss": 3.493,
+      "step": 6709
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0108752284389956e-06,
+      "loss": 3.5121,
+      "step": 6710
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0096646871330307e-06,
+      "loss": 3.6556,
+      "step": 6711
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0084547896768587e-06,
+      "loss": 3.4874,
+      "step": 6712
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0072455362656963e-06,
+      "loss": 3.517,
+      "step": 6713
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0060369270946623e-06,
+      "loss": 3.5324,
+      "step": 6714
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.004828962358766e-06,
+      "loss": 3.4624,
+      "step": 6715
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0036216422529172e-06,
+      "loss": 3.5482,
+      "step": 6716
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.002414966971918e-06,
+      "loss": 3.5559,
+      "step": 6717
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0012089367104705e-06,
+      "loss": 3.5246,
+      "step": 6718
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.000003551663168e-06,
+      "loss": 3.4878,
+      "step": 6719
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.98798812024504e-07,
+      "loss": 3.4724,
+      "step": 6720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.975947179888646e-07,
+      "loss": 3.6305,
+      "step": 6721
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.96391269750535e-07,
+      "loss": 3.5593,
+      "step": 6722
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.951884675036932e-07,
+      "loss": 3.5775,
+      "step": 6723
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.93986311442413e-07,
+      "loss": 3.3486,
+      "step": 6724
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.927848017606672e-07,
+      "loss": 3.4604,
+      "step": 6725
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.915839386523212e-07,
+      "loss": 3.5882,
+      "step": 6726
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.903837223111363e-07,
+      "loss": 3.5076,
+      "step": 6727
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.891841529307716e-07,
+      "loss": 3.498,
+      "step": 6728
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.879852307047783e-07,
+      "loss": 3.5374,
+      "step": 6729
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.86786955826608e-07,
+      "loss": 3.5387,
+      "step": 6730
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.855893284896024e-07,
+      "loss": 3.6419,
+      "step": 6731
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.843923488870038e-07,
+      "loss": 3.5583,
+      "step": 6732
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.831960172119455e-07,
+      "loss": 3.5162,
+      "step": 6733
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.820003336574601e-07,
+      "loss": 3.3851,
+      "step": 6734
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.80805298416473e-07,
+      "loss": 3.5116,
+      "step": 6735
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.796109116818065e-07,
+      "loss": 3.5394,
+      "step": 6736
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.784171736461762e-07,
+      "loss": 3.4926,
+      "step": 6737
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.772240845021968e-07,
+      "loss": 3.4604,
+      "step": 6738
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.760316444423734e-07,
+      "loss": 3.5874,
+      "step": 6739
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.74839853659112e-07,
+      "loss": 3.5305,
+      "step": 6740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.73648712344707e-07,
+      "loss": 3.5082,
+      "step": 6741
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.724582206913558e-07,
+      "loss": 3.6166,
+      "step": 6742
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.71268378891143e-07,
+      "loss": 3.4878,
+      "step": 6743
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.70079187136056e-07,
+      "loss": 3.6041,
+      "step": 6744
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.688906456179709e-07,
+      "loss": 3.4709,
+      "step": 6745
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.677027545286639e-07,
+      "loss": 3.6029,
+      "step": 6746
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.665155140598009e-07,
+      "loss": 3.6023,
+      "step": 6747
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.653289244029497e-07,
+      "loss": 3.4323,
+      "step": 6748
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.641429857495655e-07,
+      "loss": 3.4849,
+      "step": 6749
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.62957698291005e-07,
+      "loss": 3.5354,
+      "step": 6750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.61773062218515e-07,
+      "loss": 3.5064,
+      "step": 6751
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.605890777232423e-07,
+      "loss": 3.4948,
+      "step": 6752
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.594057449962219e-07,
+      "loss": 3.5348,
+      "step": 6753
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.582230642283895e-07,
+      "loss": 3.5841,
+      "step": 6754
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.570410356105725e-07,
+      "loss": 3.4223,
+      "step": 6755
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.558596593334952e-07,
+      "loss": 3.5855,
+      "step": 6756
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.546789355877728e-07,
+      "loss": 3.5561,
+      "step": 6757
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.534988645639209e-07,
+      "loss": 3.6243,
+      "step": 6758
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.52319446452345e-07,
+      "loss": 3.5918,
+      "step": 6759
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.51140681443346e-07,
+      "loss": 3.445,
+      "step": 6760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.499625697271225e-07,
+      "loss": 3.5405,
+      "step": 6761
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.487851114937635e-07,
+      "loss": 3.5383,
+      "step": 6762
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.476083069332565e-07,
+      "loss": 3.4711,
+      "step": 6763
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.464321562354811e-07,
+      "loss": 3.4628,
+      "step": 6764
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.452566595902102e-07,
+      "loss": 3.4724,
+      "step": 6765
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.440818171871152e-07,
+      "loss": 3.682,
+      "step": 6766
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.42907629215758e-07,
+      "loss": 3.5616,
+      "step": 6767
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.417340958655979e-07,
+      "loss": 3.5602,
+      "step": 6768
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.405612173259854e-07,
+      "loss": 3.4972,
+      "step": 6769
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.393889937861694e-07,
+      "loss": 3.6336,
+      "step": 6770
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.382174254352883e-07,
+      "loss": 3.4751,
+      "step": 6771
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.370465124623806e-07,
+      "loss": 3.59,
+      "step": 6772
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.358762550563721e-07,
+      "loss": 3.6808,
+      "step": 6773
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.347066534060901e-07,
+      "loss": 3.4975,
+      "step": 6774
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.335377077002495e-07,
+      "loss": 3.4565,
+      "step": 6775
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.323694181274645e-07,
+      "loss": 3.5986,
+      "step": 6776
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.3120178487624e-07,
+      "loss": 3.4876,
+      "step": 6777
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.300348081349775e-07,
+      "loss": 3.4428,
+      "step": 6778
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.288684880919696e-07,
+      "loss": 3.5059,
+      "step": 6779
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.27702824935407e-07,
+      "loss": 3.5683,
+      "step": 6780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.265378188533697e-07,
+      "loss": 3.5818,
+      "step": 6781
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.253734700338368e-07,
+      "loss": 3.6133,
+      "step": 6782
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.242097786646758e-07,
+      "loss": 3.5532,
+      "step": 6783
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.230467449336544e-07,
+      "loss": 3.5518,
+      "step": 6784
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.218843690284268e-07,
+      "loss": 3.5731,
+      "step": 6785
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.207226511365485e-07,
+      "loss": 3.6016,
+      "step": 6786
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.195615914454625e-07,
+      "loss": 3.665,
+      "step": 6787
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.18401190142511e-07,
+      "loss": 3.6253,
+      "step": 6788
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.172414474149255e-07,
+      "loss": 3.4326,
+      "step": 6789
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.160823634498345e-07,
+      "loss": 3.5594,
+      "step": 6790
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.149239384342573e-07,
+      "loss": 3.5712,
+      "step": 6791
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.137661725551111e-07,
+      "loss": 3.527,
+      "step": 6792
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.12609065999201e-07,
+      "loss": 3.5684,
+      "step": 6793
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.114526189532313e-07,
+      "loss": 3.5287,
+      "step": 6794
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.102968316037964e-07,
+      "loss": 3.4234,
+      "step": 6795
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.091417041373834e-07,
+      "loss": 3.5388,
+      "step": 6796
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.079872367403786e-07,
+      "loss": 3.5515,
+      "step": 6797
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.068334295990544e-07,
+      "loss": 3.5016,
+      "step": 6798
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.056802828995826e-07,
+      "loss": 3.5986,
+      "step": 6799
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.045277968280259e-07,
+      "loss": 3.4697,
+      "step": 6800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.033759715703383e-07,
+      "loss": 3.5961,
+      "step": 6801
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.022248073123724e-07,
+      "loss": 3.5278,
+      "step": 6802
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.010743042398684e-07,
+      "loss": 3.479,
+      "step": 6803
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.999244625384657e-07,
+      "loss": 3.4965,
+      "step": 6804
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.987752823936907e-07,
+      "loss": 3.532,
+      "step": 6805
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.976267639909697e-07,
+      "loss": 3.5674,
+      "step": 6806
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.964789075156149e-07,
+      "loss": 3.5644,
+      "step": 6807
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.953317131528394e-07,
+      "loss": 3.5446,
+      "step": 6808
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.941851810877427e-07,
+      "loss": 3.5294,
+      "step": 6809
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.930393115053226e-07,
+      "loss": 3.5142,
+      "step": 6810
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.918941045904661e-07,
+      "loss": 3.5185,
+      "step": 6811
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.907495605279565e-07,
+      "loss": 3.5585,
+      "step": 6812
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.896056795024661e-07,
+      "loss": 3.5313,
+      "step": 6813
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.88462461698566e-07,
+      "loss": 3.5285,
+      "step": 6814
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.873199073007143e-07,
+      "loss": 3.5147,
+      "step": 6815
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.861780164932671e-07,
+      "loss": 3.656,
+      "step": 6816
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.850367894604689e-07,
+      "loss": 3.5157,
+      "step": 6817
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.838962263864614e-07,
+      "loss": 3.5083,
+      "step": 6818
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.827563274552747e-07,
+      "loss": 3.4586,
+      "step": 6819
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.816170928508367e-07,
+      "loss": 3.5271,
+      "step": 6820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.804785227569628e-07,
+      "loss": 3.575,
+      "step": 6821
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.793406173573671e-07,
+      "loss": 3.5604,
+      "step": 6822
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.782033768356497e-07,
+      "loss": 3.4404,
+      "step": 6823
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.770668013753103e-07,
+      "loss": 3.5051,
+      "step": 6824
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.759308911597353e-07,
+      "loss": 3.4654,
+      "step": 6825
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.74795646372209e-07,
+      "loss": 3.5428,
+      "step": 6826
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.736610671959028e-07,
+      "loss": 3.6058,
+      "step": 6827
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.725271538138869e-07,
+      "loss": 3.5516,
+      "step": 6828
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.713939064091176e-07,
+      "loss": 3.4209,
+      "step": 6829
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.702613251644498e-07,
+      "loss": 3.5887,
+      "step": 6830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.691294102626268e-07,
+      "loss": 3.6296,
+      "step": 6831
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.679981618862842e-07,
+      "loss": 3.5418,
+      "step": 6832
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.66867580217955e-07,
+      "loss": 3.5362,
+      "step": 6833
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.657376654400573e-07,
+      "loss": 3.546,
+      "step": 6834
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.64608417734909e-07,
+      "loss": 3.446,
+      "step": 6835
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.634798372847148e-07,
+      "loss": 3.476,
+      "step": 6836
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.623519242715734e-07,
+      "loss": 3.6106,
+      "step": 6837
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.612246788774775e-07,
+      "loss": 3.5626,
+      "step": 6838
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.600981012843091e-07,
+      "loss": 3.5756,
+      "step": 6839
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.589721916738463e-07,
+      "loss": 3.5884,
+      "step": 6840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.578469502277548e-07,
+      "loss": 3.5774,
+      "step": 6841
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.567223771275973e-07,
+      "loss": 3.5483,
+      "step": 6842
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.555984725548239e-07,
+      "loss": 3.4994,
+      "step": 6843
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.544752366907816e-07,
+      "loss": 3.4765,
+      "step": 6844
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.53352669716705e-07,
+      "loss": 3.4906,
+      "step": 6845
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.522307718137246e-07,
+      "loss": 3.5058,
+      "step": 6846
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.511095431628591e-07,
+      "loss": 3.5208,
+      "step": 6847
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.499889839450237e-07,
+      "loss": 3.4906,
+      "step": 6848
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.488690943410205e-07,
+      "loss": 3.5359,
+      "step": 6849
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.477498745315499e-07,
+      "loss": 3.5022,
+      "step": 6850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.466313246971969e-07,
+      "loss": 3.5377,
+      "step": 6851
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.455134450184449e-07,
+      "loss": 3.5827,
+      "step": 6852
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.443962356756635e-07,
+      "loss": 3.5638,
+      "step": 6853
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.432796968491209e-07,
+      "loss": 3.4904,
+      "step": 6854
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.421638287189687e-07,
+      "loss": 3.6521,
+      "step": 6855
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.410486314652589e-07,
+      "loss": 3.4847,
+      "step": 6856
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.399341052679283e-07,
+      "loss": 3.4476,
+      "step": 6857
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.388202503068099e-07,
+      "loss": 3.5093,
+      "step": 6858
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.377070667616255e-07,
+      "loss": 3.4341,
+      "step": 6859
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.365945548119914e-07,
+      "loss": 3.5066,
+      "step": 6860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.354827146374117e-07,
+      "loss": 3.5595,
+      "step": 6861
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.343715464172874e-07,
+      "loss": 3.5097,
+      "step": 6862
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.332610503309046e-07,
+      "loss": 3.5897,
+      "step": 6863
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.321512265574478e-07,
+      "loss": 3.4906,
+      "step": 6864
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.310420752759868e-07,
+      "loss": 3.5925,
+      "step": 6865
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.299335966654882e-07,
+      "loss": 3.5365,
+      "step": 6866
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.288257909048048e-07,
+      "loss": 3.4695,
+      "step": 6867
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.277186581726865e-07,
+      "loss": 3.5363,
+      "step": 6868
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.266121986477699e-07,
+      "loss": 3.4498,
+      "step": 6869
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.255064125085844e-07,
+      "loss": 3.4297,
+      "step": 6870
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.244012999335532e-07,
+      "loss": 3.4756,
+      "step": 6871
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.232968611009873e-07,
+      "loss": 3.5156,
+      "step": 6872
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.221930961890895e-07,
+      "loss": 3.4228,
+      "step": 6873
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.210900053759569e-07,
+      "loss": 3.5003,
+      "step": 6874
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.199875888395736e-07,
+      "loss": 3.5907,
+      "step": 6875
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.188858467578197e-07,
+      "loss": 3.5052,
+      "step": 6876
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.177847793084614e-07,
+      "loss": 3.6077,
+      "step": 6877
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.166843866691604e-07,
+      "loss": 3.4871,
+      "step": 6878
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.155846690174657e-07,
+      "loss": 3.5393,
+      "step": 6879
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.14485626530821e-07,
+      "loss": 3.5994,
+      "step": 6880
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.133872593865571e-07,
+      "loss": 3.581,
+      "step": 6881
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.122895677619014e-07,
+      "loss": 3.4127,
+      "step": 6882
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.111925518339653e-07,
+      "loss": 3.5568,
+      "step": 6883
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.100962117797584e-07,
+      "loss": 3.4904,
+      "step": 6884
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.090005477761742e-07,
+      "loss": 3.4871,
+      "step": 6885
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.079055600000036e-07,
+      "loss": 3.5513,
+      "step": 6886
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.068112486279234e-07,
+      "loss": 3.6185,
+      "step": 6887
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.057176138365047e-07,
+      "loss": 3.52,
+      "step": 6888
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.046246558022064e-07,
+      "loss": 3.5152,
+      "step": 6889
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.035323747013812e-07,
+      "loss": 3.4887,
+      "step": 6890
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.0244077071027e-07,
+      "loss": 3.5505,
+      "step": 6891
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.013498440050077e-07,
+      "loss": 3.6209,
+      "step": 6892
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.002595947616148e-07,
+      "loss": 3.4721,
+      "step": 6893
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.991700231560084e-07,
+      "loss": 3.6075,
+      "step": 6894
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.980811293639906e-07,
+      "loss": 3.5144,
+      "step": 6895
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.969929135612603e-07,
+      "loss": 3.5787,
+      "step": 6896
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.959053759233997e-07,
+      "loss": 3.4923,
+      "step": 6897
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.948185166258898e-07,
+      "loss": 3.5947,
+      "step": 6898
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.937323358440935e-07,
+      "loss": 3.5055,
+      "step": 6899
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.926468337532722e-07,
+      "loss": 3.6426,
+      "step": 6900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.915620105285715e-07,
+      "loss": 3.6494,
+      "step": 6901
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.904778663450325e-07,
+      "loss": 3.6097,
+      "step": 6902
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.893944013775817e-07,
+      "loss": 3.5685,
+      "step": 6903
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.883116158010417e-07,
+      "loss": 3.5919,
+      "step": 6904
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.872295097901205e-07,
+      "loss": 3.5787,
+      "step": 6905
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.86148083519418e-07,
+      "loss": 3.4809,
+      "step": 6906
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.85067337163426e-07,
+      "loss": 3.6209,
+      "step": 6907
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.839872708965257e-07,
+      "loss": 3.5406,
+      "step": 6908
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.82907884892986e-07,
+      "loss": 3.6026,
+      "step": 6909
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.818291793269711e-07,
+      "loss": 3.512,
+      "step": 6910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.807511543725299e-07,
+      "loss": 3.4992,
+      "step": 6911
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.796738102036072e-07,
+      "loss": 3.5125,
+      "step": 6912
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.785971469940323e-07,
+      "loss": 3.4454,
+      "step": 6913
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.775211649175296e-07,
+      "loss": 3.4815,
+      "step": 6914
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.76445864147709e-07,
+      "loss": 3.5187,
+      "step": 6915
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.753712448580753e-07,
+      "loss": 3.3096,
+      "step": 6916
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.742973072220178e-07,
+      "loss": 3.4751,
+      "step": 6917
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.732240514128214e-07,
+      "loss": 3.4499,
+      "step": 6918
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.721514776036565e-07,
+      "loss": 3.4858,
+      "step": 6919
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.710795859675879e-07,
+      "loss": 3.5015,
+      "step": 6920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.700083766775646e-07,
+      "loss": 3.5881,
+      "step": 6921
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.68937849906431e-07,
+      "loss": 3.4377,
+      "step": 6922
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.678680058269173e-07,
+      "loss": 3.5318,
+      "step": 6923
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.667988446116475e-07,
+      "loss": 3.5189,
+      "step": 6924
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.657303664331311e-07,
+      "loss": 3.5043,
+      "step": 6925
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.646625714637712e-07,
+      "loss": 3.4959,
+      "step": 6926
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.635954598758577e-07,
+      "loss": 3.4584,
+      "step": 6927
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.625290318415723e-07,
+      "loss": 3.5812,
+      "step": 6928
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.614632875329847e-07,
+      "loss": 3.4409,
+      "step": 6929
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.603982271220572e-07,
+      "loss": 3.4814,
+      "step": 6930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.593338507806369e-07,
+      "loss": 3.4937,
+      "step": 6931
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.582701586804658e-07,
+      "loss": 3.5119,
+      "step": 6932
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.572071509931711e-07,
+      "loss": 3.4173,
+      "step": 6933
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.561448278902739e-07,
+      "loss": 3.5399,
+      "step": 6934
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.550831895431799e-07,
+      "loss": 3.4079,
+      "step": 6935
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.540222361231892e-07,
+      "loss": 3.3771,
+      "step": 6936
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.529619678014872e-07,
+      "loss": 3.5547,
+      "step": 6937
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.519023847491525e-07,
+      "loss": 3.4895,
+      "step": 6938
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.508434871371484e-07,
+      "loss": 3.6111,
+      "step": 6939
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.497852751363344e-07,
+      "loss": 3.4796,
+      "step": 6940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.487277489174527e-07,
+      "loss": 3.5931,
+      "step": 6941
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.476709086511369e-07,
+      "loss": 3.445,
+      "step": 6942
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.466147545079133e-07,
+      "loss": 3.5892,
+      "step": 6943
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.455592866581929e-07,
+      "loss": 3.4204,
+      "step": 6944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.445045052722788e-07,
+      "loss": 3.5432,
+      "step": 6945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.434504105203622e-07,
+      "loss": 3.4175,
+      "step": 6946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.423970025725225e-07,
+      "loss": 3.433,
+      "step": 6947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.413442815987315e-07,
+      "loss": 3.5792,
+      "step": 6948
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.402922477688468e-07,
+      "loss": 3.4952,
+      "step": 6949
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.392409012526163e-07,
+      "loss": 3.4977,
+      "step": 6950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.381902422196785e-07,
+      "loss": 3.3721,
+      "step": 6951
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.371402708395581e-07,
+      "loss": 3.5656,
+      "step": 6952
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.360909872816724e-07,
+      "loss": 3.5353,
+      "step": 6953
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.350423917153227e-07,
+      "loss": 3.5299,
+      "step": 6954
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.339944843097057e-07,
+      "loss": 3.5633,
+      "step": 6955
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.329472652339009e-07,
+      "loss": 3.5259,
+      "step": 6956
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.319007346568818e-07,
+      "loss": 3.4259,
+      "step": 6957
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.308548927475062e-07,
+      "loss": 3.54,
+      "step": 6958
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.298097396745257e-07,
+      "loss": 3.5643,
+      "step": 6959
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.287652756065755e-07,
+      "loss": 3.4967,
+      "step": 6960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.277215007121847e-07,
+      "loss": 3.5385,
+      "step": 6961
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.266784151597666e-07,
+      "loss": 3.3815,
+      "step": 6962
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.256360191176276e-07,
+      "loss": 3.4471,
+      "step": 6963
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.245943127539585e-07,
+      "loss": 3.4672,
+      "step": 6964
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.23553296236843e-07,
+      "loss": 3.5718,
+      "step": 6965
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.225129697342492e-07,
+      "loss": 3.5541,
+      "step": 6966
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.214733334140395e-07,
+      "loss": 3.4365,
+      "step": 6967
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.204343874439578e-07,
+      "loss": 3.5667,
+      "step": 6968
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.193961319916442e-07,
+      "loss": 3.5391,
+      "step": 6969
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.183585672246201e-07,
+      "loss": 3.4301,
+      "step": 6970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.173216933103022e-07,
+      "loss": 3.5529,
+      "step": 6971
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.162855104159894e-07,
+      "loss": 3.4267,
+      "step": 6972
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.152500187088746e-07,
+      "loss": 3.5903,
+      "step": 6973
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.142152183560353e-07,
+      "loss": 3.5831,
+      "step": 6974
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.131811095244401e-07,
+      "loss": 3.453,
+      "step": 6975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.121476923809434e-07,
+      "loss": 3.5015,
+      "step": 6976
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.111149670922918e-07,
+      "loss": 3.4657,
+      "step": 6977
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.100829338251147e-07,
+      "loss": 3.4781,
+      "step": 6978
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.090515927459357e-07,
+      "loss": 3.3812,
+      "step": 6979
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.080209440211627e-07,
+      "loss": 3.5159,
+      "step": 6980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.069909878170938e-07,
+      "loss": 3.5043,
+      "step": 6981
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.059617242999156e-07,
+      "loss": 3.58,
+      "step": 6982
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.049331536356991e-07,
+      "loss": 3.5952,
+      "step": 6983
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.039052759904102e-07,
+      "loss": 3.4957,
+      "step": 6984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.028780915298977e-07,
+      "loss": 3.5454,
+      "step": 6985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.018516004198994e-07,
+      "loss": 3.5906,
+      "step": 6986
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.008258028260434e-07,
+      "loss": 3.467,
+      "step": 6987
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.998006989138428e-07,
+      "loss": 3.5693,
+      "step": 6988
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.987762888487032e-07,
+      "loss": 3.5253,
+      "step": 6989
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.977525727959128e-07,
+      "loss": 3.5335,
+      "step": 6990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.967295509206523e-07,
+      "loss": 3.5194,
+      "step": 6991
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.95707223387987e-07,
+      "loss": 3.5322,
+      "step": 6992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.946855903628741e-07,
+      "loss": 3.4457,
+      "step": 6993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.936646520101543e-07,
+      "loss": 3.5481,
+      "step": 6994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.926444084945593e-07,
+      "loss": 3.5503,
+      "step": 6995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.916248599807074e-07,
+      "loss": 3.5438,
+      "step": 6996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.906060066331066e-07,
+      "loss": 3.5583,
+      "step": 6997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.895878486161484e-07,
+      "loss": 3.5146,
+      "step": 6998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.885703860941173e-07,
+      "loss": 3.4962,
+      "step": 6999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.875536192311822e-07,
+      "loss": 3.5536,
+      "step": 7000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.865375481914017e-07,
+      "loss": 3.4362,
+      "step": 7001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.85522173138719e-07,
+      "loss": 3.5488,
+      "step": 7002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.8450749423697e-07,
+      "loss": 3.6008,
+      "step": 7003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.834935116498736e-07,
+      "loss": 3.421,
+      "step": 7004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.824802255410395e-07,
+      "loss": 3.577,
+      "step": 7005
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.814676360739619e-07,
+      "loss": 3.5669,
+      "step": 7006
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.804557434120269e-07,
+      "loss": 3.5411,
+      "step": 7007
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.79444547718503e-07,
+      "loss": 3.6094,
+      "step": 7008
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.784340491565517e-07,
+      "loss": 3.4627,
+      "step": 7009
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.774242478892168e-07,
+      "loss": 3.5867,
+      "step": 7010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.764151440794342e-07,
+      "loss": 3.5555,
+      "step": 7011
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.75406737890023e-07,
+      "loss": 3.4048,
+      "step": 7012
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.743990294836944e-07,
+      "loss": 3.5129,
+      "step": 7013
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.73392019023042e-07,
+      "loss": 3.4852,
+      "step": 7014
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.723857066705513e-07,
+      "loss": 3.4631,
+      "step": 7015
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.713800925885905e-07,
+      "loss": 3.5188,
+      "step": 7016
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.703751769394212e-07,
+      "loss": 3.5155,
+      "step": 7017
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.693709598851866e-07,
+      "loss": 3.4892,
+      "step": 7018
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.683674415879188e-07,
+      "loss": 3.3931,
+      "step": 7019
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.673646222095398e-07,
+      "loss": 3.4659,
+      "step": 7020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.663625019118564e-07,
+      "loss": 3.5901,
+      "step": 7021
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.653610808565608e-07,
+      "loss": 3.4619,
+      "step": 7022
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.643603592052378e-07,
+      "loss": 3.6014,
+      "step": 7023
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.633603371193531e-07,
+      "loss": 3.5796,
+      "step": 7024
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.623610147602655e-07,
+      "loss": 3.5092,
+      "step": 7025
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.613623922892154e-07,
+      "loss": 3.5175,
+      "step": 7026
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.603644698673351e-07,
+      "loss": 3.5145,
+      "step": 7027
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.593672476556396e-07,
+      "loss": 3.4903,
+      "step": 7028
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.583707258150351e-07,
+      "loss": 3.5438,
+      "step": 7029
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.573749045063099e-07,
+      "loss": 3.3977,
+      "step": 7030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.563797838901453e-07,
+      "loss": 3.6392,
+      "step": 7031
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.553853641271035e-07,
+      "loss": 3.5112,
+      "step": 7032
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.543916453776389e-07,
+      "loss": 3.4905,
+      "step": 7033
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.533986278020876e-07,
+      "loss": 3.5003,
+      "step": 7034
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.524063115606782e-07,
+      "loss": 3.5044,
+      "step": 7035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.514146968135205e-07,
+      "loss": 3.5792,
+      "step": 7036
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.504237837206156e-07,
+      "loss": 3.5151,
+      "step": 7037
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.494335724418488e-07,
+      "loss": 3.4844,
+      "step": 7038
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.484440631369937e-07,
+      "loss": 3.4522,
+      "step": 7039
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.474552559657088e-07,
+      "loss": 3.563,
+      "step": 7040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.464671510875419e-07,
+      "loss": 3.4844,
+      "step": 7041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.454797486619241e-07,
+      "loss": 3.595,
+      "step": 7042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.444930488481771e-07,
+      "loss": 3.5808,
+      "step": 7043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.435070518055053e-07,
+      "loss": 3.4534,
+      "step": 7044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.425217576930037e-07,
+      "loss": 3.526,
+      "step": 7045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.415371666696491e-07,
+      "loss": 3.3662,
+      "step": 7046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.405532788943103e-07,
+      "loss": 3.7109,
+      "step": 7047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.395700945257377e-07,
+      "loss": 3.46,
+      "step": 7048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.385876137225716e-07,
+      "loss": 3.4804,
+      "step": 7049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.376058366433368e-07,
+      "loss": 3.5087,
+      "step": 7050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.366247634464462e-07,
+      "loss": 3.5283,
+      "step": 7051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.356443942901969e-07,
+      "loss": 3.5073,
+      "step": 7052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.346647293327756e-07,
+      "loss": 3.4701,
+      "step": 7053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.336857687322523e-07,
+      "loss": 3.4782,
+      "step": 7054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.327075126465832e-07,
+      "loss": 3.5072,
+      "step": 7055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.317299612336147e-07,
+      "loss": 3.4617,
+      "step": 7056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.307531146510754e-07,
+      "loss": 3.5187,
+      "step": 7057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.297769730565829e-07,
+      "loss": 3.4934,
+      "step": 7058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.288015366076394e-07,
+      "loss": 3.5683,
+      "step": 7059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.278268054616321e-07,
+      "loss": 3.5173,
+      "step": 7060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.268527797758389e-07,
+      "loss": 3.5478,
+      "step": 7061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.258794597074192e-07,
+      "loss": 3.4875,
+      "step": 7062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.249068454134216e-07,
+      "loss": 3.5942,
+      "step": 7063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.23934937050778e-07,
+      "loss": 3.503,
+      "step": 7064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.229637347763106e-07,
+      "loss": 3.4316,
+      "step": 7065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.219932387467226e-07,
+      "loss": 3.4583,
+      "step": 7066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.210234491186079e-07,
+      "loss": 3.565,
+      "step": 7067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.200543660484421e-07,
+      "loss": 3.3476,
+      "step": 7068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.19085989692591e-07,
+      "loss": 3.6306,
+      "step": 7069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.181183202073027e-07,
+      "loss": 3.6128,
+      "step": 7070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.171513577487148e-07,
+      "loss": 3.5518,
+      "step": 7071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.161851024728466e-07,
+      "loss": 3.5097,
+      "step": 7072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.152195545356082e-07,
+      "loss": 3.584,
+      "step": 7073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.142547140927902e-07,
+      "loss": 3.5308,
+      "step": 7074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.132905813000745e-07,
+      "loss": 3.5414,
+      "step": 7075
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.123271563130234e-07,
+      "loss": 3.4219,
+      "step": 7076
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.11364439287091e-07,
+      "loss": 3.5422,
+      "step": 7077
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.104024303776107e-07,
+      "loss": 3.546,
+      "step": 7078
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.094411297398073e-07,
+      "loss": 3.409,
+      "step": 7079
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.084805375287867e-07,
+      "loss": 3.5375,
+      "step": 7080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.075206538995448e-07,
+      "loss": 3.5383,
+      "step": 7081
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.065614790069591e-07,
+      "loss": 3.5093,
+      "step": 7082
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.056030130057972e-07,
+      "loss": 3.6333,
+      "step": 7083
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.046452560507065e-07,
+      "loss": 3.4851,
+      "step": 7084
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.036882082962264e-07,
+      "loss": 3.655,
+      "step": 7085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.02731869896776e-07,
+      "loss": 3.4762,
+      "step": 7086
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.017762410066657e-07,
+      "loss": 3.4588,
+      "step": 7087
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.008213217800851e-07,
+      "loss": 3.4082,
+      "step": 7088
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.998671123711158e-07,
+      "loss": 3.5099,
+      "step": 7089
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.989136129337203e-07,
+      "loss": 3.6024,
+      "step": 7090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.979608236217465e-07,
+      "loss": 3.6275,
+      "step": 7091
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.970087445889317e-07,
+      "loss": 3.5345,
+      "step": 7092
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.960573759888938e-07,
+      "loss": 3.5192,
+      "step": 7093
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.951067179751402e-07,
+      "loss": 3.5546,
+      "step": 7094
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.94156770701061e-07,
+      "loss": 3.569,
+      "step": 7095
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.932075343199311e-07,
+      "loss": 3.6242,
+      "step": 7096
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.922590089849145e-07,
+      "loss": 3.5458,
+      "step": 7097
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.913111948490551e-07,
+      "loss": 3.525,
+      "step": 7098
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.903640920652881e-07,
+      "loss": 3.5651,
+      "step": 7099
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.894177007864272e-07,
+      "loss": 3.3773,
+      "step": 7100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.884720211651784e-07,
+      "loss": 3.5269,
+      "step": 7101
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.87527053354126e-07,
+      "loss": 3.5091,
+      "step": 7102
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.865827975057454e-07,
+      "loss": 3.4947,
+      "step": 7103
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.85639253772392e-07,
+      "loss": 3.572,
+      "step": 7104
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.846964223063117e-07,
+      "loss": 3.5536,
+      "step": 7105
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.837543032596293e-07,
+      "loss": 3.6597,
+      "step": 7106
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.82812896784361e-07,
+      "loss": 3.4524,
+      "step": 7107
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.81872203032402e-07,
+      "loss": 3.4667,
+      "step": 7108
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.80932222155538e-07,
+      "loss": 3.4992,
+      "step": 7109
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.799929543054345e-07,
+      "loss": 3.4977,
+      "step": 7110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.790543996336468e-07,
+      "loss": 3.5477,
+      "step": 7111
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.781165582916115e-07,
+      "loss": 3.6042,
+      "step": 7112
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.771794304306522e-07,
+      "loss": 3.427,
+      "step": 7113
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.762430162019761e-07,
+      "loss": 3.5467,
+      "step": 7114
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.753073157566763e-07,
+      "loss": 3.5539,
+      "step": 7115
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.743723292457293e-07,
+      "loss": 3.516,
+      "step": 7116
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.734380568199988e-07,
+      "loss": 3.425,
+      "step": 7117
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.725044986302302e-07,
+      "loss": 3.639,
+      "step": 7118
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.715716548270572e-07,
+      "loss": 3.5817,
+      "step": 7119
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.706395255609936e-07,
+      "loss": 3.615,
+      "step": 7120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.697081109824437e-07,
+      "loss": 3.4444,
+      "step": 7121
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.687774112416905e-07,
+      "loss": 3.5134,
+      "step": 7122
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.678474264889066e-07,
+      "loss": 3.5648,
+      "step": 7123
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.669181568741461e-07,
+      "loss": 3.4912,
+      "step": 7124
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.659896025473499e-07,
+      "loss": 3.6357,
+      "step": 7125
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.650617636583406e-07,
+      "loss": 3.5313,
+      "step": 7126
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.641346403568287e-07,
+      "loss": 3.4555,
+      "step": 7127
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.632082327924077e-07,
+      "loss": 3.5235,
+      "step": 7128
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.622825411145538e-07,
+      "loss": 3.591,
+      "step": 7129
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.613575654726316e-07,
+      "loss": 3.5585,
+      "step": 7130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.604333060158879e-07,
+      "loss": 3.5106,
+      "step": 7131
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.595097628934515e-07,
+      "loss": 3.478,
+      "step": 7132
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.585869362543416e-07,
+      "loss": 3.444,
+      "step": 7133
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.576648262474554e-07,
+      "loss": 3.4395,
+      "step": 7134
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.567434330215804e-07,
+      "loss": 3.4894,
+      "step": 7135
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.558227567253832e-07,
+      "loss": 3.4873,
+      "step": 7136
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.549027975074184e-07,
+      "loss": 3.6527,
+      "step": 7137
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.539835555161221e-07,
+      "loss": 3.6101,
+      "step": 7138
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.530650308998187e-07,
+      "loss": 3.4642,
+      "step": 7139
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.521472238067111e-07,
+      "loss": 3.5777,
+      "step": 7140
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.512301343848919e-07,
+      "loss": 3.4526,
+      "step": 7141
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.50313762782334e-07,
+      "loss": 3.6083,
+      "step": 7142
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.493981091468981e-07,
+      "loss": 3.4774,
+      "step": 7143
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.484831736263246e-07,
+      "loss": 3.4715,
+      "step": 7144
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.475689563682429e-07,
+      "loss": 3.5344,
+      "step": 7145
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.466554575201615e-07,
+      "loss": 3.6394,
+      "step": 7146
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.457426772294777e-07,
+      "loss": 3.5572,
+      "step": 7147
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.448306156434691e-07,
+      "loss": 3.455,
+      "step": 7148
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.439192729093007e-07,
+      "loss": 3.6685,
+      "step": 7149
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.430086491740177e-07,
+      "loss": 3.5089,
+      "step": 7150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.420987445845533e-07,
+      "loss": 3.494,
+      "step": 7151
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.411895592877209e-07,
+      "loss": 3.6033,
+      "step": 7152
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.402810934302217e-07,
+      "loss": 3.6029,
+      "step": 7153
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.39373347158636e-07,
+      "loss": 3.4439,
+      "step": 7154
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.38466320619434e-07,
+      "loss": 3.4882,
+      "step": 7155
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.375600139589632e-07,
+      "loss": 3.4472,
+      "step": 7156
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.366544273234614e-07,
+      "loss": 3.472,
+      "step": 7157
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.357495608590441e-07,
+      "loss": 3.6866,
+      "step": 7158
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.348454147117166e-07,
+      "loss": 3.4112,
+      "step": 7159
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.339419890273623e-07,
+      "loss": 3.5504,
+      "step": 7160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.330392839517529e-07,
+      "loss": 3.6507,
+      "step": 7161
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.321372996305407e-07,
+      "loss": 3.5772,
+      "step": 7162
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.312360362092639e-07,
+      "loss": 3.5677,
+      "step": 7163
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.303354938333432e-07,
+      "loss": 3.5288,
+      "step": 7164
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.29435672648082e-07,
+      "loss": 3.6044,
+      "step": 7165
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.285365727986708e-07,
+      "loss": 3.5459,
+      "step": 7166
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.276381944301795e-07,
+      "loss": 3.5772,
+      "step": 7167
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.267405376875634e-07,
+      "loss": 3.5333,
+      "step": 7168
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.258436027156632e-07,
+      "loss": 3.5029,
+      "step": 7169
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.249473896591995e-07,
+      "loss": 3.5433,
+      "step": 7170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.240518986627807e-07,
+      "loss": 3.482,
+      "step": 7171
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.231571298708932e-07,
+      "loss": 3.5888,
+      "step": 7172
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.22263083427913e-07,
+      "loss": 3.5486,
+      "step": 7173
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.213697594780942e-07,
+      "loss": 3.5446,
+      "step": 7174
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.204771581655794e-07,
+      "loss": 3.5216,
+      "step": 7175
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.195852796343881e-07,
+      "loss": 3.539,
+      "step": 7176
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.18694124028431e-07,
+      "loss": 3.5185,
+      "step": 7177
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.178036914914947e-07,
+      "loss": 3.5626,
+      "step": 7178
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.169139821672553e-07,
+      "loss": 3.5645,
+      "step": 7179
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.160249961992669e-07,
+      "loss": 3.4774,
+      "step": 7180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.151367337309721e-07,
+      "loss": 3.5214,
+      "step": 7181
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.142491949056916e-07,
+      "loss": 3.5354,
+      "step": 7182
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.133623798666338e-07,
+      "loss": 3.455,
+      "step": 7183
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.124762887568862e-07,
+      "loss": 3.5587,
+      "step": 7184
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.115909217194243e-07,
+      "loss": 3.5509,
+      "step": 7185
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.107062788971017e-07,
+      "loss": 3.6148,
+      "step": 7186
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.098223604326596e-07,
+      "loss": 3.5398,
+      "step": 7187
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.089391664687182e-07,
+      "loss": 3.6247,
+      "step": 7188
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.080566971477851e-07,
+      "loss": 3.5625,
+      "step": 7189
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.071749526122466e-07,
+      "loss": 3.419,
+      "step": 7190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.062939330043764e-07,
+      "loss": 3.5184,
+      "step": 7191
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.054136384663267e-07,
+      "loss": 3.497,
+      "step": 7192
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.045340691401369e-07,
+      "loss": 3.5446,
+      "step": 7193
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.036552251677262e-07,
+      "loss": 3.5563,
+      "step": 7194
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.027771066908998e-07,
+      "loss": 3.5072,
+      "step": 7195
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.01899713851342e-07,
+      "loss": 3.5153,
+      "step": 7196
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.010230467906241e-07,
+      "loss": 3.5461,
+      "step": 7197
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.001471056501966e-07,
+      "loss": 3.4864,
+      "step": 7198
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.992718905713967e-07,
+      "loss": 3.5334,
+      "step": 7199
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.983974016954407e-07,
+      "loss": 3.5508,
+      "step": 7200
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.975236391634292e-07,
+      "loss": 3.5118,
+      "step": 7201
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.966506031163476e-07,
+      "loss": 3.5417,
+      "step": 7202
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.957782936950612e-07,
+      "loss": 3.4914,
+      "step": 7203
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.949067110403177e-07,
+      "loss": 3.5094,
+      "step": 7204
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.940358552927515e-07,
+      "loss": 3.4058,
+      "step": 7205
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.931657265928752e-07,
+      "loss": 3.5426,
+      "step": 7206
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.922963250810881e-07,
+      "loss": 3.5237,
+      "step": 7207
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.914276508976673e-07,
+      "loss": 3.5607,
+      "step": 7208
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.905597041827781e-07,
+      "loss": 3.4581,
+      "step": 7209
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.89692485076464e-07,
+      "loss": 3.4805,
+      "step": 7210
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.888259937186546e-07,
+      "loss": 3.4441,
+      "step": 7211
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.879602302491576e-07,
+      "loss": 3.5436,
+      "step": 7212
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.870951948076686e-07,
+      "loss": 3.5823,
+      "step": 7213
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.862308875337607e-07,
+      "loss": 3.4986,
+      "step": 7214
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.853673085668947e-07,
+      "loss": 3.391,
+      "step": 7215
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.845044580464075e-07,
+      "loss": 3.4443,
+      "step": 7216
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.836423361115255e-07,
+      "loss": 3.5138,
+      "step": 7217
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.827809429013513e-07,
+      "loss": 3.4942,
+      "step": 7218
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.819202785548755e-07,
+      "loss": 3.4521,
+      "step": 7219
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.810603432109651e-07,
+      "loss": 3.4033,
+      "step": 7220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.802011370083748e-07,
+      "loss": 3.5692,
+      "step": 7221
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.793426600857381e-07,
+      "loss": 3.5229,
+      "step": 7222
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.784849125815744e-07,
+      "loss": 3.5068,
+      "step": 7223
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.776278946342805e-07,
+      "loss": 3.4277,
+      "step": 7224
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7677160638214093e-07,
+      "loss": 3.6016,
+      "step": 7225
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.759160479633174e-07,
+      "loss": 3.554,
+      "step": 7226
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.750612195158577e-07,
+      "loss": 3.4712,
+      "step": 7227
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.742071211776894e-07,
+      "loss": 3.5004,
+      "step": 7228
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7335375308662447e-07,
+      "loss": 3.4997,
+      "step": 7229
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7250111538035437e-07,
+      "loss": 3.6271,
+      "step": 7230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.716492081964552e-07,
+      "loss": 3.4854,
+      "step": 7231
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.707980316723837e-07,
+      "loss": 3.5053,
+      "step": 7232
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6994758594547996e-07,
+      "loss": 3.5299,
+      "step": 7233
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6909787115296366e-07,
+      "loss": 3.47,
+      "step": 7234
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.682488874319402e-07,
+      "loss": 3.5626,
+      "step": 7235
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.674006349193938e-07,
+      "loss": 3.4668,
+      "step": 7236
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.665531137521917e-07,
+      "loss": 3.5379,
+      "step": 7237
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.65706324067085e-07,
+      "loss": 3.6269,
+      "step": 7238
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.648602660007029e-07,
+      "loss": 3.5737,
+      "step": 7239
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.64014939689561e-07,
+      "loss": 3.4756,
+      "step": 7240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6317034527005424e-07,
+      "loss": 3.6668,
+      "step": 7241
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6232648287845796e-07,
+      "loss": 3.4426,
+      "step": 7242
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6148335265093324e-07,
+      "loss": 3.51,
+      "step": 7243
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6064095472352024e-07,
+      "loss": 3.5381,
+      "step": 7244
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.597992892321429e-07,
+      "loss": 3.4369,
+      "step": 7245
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5895835631260375e-07,
+      "loss": 3.4922,
+      "step": 7246
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5811815610059205e-07,
+      "loss": 3.6685,
+      "step": 7247
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5727868873167324e-07,
+      "loss": 3.5633,
+      "step": 7248
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.564399543412995e-07,
+      "loss": 3.5491,
+      "step": 7249
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5560195306480095e-07,
+      "loss": 3.4225,
+      "step": 7250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5476468503739267e-07,
+      "loss": 3.5336,
+      "step": 7251
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5392815039416784e-07,
+      "loss": 3.4517,
+      "step": 7252
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5309234927010503e-07,
+      "loss": 3.5854,
+      "step": 7253
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5225728180006146e-07,
+      "loss": 3.5689,
+      "step": 7254
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5142294811877816e-07,
+      "loss": 3.4573,
+      "step": 7255
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.505893483608759e-07,
+      "loss": 3.5387,
+      "step": 7256
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4975648266085915e-07,
+      "loss": 3.7011,
+      "step": 7257
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.489243511531105e-07,
+      "loss": 3.4765,
+      "step": 7258
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.480929539718987e-07,
+      "loss": 3.5958,
+      "step": 7259
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.472622912513697e-07,
+      "loss": 3.6145,
+      "step": 7260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.464323631255546e-07,
+      "loss": 3.5326,
+      "step": 7261
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.456031697283625e-07,
+      "loss": 3.5565,
+      "step": 7262
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.447747111935874e-07,
+      "loss": 3.471,
+      "step": 7263
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.439469876549013e-07,
+      "loss": 3.5747,
+      "step": 7264
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.431199992458607e-07,
+      "loss": 3.6146,
+      "step": 7265
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.422937460999005e-07,
+      "loss": 3.5307,
+      "step": 7266
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4146822835034074e-07,
+      "loss": 3.4973,
+      "step": 7267
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4064344613037824e-07,
+      "loss": 3.3956,
+      "step": 7268
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3981939957309537e-07,
+      "loss": 3.5131,
+      "step": 7269
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.389960888114525e-07,
+      "loss": 3.5522,
+      "step": 7270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3817351397829446e-07,
+      "loss": 3.6139,
+      "step": 7271
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3735167520634445e-07,
+      "loss": 3.4641,
+      "step": 7272
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.365305726282071e-07,
+      "loss": 3.407,
+      "step": 7273
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.357102063763713e-07,
+      "loss": 3.5896,
+      "step": 7274
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3489057658320344e-07,
+      "loss": 3.499,
+      "step": 7275
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3407168338095327e-07,
+      "loss": 3.5284,
+      "step": 7276
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3325352690175183e-07,
+      "loss": 3.678,
+      "step": 7277
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.324361072776084e-07,
+      "loss": 3.5105,
+      "step": 7278
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.316194246404182e-07,
+      "loss": 3.5971,
+      "step": 7279
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.308034791219523e-07,
+      "loss": 3.383,
+      "step": 7280
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2998827085386773e-07,
+      "loss": 3.6228,
+      "step": 7281
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.291737999676981e-07,
+      "loss": 3.5881,
+      "step": 7282
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2836006659486227e-07,
+      "loss": 3.5192,
+      "step": 7283
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2754707086665626e-07,
+      "loss": 3.5996,
+      "step": 7284
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2673481291425957e-07,
+      "loss": 3.511,
+      "step": 7285
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.259232928687318e-07,
+      "loss": 3.5325,
+      "step": 7286
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.251125108610138e-07,
+      "loss": 3.5107,
+      "step": 7287
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.243024670219259e-07,
+      "loss": 3.5176,
+      "step": 7288
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.23493161482173e-07,
+      "loss": 3.5216,
+      "step": 7289
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.226845943723357e-07,
+      "loss": 3.5861,
+      "step": 7290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2187676582288005e-07,
+      "loss": 3.5452,
+      "step": 7291
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2106967596414974e-07,
+      "loss": 3.5311,
+      "step": 7292
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.202633249263721e-07,
+      "loss": 3.5177,
+      "step": 7293
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.194577128396521e-07,
+      "loss": 3.6233,
+      "step": 7294
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.186528398339784e-07,
+      "loss": 3.5112,
+      "step": 7295
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.178487060392178e-07,
+      "loss": 3.5837,
+      "step": 7296
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1704531158512086e-07,
+      "loss": 3.5685,
+      "step": 7297
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.16242656601315e-07,
+      "loss": 3.4794,
+      "step": 7298
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.154407412173128e-07,
+      "loss": 3.5548,
+      "step": 7299
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1463956556250296e-07,
+      "loss": 3.4547,
+      "step": 7300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1383912976615927e-07,
+      "loss": 3.4492,
+      "step": 7301
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1303943395743173e-07,
+      "loss": 3.617,
+      "step": 7302
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.122404782653544e-07,
+      "loss": 3.5247,
+      "step": 7303
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.114422628188397e-07,
+      "loss": 3.5682,
+      "step": 7304
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.106447877466829e-07,
+      "loss": 3.5313,
+      "step": 7305
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.098480531775567e-07,
+      "loss": 3.4115,
+      "step": 7306
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0905205924001767e-07,
+      "loss": 3.6081,
+      "step": 7307
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0825680606249974e-07,
+      "loss": 3.4588,
+      "step": 7308
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.074622937733208e-07,
+      "loss": 3.5544,
+      "step": 7309
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.066685225006761e-07,
+      "loss": 3.6136,
+      "step": 7310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0587549237264154e-07,
+      "loss": 3.4862,
+      "step": 7311
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0508320351717587e-07,
+      "loss": 3.4693,
+      "step": 7312
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.042916560621163e-07,
+      "loss": 3.3652,
+      "step": 7313
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0350085013518005e-07,
+      "loss": 3.5937,
+      "step": 7314
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0271078586396675e-07,
+      "loss": 3.6644,
+      "step": 7315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.019214633759533e-07,
+      "loss": 3.3289,
+      "step": 7316
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0113288279850115e-07,
+      "loss": 3.5305,
+      "step": 7317
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.003450442588469e-07,
+      "loss": 3.5493,
+      "step": 7318
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9955794788411274e-07,
+      "loss": 3.4323,
+      "step": 7319
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.987715938012965e-07,
+      "loss": 3.5121,
+      "step": 7320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9798598213727944e-07,
+      "loss": 3.4446,
+      "step": 7321
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.972011130188208e-07,
+      "loss": 3.553,
+      "step": 7322
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.96416986572562e-07,
+      "loss": 3.6447,
+      "step": 7323
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9563360292502297e-07,
+      "loss": 3.5347,
+      "step": 7324
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.948509622026048e-07,
+      "loss": 3.4872,
+      "step": 7325
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.940690645315881e-07,
+      "loss": 3.4966,
+      "step": 7326
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9328791003813426e-07,
+      "loss": 3.576,
+      "step": 7327
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9250749884828353e-07,
+      "loss": 3.567,
+      "step": 7328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9172783108795854e-07,
+      "loss": 3.5561,
+      "step": 7329
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.909489068829586e-07,
+      "loss": 3.4498,
+      "step": 7330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9017072635896716e-07,
+      "loss": 3.496,
+      "step": 7331
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8939328964154253e-07,
+      "loss": 3.522,
+      "step": 7332
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8861659685612895e-07,
+      "loss": 3.4927,
+      "step": 7333
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.878406481280456e-07,
+      "loss": 3.5498,
+      "step": 7334
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.870654435824944e-07,
+      "loss": 3.511,
+      "step": 7335
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.862909833445561e-07,
+      "loss": 3.5203,
+      "step": 7336
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.855172675391916e-07,
+      "loss": 3.4424,
+      "step": 7337
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8474429629124175e-07,
+      "loss": 3.5733,
+      "step": 7338
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8397206972542787e-07,
+      "loss": 3.4921,
+      "step": 7339
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.832005879663492e-07,
+      "loss": 3.5094,
+      "step": 7340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.824298511384877e-07,
+      "loss": 3.6307,
+      "step": 7341
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.816598593662024e-07,
+      "loss": 3.4461,
+      "step": 7342
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8089061277373385e-07,
+      "loss": 3.548,
+      "step": 7343
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.801221114852005e-07,
+      "loss": 3.405,
+      "step": 7344
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.793543556246038e-07,
+      "loss": 3.5643,
+      "step": 7345
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7858734531582175e-07,
+      "loss": 3.4926,
+      "step": 7346
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.778210806826127e-07,
+      "loss": 3.4388,
+      "step": 7347
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.770555618486166e-07,
+      "loss": 3.584,
+      "step": 7348
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7629078893735127e-07,
+      "loss": 3.4916,
+      "step": 7349
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7552676207221303e-07,
+      "loss": 3.4697,
+      "step": 7350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7476348137648166e-07,
+      "loss": 3.5208,
+      "step": 7351
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7400094697331246e-07,
+      "loss": 3.4131,
+      "step": 7352
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7323915898574316e-07,
+      "loss": 3.5406,
+      "step": 7353
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.724781175366893e-07,
+      "loss": 3.5737,
+      "step": 7354
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.717178227489482e-07,
+      "loss": 3.5229,
+      "step": 7355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7095827474519287e-07,
+      "loss": 3.4555,
+      "step": 7356
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.701994736479797e-07,
+      "loss": 3.5864,
+      "step": 7357
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6944141957974245e-07,
+      "loss": 3.4534,
+      "step": 7358
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.686841126627955e-07,
+      "loss": 3.4754,
+      "step": 7359
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6792755301933104e-07,
+      "loss": 3.602,
+      "step": 7360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6717174077142317e-07,
+      "loss": 3.5295,
+      "step": 7361
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6641667604102205e-07,
+      "loss": 3.4468,
+      "step": 7362
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6566235894996083e-07,
+      "loss": 3.5036,
+      "step": 7363
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.649087896199488e-07,
+      "loss": 3.613,
+      "step": 7364
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.641559681725776e-07,
+      "loss": 3.5446,
+      "step": 7365
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6340389472931557e-07,
+      "loss": 3.5064,
+      "step": 7366
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.626525694115124e-07,
+      "loss": 3.5158,
+      "step": 7367
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.619019923403949e-07,
+      "loss": 3.4913,
+      "step": 7368
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6115216363707184e-07,
+      "loss": 3.4506,
+      "step": 7369
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.604030834225286e-07,
+      "loss": 3.5984,
+      "step": 7370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5965475181763243e-07,
+      "loss": 3.6924,
+      "step": 7371
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.589071689431267e-07,
+      "loss": 3.5041,
+      "step": 7372
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.581603349196372e-07,
+      "loss": 3.39,
+      "step": 7373
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5741424986766527e-07,
+      "loss": 3.6537,
+      "step": 7374
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.566689139075963e-07,
+      "loss": 3.5639,
+      "step": 7375
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5592432715968906e-07,
+      "loss": 3.5311,
+      "step": 7376
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5518048974408637e-07,
+      "loss": 3.4107,
+      "step": 7377
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5443740178080666e-07,
+      "loss": 3.5423,
+      "step": 7378
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.536950633897507e-07,
+      "loss": 3.6412,
+      "step": 7379
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5295347469069427e-07,
+      "loss": 3.4322,
+      "step": 7380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.522126358032968e-07,
+      "loss": 3.549,
+      "step": 7381
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5147254684709254e-07,
+      "loss": 3.5345,
+      "step": 7382
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5073320794149655e-07,
+      "loss": 3.446,
+      "step": 7383
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.499946192058046e-07,
+      "loss": 3.4768,
+      "step": 7384
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.492567807591879e-07,
+      "loss": 3.6373,
+      "step": 7385
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.485196927206985e-07,
+      "loss": 3.5037,
+      "step": 7386
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4778335520926854e-07,
+      "loss": 3.5715,
+      "step": 7387
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4704776834370625e-07,
+      "loss": 3.5242,
+      "step": 7388
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.463129322427017e-07,
+      "loss": 3.4104,
+      "step": 7389
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4557884702482114e-07,
+      "loss": 3.5172,
+      "step": 7390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.448455128085121e-07,
+      "loss": 3.5371,
+      "step": 7391
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.441129297120982e-07,
+      "loss": 3.6341,
+      "step": 7392
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.433810978537849e-07,
+      "loss": 3.4845,
+      "step": 7393
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.426500173516539e-07,
+      "loss": 3.505,
+      "step": 7394
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.419196883236675e-07,
+      "loss": 3.4284,
+      "step": 7395
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.411901108876653e-07,
+      "loss": 3.5014,
+      "step": 7396
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.404612851613676e-07,
+      "loss": 3.5165,
+      "step": 7397
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3973321126237024e-07,
+      "loss": 3.48,
+      "step": 7398
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3900588930815104e-07,
+      "loss": 3.4895,
+      "step": 7399
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3827931941606385e-07,
+      "loss": 3.5901,
+      "step": 7400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3755350170334434e-07,
+      "loss": 3.6133,
+      "step": 7401
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3682843628710327e-07,
+      "loss": 3.5523,
+      "step": 7402
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.361041232843315e-07,
+      "loss": 3.5351,
+      "step": 7403
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3538056281189945e-07,
+      "loss": 3.5103,
+      "step": 7404
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3465775498655427e-07,
+      "loss": 3.5509,
+      "step": 7405
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3393569992492435e-07,
+      "loss": 3.41,
+      "step": 7406
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3321439774351315e-07,
+      "loss": 3.4993,
+      "step": 7407
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.324938485587059e-07,
+      "loss": 3.4738,
+      "step": 7408
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.31774052486763e-07,
+      "loss": 3.58,
+      "step": 7409
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.310550096438275e-07,
+      "loss": 3.4721,
+      "step": 7410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3033672014591677e-07,
+      "loss": 3.4637,
+      "step": 7411
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2961918410892967e-07,
+      "loss": 3.5492,
+      "step": 7412
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2890240164864085e-07,
+      "loss": 3.3575,
+      "step": 7413
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.281863728807072e-07,
+      "loss": 3.4981,
+      "step": 7414
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2747109792065855e-07,
+      "loss": 3.3899,
+      "step": 7415
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.267565768839093e-07,
+      "loss": 3.5075,
+      "step": 7416
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2604280988574653e-07,
+      "loss": 3.4172,
+      "step": 7417
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2532979704133994e-07,
+      "loss": 3.4648,
+      "step": 7418
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2461753846573463e-07,
+      "loss": 3.5146,
+      "step": 7419
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.239060342738559e-07,
+      "loss": 3.5516,
+      "step": 7420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2319528458050585e-07,
+      "loss": 3.5194,
+      "step": 7421
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.224852895003666e-07,
+      "loss": 3.5277,
+      "step": 7422
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.217760491479971e-07,
+      "loss": 3.5039,
+      "step": 7423
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2106756363783355e-07,
+      "loss": 3.481,
+      "step": 7424
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2035983308419393e-07,
+      "loss": 3.4775,
+      "step": 7425
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1965285760127086e-07,
+      "loss": 3.4993,
+      "step": 7426
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1894663730313637e-07,
+      "loss": 3.5324,
+      "step": 7427
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.182411723037421e-07,
+      "loss": 3.4101,
+      "step": 7428
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.175364627169142e-07,
+      "loss": 3.5206,
+      "step": 7429
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.168325086563612e-07,
+      "loss": 3.3829,
+      "step": 7430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1612931023566664e-07,
+      "loss": 3.6123,
+      "step": 7431
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1542686756829423e-07,
+      "loss": 3.4694,
+      "step": 7432
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1472518076758274e-07,
+      "loss": 3.5378,
+      "step": 7433
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.140242499467533e-07,
+      "loss": 3.4953,
+      "step": 7434
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1332407521890096e-07,
+      "loss": 3.6206,
+      "step": 7435
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1262465669700203e-07,
+      "loss": 3.563,
+      "step": 7436
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.119259944939074e-07,
+      "loss": 3.6537,
+      "step": 7437
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.112280887223501e-07,
+      "loss": 3.5004,
+      "step": 7438
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.105309394949363e-07,
+      "loss": 3.6233,
+      "step": 7439
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.098345469241548e-07,
+      "loss": 3.4636,
+      "step": 7440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0913891112236915e-07,
+      "loss": 3.6294,
+      "step": 7441
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0844403220182227e-07,
+      "loss": 3.4988,
+      "step": 7442
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.077499102746334e-07,
+      "loss": 3.4096,
+      "step": 7443
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.070565454528024e-07,
+      "loss": 3.577,
+      "step": 7444
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0636393784820306e-07,
+      "loss": 3.4276,
+      "step": 7445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.056720875725916e-07,
+      "loss": 3.5022,
+      "step": 7446
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.049809947375981e-07,
+      "loss": 3.5754,
+      "step": 7447
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.042906594547329e-07,
+      "loss": 3.6575,
+      "step": 7448
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0360108183538185e-07,
+      "loss": 3.4963,
+      "step": 7449
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.029122619908115e-07,
+      "loss": 3.4456,
+      "step": 7450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.022242000321629e-07,
+      "loss": 3.5885,
+      "step": 7451
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.015368960704584e-07,
+      "loss": 3.5075,
+      "step": 7452
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.008503502165938e-07,
+      "loss": 3.4637,
+      "step": 7453
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.001645625813471e-07,
+      "loss": 3.4727,
+      "step": 7454
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.994795332753703e-07,
+      "loss": 3.6419,
+      "step": 7455
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9879526240919497e-07,
+      "loss": 3.4404,
+      "step": 7456
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.98111750093229e-07,
+      "loss": 3.4881,
+      "step": 7457
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.974289964377597e-07,
+      "loss": 3.5278,
+      "step": 7458
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.967470015529511e-07,
+      "loss": 3.4847,
+      "step": 7459
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.96065765548843e-07,
+      "loss": 3.5537,
+      "step": 7460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9538528853535643e-07,
+      "loss": 3.6009,
+      "step": 7461
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.947055706222868e-07,
+      "loss": 3.6518,
+      "step": 7462
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9402661191930803e-07,
+      "loss": 3.5432,
+      "step": 7463
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9334841253597203e-07,
+      "loss": 3.5767,
+      "step": 7464
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9267097258170784e-07,
+      "loss": 3.4932,
+      "step": 7465
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9199429216582196e-07,
+      "loss": 3.4697,
+      "step": 7466
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9131837139749764e-07,
+      "loss": 3.4999,
+      "step": 7467
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.906432103857981e-07,
+      "loss": 3.5287,
+      "step": 7468
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.899688092396596e-07,
+      "loss": 3.566,
+      "step": 7469
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.892951680679007e-07,
+      "loss": 3.4593,
+      "step": 7470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.886222869792127e-07,
+      "loss": 3.5044,
+      "step": 7471
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8795016608216876e-07,
+      "loss": 3.5643,
+      "step": 7472
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.872788054852155e-07,
+      "loss": 3.5206,
+      "step": 7473
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8660820529667953e-07,
+      "loss": 3.6356,
+      "step": 7474
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8593836562476274e-07,
+      "loss": 3.5729,
+      "step": 7475
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8526928657754636e-07,
+      "loss": 3.38,
+      "step": 7476
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.846009682629869e-07,
+      "loss": 3.5384,
+      "step": 7477
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8393341078891967e-07,
+      "loss": 3.5446,
+      "step": 7478
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8326661426305637e-07,
+      "loss": 3.4529,
+      "step": 7479
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.826005787929864e-07,
+      "loss": 3.4839,
+      "step": 7480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.81935304486175e-07,
+      "loss": 3.4006,
+      "step": 7481
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.812707914499679e-07,
+      "loss": 3.5606,
+      "step": 7482
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8060703979158334e-07,
+      "loss": 3.5569,
+      "step": 7483
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.799440496181216e-07,
+      "loss": 3.5337,
+      "step": 7484
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.79281821036555e-07,
+      "loss": 3.5259,
+      "step": 7485
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7862035415373855e-07,
+      "loss": 3.5655,
+      "step": 7486
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.779596490763986e-07,
+      "loss": 3.5872,
+      "step": 7487
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.772997059111437e-07,
+      "loss": 3.4777,
+      "step": 7488
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7664052476445604e-07,
+      "loss": 3.4642,
+      "step": 7489
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7598210574269646e-07,
+      "loss": 3.5503,
+      "step": 7490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7532444895210177e-07,
+      "loss": 3.5193,
+      "step": 7491
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7466755449878756e-07,
+      "loss": 3.5185,
+      "step": 7492
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.740114224887441e-07,
+      "loss": 3.4693,
+      "step": 7493
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7335605302784063e-07,
+      "loss": 3.6075,
+      "step": 7494
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.727014462218225e-07,
+      "loss": 3.6222,
+      "step": 7495
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7204760217631074e-07,
+      "loss": 3.5826,
+      "step": 7496
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.713945209968061e-07,
+      "loss": 3.6218,
+      "step": 7497
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.707422027886841e-07,
+      "loss": 3.4662,
+      "step": 7498
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.700906476571974e-07,
+      "loss": 3.4627,
+      "step": 7499
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6943985570747735e-07,
+      "loss": 3.5069,
+      "step": 7500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.687898270445283e-07,
+      "loss": 3.5046,
+      "step": 7501
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6814056177323635e-07,
+      "loss": 3.5122,
+      "step": 7502
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.674920599983605e-07,
+      "loss": 3.5473,
+      "step": 7503
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.668443218245392e-07,
+      "loss": 3.4544,
+      "step": 7504
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6619734735628456e-07,
+      "loss": 3.6078,
+      "step": 7505
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.655511366979896e-07,
+      "loss": 3.5788,
+      "step": 7506
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6490568995391986e-07,
+      "loss": 3.4351,
+      "step": 7507
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6426100722822145e-07,
+      "loss": 3.4702,
+      "step": 7508
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6361708862491396e-07,
+      "loss": 3.5101,
+      "step": 7509
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6297393424789654e-07,
+      "loss": 3.6489,
+      "step": 7510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.623315442009422e-07,
+      "loss": 3.5538,
+      "step": 7511
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.616899185877031e-07,
+      "loss": 3.4679,
+      "step": 7512
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.610490575117064e-07,
+      "loss": 3.5741,
+      "step": 7513
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6040896107635714e-07,
+      "loss": 3.5421,
+      "step": 7514
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5976962938493553e-07,
+      "loss": 3.5343,
+      "step": 7515
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5913106254060015e-07,
+      "loss": 3.5212,
+      "step": 7516
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5849326064638413e-07,
+      "loss": 3.4674,
+      "step": 7517
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.578562238051996e-07,
+      "loss": 3.516,
+      "step": 7518
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.572199521198326e-07,
+      "loss": 3.616,
+      "step": 7519
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.565844456929478e-07,
+      "loss": 3.5806,
+      "step": 7520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5594970462708533e-07,
+      "loss": 3.6167,
+      "step": 7521
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5531572902466274e-07,
+      "loss": 3.5329,
+      "step": 7522
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.546825189879726e-07,
+      "loss": 3.5854,
+      "step": 7523
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.540500746191854e-07,
+      "loss": 3.6535,
+      "step": 7524
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.534183960203468e-07,
+      "loss": 3.5682,
+      "step": 7525
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5278748329338077e-07,
+      "loss": 3.6216,
+      "step": 7526
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.521573365400848e-07,
+      "loss": 3.5773,
+      "step": 7527
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5152795586213694e-07,
+      "loss": 3.5554,
+      "step": 7528
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5089934136108665e-07,
+      "loss": 3.5222,
+      "step": 7529
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.502714931383643e-07,
+      "loss": 3.518,
+      "step": 7530
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4964441129527337e-07,
+      "loss": 3.492,
+      "step": 7531
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.490180959329952e-07,
+      "loss": 3.6015,
+      "step": 7532
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.483925471525883e-07,
+      "loss": 3.589,
+      "step": 7533
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4776776505498477e-07,
+      "loss": 3.4799,
+      "step": 7534
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4714374974099567e-07,
+      "loss": 3.525,
+      "step": 7535
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4652050131130756e-07,
+      "loss": 3.5002,
+      "step": 7536
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.458980198664812e-07,
+      "loss": 3.5366,
+      "step": 7537
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4527630550695793e-07,
+      "loss": 3.3655,
+      "step": 7538
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.446553583330508e-07,
+      "loss": 3.5664,
+      "step": 7539
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.440351784449524e-07,
+      "loss": 3.5031,
+      "step": 7540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4341576594272944e-07,
+      "loss": 3.5265,
+      "step": 7541
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.427971209263258e-07,
+      "loss": 3.4906,
+      "step": 7542
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.421792434955611e-07,
+      "loss": 3.5278,
+      "step": 7543
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.415621337501317e-07,
+      "loss": 3.5748,
+      "step": 7544
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4094579178960854e-07,
+      "loss": 3.5395,
+      "step": 7545
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4033021771344143e-07,
+      "loss": 3.4684,
+      "step": 7546
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3971541162095324e-07,
+      "loss": 3.3566,
+      "step": 7547
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.391013736113462e-07,
+      "loss": 3.557,
+      "step": 7548
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.384881037836945e-07,
+      "loss": 3.4834,
+      "step": 7549
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3787560223695216e-07,
+      "loss": 3.5315,
+      "step": 7550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3726386906994692e-07,
+      "loss": 3.4666,
+      "step": 7551
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3665290438138367e-07,
+      "loss": 3.4104,
+      "step": 7552
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3604270826984298e-07,
+      "loss": 3.5576,
+      "step": 7553
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.354332808337817e-07,
+      "loss": 3.5315,
+      "step": 7554
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3482462217153168e-07,
+      "loss": 3.508,
+      "step": 7555
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3421673238130217e-07,
+      "loss": 3.4689,
+      "step": 7556
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.336096115611769e-07,
+      "loss": 3.4912,
+      "step": 7557
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3300325980911642e-07,
+      "loss": 3.4948,
+      "step": 7558
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3239767722295692e-07,
+      "loss": 3.4393,
+      "step": 7559
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3179286390041134e-07,
+      "loss": 3.5762,
+      "step": 7560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3118881993906607e-07,
+      "loss": 3.4798,
+      "step": 7561
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.30585545436387e-07,
+      "loss": 3.5055,
+      "step": 7562
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2998304048971188e-07,
+      "loss": 3.5673,
+      "step": 7563
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.293813051962579e-07,
+      "loss": 3.4915,
+      "step": 7564
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2878033965311518e-07,
+      "loss": 3.4819,
+      "step": 7565
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2818014395725174e-07,
+      "loss": 3.5215,
+      "step": 7566
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2758071820551064e-07,
+      "loss": 3.508,
+      "step": 7567
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.269820624946095e-07,
+      "loss": 3.4133,
+      "step": 7568
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2638417692114388e-07,
+      "loss": 3.5944,
+      "step": 7569
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2578706158158324e-07,
+      "loss": 3.5766,
+      "step": 7570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2519071657227443e-07,
+      "loss": 3.4827,
+      "step": 7571
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2459514198943876e-07,
+      "loss": 3.5208,
+      "step": 7572
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2400033792917275e-07,
+      "loss": 3.4994,
+      "step": 7573
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2340630448745015e-07,
+      "loss": 3.4078,
+      "step": 7574
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.228130417601193e-07,
+      "loss": 3.5171,
+      "step": 7575
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.222205498429053e-07,
+      "loss": 3.492,
+      "step": 7576
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.216288288314067e-07,
+      "loss": 3.4321,
+      "step": 7577
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2103787882110096e-07,
+      "loss": 3.4601,
+      "step": 7578
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2044769990733683e-07,
+      "loss": 3.5273,
+      "step": 7579
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1985829218534315e-07,
+      "loss": 3.507,
+      "step": 7580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1926965575022107e-07,
+      "loss": 3.5663,
+      "step": 7581
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1868179069694906e-07,
+      "loss": 3.5865,
+      "step": 7582
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.18094697120379e-07,
+      "loss": 3.5427,
+      "step": 7583
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1750837511524235e-07,
+      "loss": 3.5606,
+      "step": 7584
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1692282477614123e-07,
+      "loss": 3.5695,
+      "step": 7585
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1633804619755728e-07,
+      "loss": 3.574,
+      "step": 7586
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1575403947384389e-07,
+      "loss": 3.5626,
+      "step": 7587
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1517080469923403e-07,
+      "loss": 3.4522,
+      "step": 7588
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.145883419678324e-07,
+      "loss": 3.3934,
+      "step": 7589
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1400665137362165e-07,
+      "loss": 3.5978,
+      "step": 7590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1342573301045832e-07,
+      "loss": 3.6077,
+      "step": 7591
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1284558697207524e-07,
+      "loss": 3.5081,
+      "step": 7592
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1226621335207976e-07,
+      "loss": 3.5098,
+      "step": 7593
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.116876122439565e-07,
+      "loss": 3.5632,
+      "step": 7594
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1110978374106195e-07,
+      "loss": 3.493,
+      "step": 7595
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.105327279366326e-07,
+      "loss": 3.5742,
+      "step": 7596
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0995644492377509e-07,
+      "loss": 3.473,
+      "step": 7597
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0938093479547616e-07,
+      "loss": 3.589,
+      "step": 7598
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0880619764459432e-07,
+      "loss": 3.5034,
+      "step": 7599
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.082322335638659e-07,
+      "loss": 3.5422,
+      "step": 7600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.076590426459002e-07,
+      "loss": 3.5377,
+      "step": 7601
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0708662498318377e-07,
+      "loss": 3.4864,
+      "step": 7602
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0651498066807717e-07,
+      "loss": 3.4321,
+      "step": 7603
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0594410979281665e-07,
+      "loss": 3.5454,
+      "step": 7604
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0537401244951293e-07,
+      "loss": 3.5778,
+      "step": 7605
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0480468873015303e-07,
+      "loss": 3.4743,
+      "step": 7606
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0423613872659897e-07,
+      "loss": 3.5332,
+      "step": 7607
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.036683625305874e-07,
+      "loss": 3.5063,
+      "step": 7608
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.031013602337295e-07,
+      "loss": 3.4305,
+      "step": 7609
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0253513192751374e-07,
+      "loss": 3.4714,
+      "step": 7610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.019696777033009e-07,
+      "loss": 3.4948,
+      "step": 7611
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0140499765232968e-07,
+      "loss": 3.5264,
+      "step": 7612
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.008410918657111e-07,
+      "loss": 3.4637,
+      "step": 7613
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.002779604344346e-07,
+      "loss": 3.5467,
+      "step": 7614
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.997156034493608e-07,
+      "loss": 3.5233,
+      "step": 7615
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9915402100122881e-07,
+      "loss": 3.558,
+      "step": 7616
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.985932131806495e-07,
+      "loss": 3.5834,
+      "step": 7617
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9803318007811267e-07,
+      "loss": 3.5097,
+      "step": 7618
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9747392178397885e-07,
+      "loss": 3.5766,
+      "step": 7619
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9691543838848747e-07,
+      "loss": 3.4906,
+      "step": 7620
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9635772998174985e-07,
+      "loss": 3.5083,
+      "step": 7621
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9580079665375395e-07,
+      "loss": 3.4965,
+      "step": 7622
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9524463849436237e-07,
+      "loss": 3.5327,
+      "step": 7623
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9468925559331275e-07,
+      "loss": 3.6442,
+      "step": 7624
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.941346480402162e-07,
+      "loss": 3.5022,
+      "step": 7625
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9358081592456167e-07,
+      "loss": 3.6428,
+      "step": 7626
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9302775933570938e-07,
+      "loss": 3.5826,
+      "step": 7627
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9247547836289792e-07,
+      "loss": 3.5874,
+      "step": 7628
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9192397309523824e-07,
+      "loss": 3.5536,
+      "step": 7629
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.913732436217175e-07,
+      "loss": 3.4666,
+      "step": 7630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.908232900311957e-07,
+      "loss": 3.4821,
+      "step": 7631
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9027411241241133e-07,
+      "loss": 3.5741,
+      "step": 7632
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8972571085397406e-07,
+      "loss": 3.5737,
+      "step": 7633
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8917808544437034e-07,
+      "loss": 3.4975,
+      "step": 7634
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8863123627196002e-07,
+      "loss": 3.4408,
+      "step": 7635
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8808516342497974e-07,
+      "loss": 3.4073,
+      "step": 7636
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8753986699153792e-07,
+      "loss": 3.5753,
+      "step": 7637
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8699534705962085e-07,
+      "loss": 3.5824,
+      "step": 7638
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8645160371708716e-07,
+      "loss": 3.5512,
+      "step": 7639
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8590863705167217e-07,
+      "loss": 3.5279,
+      "step": 7640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8536644715098418e-07,
+      "loss": 3.4632,
+      "step": 7641
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8482503410250598e-07,
+      "loss": 3.5033,
+      "step": 7642
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.842843979935971e-07,
+      "loss": 3.4797,
+      "step": 7643
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8374453891149e-07,
+      "loss": 3.4681,
+      "step": 7644
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8320545694329117e-07,
+      "loss": 3.5158,
+      "step": 7645
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.826671521759843e-07,
+      "loss": 3.5677,
+      "step": 7646
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8212962469642492e-07,
+      "loss": 3.5861,
+      "step": 7647
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8159287459134533e-07,
+      "loss": 3.599,
+      "step": 7648
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8105690194735015e-07,
+      "loss": 3.6499,
+      "step": 7649
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.805217068509213e-07,
+      "loss": 3.3871,
+      "step": 7650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.799872893884125e-07,
+      "loss": 3.5106,
+      "step": 7651
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7945364964605417e-07,
+      "loss": 3.4847,
+      "step": 7652
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7892078770994915e-07,
+      "loss": 3.5112,
+      "step": 7653
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7838870366607752e-07,
+      "loss": 3.5153,
+      "step": 7654
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7785739760029065e-07,
+      "loss": 3.5046,
+      "step": 7655
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7732686959831714e-07,
+      "loss": 3.5621,
+      "step": 7656
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7679711974575798e-07,
+      "loss": 3.5194,
+      "step": 7657
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.762681481280909e-07,
+      "loss": 3.5654,
+      "step": 7658
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7573995483066486e-07,
+      "loss": 3.528,
+      "step": 7659
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7521253993870668e-07,
+      "loss": 3.5819,
+      "step": 7660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7468590353731495e-07,
+      "loss": 3.4887,
+      "step": 7661
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.741600457114645e-07,
+      "loss": 3.4343,
+      "step": 7662
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7363496654600298e-07,
+      "loss": 3.4652,
+      "step": 7663
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7311066612565374e-07,
+      "loss": 3.5475,
+      "step": 7664
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7258714453501245e-07,
+      "loss": 3.5457,
+      "step": 7665
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7206440185855267e-07,
+      "loss": 3.6083,
+      "step": 7666
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7154243818061854e-07,
+      "loss": 3.4547,
+      "step": 7667
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7102125358543108e-07,
+      "loss": 3.5118,
+      "step": 7668
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7050084815708356e-07,
+      "loss": 3.5844,
+      "step": 7669
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6998122197954548e-07,
+      "loss": 3.4227,
+      "step": 7670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.694623751366592e-07,
+      "loss": 3.5849,
+      "step": 7671
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6894430771214277e-07,
+      "loss": 3.6274,
+      "step": 7672
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6842701978958597e-07,
+      "loss": 3.4288,
+      "step": 7673
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6791051145245596e-07,
+      "loss": 3.6487,
+      "step": 7674
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.673947827840916e-07,
+      "loss": 3.339,
+      "step": 7675
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6687983386770746e-07,
+      "loss": 3.4579,
+      "step": 7676
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.663656647863915e-07,
+      "loss": 3.5075,
+      "step": 7677
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.658522756231057e-07,
+      "loss": 3.519,
+      "step": 7678
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6533966646068711e-07,
+      "loss": 3.4942,
+      "step": 7679
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6482783738184628e-07,
+      "loss": 3.5234,
+      "step": 7680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6431678846916766e-07,
+      "loss": 3.5382,
+      "step": 7681
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6380651980511086e-07,
+      "loss": 3.5895,
+      "step": 7682
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6329703147200783e-07,
+      "loss": 3.462,
+      "step": 7683
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.627883235520672e-07,
+      "loss": 3.5115,
+      "step": 7684
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6228039612736835e-07,
+      "loss": 3.545,
+      "step": 7685
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6177324927986848e-07,
+      "loss": 3.4631,
+      "step": 7686
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.612668830913955e-07,
+      "loss": 3.5168,
+      "step": 7687
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6076129764365344e-07,
+      "loss": 3.5474,
+      "step": 7688
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6025649301821877e-07,
+      "loss": 3.4375,
+      "step": 7689
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5975246929654408e-07,
+      "loss": 3.562,
+      "step": 7690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5924922655995379e-07,
+      "loss": 3.4499,
+      "step": 7691
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5874676488964847e-07,
+      "loss": 3.5404,
+      "step": 7692
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5824508436669995e-07,
+      "loss": 3.5382,
+      "step": 7693
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.577441850720568e-07,
+      "loss": 3.4885,
+      "step": 7694
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5724406708653938e-07,
+      "loss": 3.6008,
+      "step": 7695
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.567447304908437e-07,
+      "loss": 3.5357,
+      "step": 7696
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5624617536553755e-07,
+      "loss": 3.4713,
+      "step": 7697
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5574840179106598e-07,
+      "loss": 3.545,
+      "step": 7698
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5525140984774368e-07,
+      "loss": 3.5613,
+      "step": 7699
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5475519961576314e-07,
+      "loss": 3.4922,
+      "step": 7700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.542597711751881e-07,
+      "loss": 3.5482,
+      "step": 7701
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5376512460595795e-07,
+      "loss": 3.5763,
+      "step": 7702
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5327125998788384e-07,
+      "loss": 3.5915,
+      "step": 7703
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5277817740065315e-07,
+      "loss": 3.4547,
+      "step": 7704
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5228587692382447e-07,
+      "loss": 3.6067,
+      "step": 7705
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5179435863683367e-07,
+      "loss": 3.5514,
+      "step": 7706
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5130362261898624e-07,
+      "loss": 3.539,
+      "step": 7707
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.508136689494649e-07,
+      "loss": 3.5938,
+      "step": 7708
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5032449770732428e-07,
+      "loss": 3.5723,
+      "step": 7709
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4983610897149336e-07,
+      "loss": 3.5486,
+      "step": 7710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4934850282077473e-07,
+      "loss": 3.5127,
+      "step": 7711
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4886167933384544e-07,
+      "loss": 3.5519,
+      "step": 7712
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4837563858925484e-07,
+      "loss": 3.4665,
+      "step": 7713
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4789038066542637e-07,
+      "loss": 3.5312,
+      "step": 7714
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4740590564065905e-07,
+      "loss": 3.4957,
+      "step": 7715
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4692221359312199e-07,
+      "loss": 3.4604,
+      "step": 7716
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4643930460086164e-07,
+      "loss": 3.5491,
+      "step": 7717
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4595717874179572e-07,
+      "loss": 3.5707,
+      "step": 7718
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.454758360937164e-07,
+      "loss": 3.5244,
+      "step": 7719
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4499527673428993e-07,
+      "loss": 3.57,
+      "step": 7720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4451550074105481e-07,
+      "loss": 3.5657,
+      "step": 7721
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4403650819142468e-07,
+      "loss": 3.5347,
+      "step": 7722
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4355829916268604e-07,
+      "loss": 3.5364,
+      "step": 7723
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4308087373199886e-07,
+      "loss": 3.485,
+      "step": 7724
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.426042319763965e-07,
+      "loss": 3.5582,
+      "step": 7725
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.421283739727869e-07,
+      "loss": 3.5248,
+      "step": 7726
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4165329979794972e-07,
+      "loss": 3.4296,
+      "step": 7727
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.411790095285409e-07,
+      "loss": 3.4562,
+      "step": 7728
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.40705503241087e-07,
+      "loss": 3.5258,
+      "step": 7729
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4023278101199024e-07,
+      "loss": 3.6132,
+      "step": 7730
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3976084291752458e-07,
+      "loss": 3.4517,
+      "step": 7731
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.392896890338391e-07,
+      "loss": 3.5005,
+      "step": 7732
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3881931943695413e-07,
+      "loss": 3.5458,
+      "step": 7733
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.383497342027673e-07,
+      "loss": 3.544,
+      "step": 7734
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3788093340704466e-07,
+      "loss": 3.6142,
+      "step": 7735
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.374129171254307e-07,
+      "loss": 3.4496,
+      "step": 7736
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.369456854334389e-07,
+      "loss": 3.5584,
+      "step": 7737
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3647923840645948e-07,
+      "loss": 3.4995,
+      "step": 7738
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.360135761197534e-07,
+      "loss": 3.5434,
+      "step": 7739
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.355486986484583e-07,
+      "loss": 3.4687,
+      "step": 7740
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3508460606758145e-07,
+      "loss": 3.5905,
+      "step": 7741
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.346212984520062e-07,
+      "loss": 3.4203,
+      "step": 7742
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3415877587648784e-07,
+      "loss": 3.4623,
+      "step": 7743
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3369703841565663e-07,
+      "loss": 3.5053,
+      "step": 7744
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3323608614401295e-07,
+      "loss": 3.5476,
+      "step": 7745
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.32775919135934e-07,
+      "loss": 3.5737,
+      "step": 7746
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.323165374656682e-07,
+      "loss": 3.5501,
+      "step": 7747
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.318579412073384e-07,
+      "loss": 3.4211,
+      "step": 7748
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3140013043493992e-07,
+      "loss": 3.4857,
+      "step": 7749
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3094310522234032e-07,
+      "loss": 3.5492,
+      "step": 7750
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3048686564328395e-07,
+      "loss": 3.5875,
+      "step": 7751
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.300314117713841e-07,
+      "loss": 3.48,
+      "step": 7752
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2957674368013095e-07,
+      "loss": 3.532,
+      "step": 7753
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2912286144288521e-07,
+      "loss": 3.5437,
+      "step": 7754
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2866976513288165e-07,
+      "loss": 3.5032,
+      "step": 7755
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2821745482322901e-07,
+      "loss": 3.5424,
+      "step": 7756
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2776593058690833e-07,
+      "loss": 3.4935,
+      "step": 7757
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2731519249677416e-07,
+      "loss": 3.4804,
+      "step": 7758
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2686524062555383e-07,
+      "loss": 3.5496,
+      "step": 7759
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.264160750458493e-07,
+      "loss": 3.5329,
+      "step": 7760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.259676958301326e-07,
+      "loss": 3.5211,
+      "step": 7761
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2552010305075247e-07,
+      "loss": 3.4669,
+      "step": 7762
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2507329677992786e-07,
+      "loss": 3.634,
+      "step": 7763
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2462727708975274e-07,
+      "loss": 3.5076,
+      "step": 7764
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2418204405219236e-07,
+      "loss": 3.5422,
+      "step": 7765
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2373759773908756e-07,
+      "loss": 3.5718,
+      "step": 7766
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2329393822214985e-07,
+      "loss": 3.4552,
+      "step": 7767
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2285106557296479e-07,
+      "loss": 3.5638,
+      "step": 7768
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2240897986299128e-07,
+      "loss": 3.5443,
+      "step": 7769
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.219676811635606e-07,
+      "loss": 3.4755,
+      "step": 7770
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2152716954587695e-07,
+      "loss": 3.5096,
+      "step": 7771
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2108744508101844e-07,
+      "loss": 3.5913,
+      "step": 7772
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2064850783993555e-07,
+      "loss": 3.5148,
+      "step": 7773
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2021035789345158e-07,
+      "loss": 3.6154,
+      "step": 7774
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1977299531226283e-07,
+      "loss": 3.409,
+      "step": 7775
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1933642016694002e-07,
+      "loss": 3.6102,
+      "step": 7776
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1890063252792406e-07,
+      "loss": 3.4813,
+      "step": 7777
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.184656324655309e-07,
+      "loss": 3.4924,
+      "step": 7778
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1803142004994827e-07,
+      "loss": 3.5143,
+      "step": 7779
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1759799535123906e-07,
+      "loss": 3.5911,
+      "step": 7780
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1716535843933507e-07,
+      "loss": 3.5387,
+      "step": 7781
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1673350938404493e-07,
+      "loss": 3.537,
+      "step": 7782
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1630244825504733e-07,
+      "loss": 3.486,
+      "step": 7783
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1587217512189553e-07,
+      "loss": 3.4548,
+      "step": 7784
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1544269005401509e-07,
+      "loss": 3.6084,
+      "step": 7785
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1501399312070504e-07,
+      "loss": 3.5422,
+      "step": 7786
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1458608439113561e-07,
+      "loss": 3.4223,
+      "step": 7787
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.14158963934351e-07,
+      "loss": 3.5913,
+      "step": 7788
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1373263181926831e-07,
+      "loss": 3.4874,
+      "step": 7789
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.133070881146775e-07,
+      "loss": 3.5519,
+      "step": 7790
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1288233288924033e-07,
+      "loss": 3.4431,
+      "step": 7791
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1245836621149308e-07,
+      "loss": 3.4685,
+      "step": 7792
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1203518814984216e-07,
+      "loss": 3.5388,
+      "step": 7793
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1161279877257014e-07,
+      "loss": 3.5027,
+      "step": 7794
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1119119814782863e-07,
+      "loss": 3.4398,
+      "step": 7795
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1077038634364601e-07,
+      "loss": 3.6699,
+      "step": 7796
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1035036342791905e-07,
+      "loss": 3.4586,
+      "step": 7797
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0993112946842133e-07,
+      "loss": 3.5885,
+      "step": 7798
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0951268453279595e-07,
+      "loss": 3.6046,
+      "step": 7799
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.090950286885606e-07,
+      "loss": 3.5624,
+      "step": 7800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.086781620031041e-07,
+      "loss": 3.6888,
+      "step": 7801
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0826208454368992e-07,
+      "loss": 3.4836,
+      "step": 7802
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0784679637745265e-07,
+      "loss": 3.4449,
+      "step": 7803
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0743229757140039e-07,
+      "loss": 3.4609,
+      "step": 7804
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0701858819241296e-07,
+      "loss": 3.5734,
+      "step": 7805
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0660566830724362e-07,
+      "loss": 3.6607,
+      "step": 7806
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0619353798251741e-07,
+      "loss": 3.4661,
+      "step": 7807
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0578219728473338e-07,
+      "loss": 3.6212,
+      "step": 7808
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0537164628026119e-07,
+      "loss": 3.4845,
+      "step": 7809
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0496188503534566e-07,
+      "loss": 3.6429,
+      "step": 7810
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.045529136161011e-07,
+      "loss": 3.5539,
+      "step": 7811
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.041447320885175e-07,
+      "loss": 3.5225,
+      "step": 7812
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0373734051845497e-07,
+      "loss": 3.5063,
+      "step": 7813
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0333073897164702e-07,
+      "loss": 3.5549,
+      "step": 7814
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0292492751370009e-07,
+      "loss": 3.4631,
+      "step": 7815
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0251990621009289e-07,
+      "loss": 3.5967,
+      "step": 7816
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0211567512617649e-07,
+      "loss": 3.4253,
+      "step": 7817
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0171223432717425e-07,
+      "loss": 3.6037,
+      "step": 7818
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.013095838781819e-07,
+      "loss": 3.4938,
+      "step": 7819
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0090772384416914e-07,
+      "loss": 3.4893,
+      "step": 7820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0050665428997575e-07,
+      "loss": 3.5415,
+      "step": 7821
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0010637528031664e-07,
+      "loss": 3.6172,
+      "step": 7822
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.970688687977681e-08,
+      "loss": 3.601,
+      "step": 7823
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.93081891528147e-08,
+      "loss": 3.4673,
+      "step": 7824
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.891028216376108e-08,
+      "loss": 3.5332,
+      "step": 7825
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.851316597681959e-08,
+      "loss": 3.4906,
+      "step": 7826
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.811684065606509e-08,
+      "loss": 3.4482,
+      "step": 7827
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.77213062654464e-08,
+      "loss": 3.6144,
+      "step": 7828
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.732656286878305e-08,
+      "loss": 3.4503,
+      "step": 7829
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.69326105297691e-08,
+      "loss": 3.5463,
+      "step": 7830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.653944931196813e-08,
+      "loss": 3.5923,
+      "step": 7831
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.614707927881827e-08,
+      "loss": 3.5869,
+      "step": 7832
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.575550049362947e-08,
+      "loss": 3.5653,
+      "step": 7833
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.53647130195845e-08,
+      "loss": 3.526,
+      "step": 7834
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.497471691973625e-08,
+      "loss": 3.5511,
+      "step": 7835
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.458551225701273e-08,
+      "loss": 3.4273,
+      "step": 7836
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.419709909421204e-08,
+      "loss": 3.5876,
+      "step": 7837
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.380947749400682e-08,
+      "loss": 3.5015,
+      "step": 7838
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.342264751893981e-08,
+      "loss": 3.4667,
+      "step": 7839
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.303660923142721e-08,
+      "loss": 3.4573,
+      "step": 7840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.265136269375641e-08,
+      "loss": 3.4684,
+      "step": 7841
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.226690796808934e-08,
+      "loss": 3.4788,
+      "step": 7842
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.188324511645752e-08,
+      "loss": 3.4554,
+      "step": 7843
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.150037420076641e-08,
+      "loss": 3.5239,
+      "step": 7844
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.111829528279215e-08,
+      "loss": 3.5298,
+      "step": 7845
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.073700842418542e-08,
+      "loss": 3.5131,
+      "step": 7846
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.035651368646647e-08,
+      "loss": 3.4559,
+      "step": 7847
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.997681113103007e-08,
+      "loss": 3.5362,
+      "step": 7848
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.95979008191411e-08,
+      "loss": 3.6076,
+      "step": 7849
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.921978281193844e-08,
+      "loss": 3.5389,
+      "step": 7850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.884245717043216e-08,
+      "loss": 3.6252,
+      "step": 7851
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.846592395550412e-08,
+      "loss": 3.5287,
+      "step": 7852
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.809018322790908e-08,
+      "loss": 3.5004,
+      "step": 7853
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.771523504827461e-08,
+      "loss": 3.5037,
+      "step": 7854
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.734107947709735e-08,
+      "loss": 3.4833,
+      "step": 7855
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.696771657475068e-08,
+      "loss": 3.5081,
+      "step": 7856
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.659514640147526e-08,
+      "loss": 3.5151,
+      "step": 7857
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.622336901738749e-08,
+      "loss": 3.3449,
+      "step": 7858
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.585238448247434e-08,
+      "loss": 3.497,
+      "step": 7859
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.548219285659464e-08,
+      "loss": 3.4161,
+      "step": 7860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.51127941994806e-08,
+      "loss": 3.414,
+      "step": 7861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.474418857073396e-08,
+      "loss": 3.4455,
+      "step": 7862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.437637602983162e-08,
+      "loss": 3.405,
+      "step": 7863
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.400935663611997e-08,
+      "loss": 3.6848,
+      "step": 7864
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.364313044881888e-08,
+      "loss": 3.5088,
+      "step": 7865
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.327769752701941e-08,
+      "loss": 3.4796,
+      "step": 7866
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.291305792968607e-08,
+      "loss": 3.5133,
+      "step": 7867
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.254921171565233e-08,
+      "loss": 3.5384,
+      "step": 7868
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.218615894362792e-08,
+      "loss": 3.5152,
+      "step": 7869
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.18238996721904e-08,
+      "loss": 3.5142,
+      "step": 7870
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.146243395979248e-08,
+      "loss": 3.6393,
+      "step": 7871
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.110176186475583e-08,
+      "loss": 3.4801,
+      "step": 7872
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.074188344527778e-08,
+      "loss": 3.6549,
+      "step": 7873
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.038279875942413e-08,
+      "loss": 3.6188,
+      "step": 7874
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.00245078651346e-08,
+      "loss": 3.4777,
+      "step": 7875
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.966701082021966e-08,
+      "loss": 3.494,
+      "step": 7876
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.931030768236314e-08,
+      "loss": 3.4837,
+      "step": 7877
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.895439850911901e-08,
+      "loss": 3.5704,
+      "step": 7878
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.859928335791467e-08,
+      "loss": 3.4022,
+      "step": 7879
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.824496228604872e-08,
+      "loss": 3.5349,
+      "step": 7880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.789143535069154e-08,
+      "loss": 3.5269,
+      "step": 7881
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.75387026088853e-08,
+      "loss": 3.5477,
+      "step": 7882
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.7186764117545e-08,
+      "loss": 3.6157,
+      "step": 7883
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.683561993345522e-08,
+      "loss": 3.5243,
+      "step": 7884
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.648527011327566e-08,
+      "loss": 3.4602,
+      "step": 7885
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.613571471353498e-08,
+      "loss": 3.461,
+      "step": 7886
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.578695379063527e-08,
+      "loss": 3.4335,
+      "step": 7887
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.543898740084931e-08,
+      "loss": 3.6393,
+      "step": 7888
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.509181560032274e-08,
+      "loss": 3.5205,
+      "step": 7889
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.474543844507243e-08,
+      "loss": 3.5151,
+      "step": 7890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.439985599098699e-08,
+      "loss": 3.4669,
+      "step": 7891
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.405506829382736e-08,
+      "loss": 3.5018,
+      "step": 7892
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.371107540922517e-08,
+      "loss": 3.548,
+      "step": 7893
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.336787739268491e-08,
+      "loss": 3.5419,
+      "step": 7894
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.302547429958284e-08,
+      "loss": 3.5652,
+      "step": 7895
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.268386618516531e-08,
+      "loss": 3.5764,
+      "step": 7896
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.23430531045527e-08,
+      "loss": 3.4113,
+      "step": 7897
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.200303511273488e-08,
+      "loss": 3.5482,
+      "step": 7898
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.166381226457575e-08,
+      "loss": 3.5599,
+      "step": 7899
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.132538461480931e-08,
+      "loss": 3.483,
+      "step": 7900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.098775221804077e-08,
+      "loss": 3.4396,
+      "step": 7901
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.065091512874934e-08,
+      "loss": 3.6735,
+      "step": 7902
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.031487340128318e-08,
+      "loss": 3.4467,
+      "step": 7903
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.997962708986339e-08,
+      "loss": 3.4276,
+      "step": 7904
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.964517624858391e-08,
+      "loss": 3.5461,
+      "step": 7905
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.931152093140825e-08,
+      "loss": 3.549,
+      "step": 7906
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.897866119217278e-08,
+      "loss": 3.4888,
+      "step": 7907
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.86465970845851e-08,
+      "loss": 3.5103,
+      "step": 7908
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.831532866222457e-08,
+      "loss": 3.5745,
+      "step": 7909
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.798485597854176e-08,
+      "loss": 3.4387,
+      "step": 7910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.76551790868596e-08,
+      "loss": 3.5223,
+      "step": 7911
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.732629804037216e-08,
+      "loss": 3.4594,
+      "step": 7912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.699821289214537e-08,
+      "loss": 3.5295,
+      "step": 7913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.667092369511574e-08,
+      "loss": 3.5304,
+      "step": 7914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.634443050209327e-08,
+      "loss": 3.4649,
+      "step": 7915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.601873336575748e-08,
+      "loss": 3.6004,
+      "step": 7916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.569383233866078e-08,
+      "loss": 3.5054,
+      "step": 7917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.536972747322623e-08,
+      "loss": 3.4956,
+      "step": 7918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.504641882174977e-08,
+      "loss": 3.5106,
+      "step": 7919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.472390643639693e-08,
+      "loss": 3.5893,
+      "step": 7920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.440219036920658e-08,
+      "loss": 3.4607,
+      "step": 7921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.408127067208835e-08,
+      "loss": 3.5174,
+      "step": 7922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.376114739682304e-08,
+      "loss": 3.5888,
+      "step": 7923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.344182059506266e-08,
+      "loss": 3.5827,
+      "step": 7924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.31232903183332e-08,
+      "loss": 3.4813,
+      "step": 7925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.280555661802857e-08,
+      "loss": 3.4771,
+      "step": 7926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.248861954541607e-08,
+      "loss": 3.5544,
+      "step": 7927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.217247915163482e-08,
+      "loss": 3.4822,
+      "step": 7928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.185713548769457e-08,
+      "loss": 3.5456,
+      "step": 7929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.154258860447682e-08,
+      "loss": 3.4047,
+      "step": 7930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.122883855273432e-08,
+      "loss": 3.5415,
+      "step": 7931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.091588538309046e-08,
+      "loss": 3.4989,
+      "step": 7932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.060372914604263e-08,
+      "loss": 3.4587,
+      "step": 7933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.029236989195719e-08,
+      "loss": 3.6048,
+      "step": 7934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.998180767107231e-08,
+      "loss": 3.5013,
+      "step": 7935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.9672042533499e-08,
+      "loss": 3.4607,
+      "step": 7936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.9363074529216746e-08,
+      "loss": 3.4571,
+      "step": 7937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.905490370808009e-08,
+      "loss": 3.5071,
+      "step": 7938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.87475301198126e-08,
+      "loss": 3.5285,
+      "step": 7939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.844095381400849e-08,
+      "loss": 3.4595,
+      "step": 7940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.8135174840135975e-08,
+      "loss": 3.5284,
+      "step": 7941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.7830193247532807e-08,
+      "loss": 3.5874,
+      "step": 7942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.752600908540851e-08,
+      "loss": 3.5969,
+      "step": 7943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.7222622402843266e-08,
+      "loss": 3.5532,
+      "step": 7944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.692003324879069e-08,
+      "loss": 3.5379,
+      "step": 7945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.66182416720723e-08,
+      "loss": 3.6074,
+      "step": 7946
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.631724772138469e-08,
+      "loss": 3.496,
+      "step": 7947
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.6017051445292345e-08,
+      "loss": 3.5993,
+      "step": 7948
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.571765289223374e-08,
+      "loss": 3.5616,
+      "step": 7949
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.54190521105169e-08,
+      "loss": 3.6225,
+      "step": 7950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.512124914832273e-08,
+      "loss": 3.4878,
+      "step": 7951
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.482424405370057e-08,
+      "loss": 3.624,
+      "step": 7952
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.452803687457486e-08,
+      "loss": 3.5618,
+      "step": 7953
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.423262765873794e-08,
+      "loss": 3.5624,
+      "step": 7954
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.3938016453856103e-08,
+      "loss": 3.4598,
+      "step": 7955
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.3644203307464115e-08,
+      "loss": 3.5181,
+      "step": 7956
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.3351188266970167e-08,
+      "loss": 3.4064,
+      "step": 7957
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.305897137965199e-08,
+      "loss": 3.5951,
+      "step": 7958
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.2767552692661314e-08,
+      "loss": 3.6542,
+      "step": 7959
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.24769322530172e-08,
+      "loss": 3.4666,
+      "step": 7960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.2187110107613794e-08,
+      "loss": 3.5545,
+      "step": 7961
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.189808630321258e-08,
+      "loss": 3.612,
+      "step": 7962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.160986088645015e-08,
+      "loss": 3.4913,
+      "step": 7963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.132243390383096e-08,
+      "loss": 3.5706,
+      "step": 7964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.103580540173292e-08,
+      "loss": 3.5615,
+      "step": 7965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.074997542640292e-08,
+      "loss": 3.5158,
+      "step": 7966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.0464944023961845e-08,
+      "loss": 3.5281,
+      "step": 7967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.018071124039958e-08,
+      "loss": 3.5476,
+      "step": 7968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.989727712157721e-08,
+      "loss": 3.5755,
+      "step": 7969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.961464171322761e-08,
+      "loss": 3.4821,
+      "step": 7970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.933280506095595e-08,
+      "loss": 3.4284,
+      "step": 7971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.90517672102353e-08,
+      "loss": 3.5459,
+      "step": 7972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8771528206412734e-08,
+      "loss": 3.4466,
+      "step": 7973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8492088094705956e-08,
+      "loss": 3.6743,
+      "step": 7974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.821344692020225e-08,
+      "loss": 3.4728,
+      "step": 7975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.793560472786174e-08,
+      "loss": 3.473,
+      "step": 7976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.765856156251469e-08,
+      "loss": 3.5469,
+      "step": 7977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.738231746886202e-08,
+      "loss": 3.547,
+      "step": 7978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.710687249147749e-08,
+      "loss": 3.5324,
+      "step": 7979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6832226674803916e-08,
+      "loss": 3.5241,
+      "step": 7980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.655838006315694e-08,
+      "loss": 3.6227,
+      "step": 7981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6285332700720665e-08,
+      "loss": 3.4451,
+      "step": 7982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6013084631554294e-08,
+      "loss": 3.7209,
+      "step": 7983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.574163589958325e-08,
+      "loss": 3.5832,
+      "step": 7984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5470986548608046e-08,
+      "loss": 3.6296,
+      "step": 7985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5201136622297636e-08,
+      "loss": 3.5362,
+      "step": 7986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4932086164193865e-08,
+      "loss": 3.4584,
+      "step": 7987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.466383521770812e-08,
+      "loss": 3.5505,
+      "step": 7988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4396383826123544e-08,
+      "loss": 3.4526,
+      "step": 7989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4129732032592854e-08,
+      "loss": 3.5555,
+      "step": 7990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3863879880142737e-08,
+      "loss": 3.4235,
+      "step": 7991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3598827411667766e-08,
+      "loss": 3.4514,
+      "step": 7992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.333457466993596e-08,
+      "loss": 3.5459,
+      "step": 7993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.307112169758376e-08,
+      "loss": 3.6537,
+      "step": 7994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2808468537120505e-08,
+      "loss": 3.5976,
+      "step": 7995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.254661523092562e-08,
+      "loss": 3.521,
+      "step": 7996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.228556182125032e-08,
+      "loss": 3.5003,
+      "step": 7997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2025308350215896e-08,
+      "loss": 3.5245,
+      "step": 7998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.176585485981488e-08,
+      "loss": 3.6258,
+      "step": 7999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.150720139191045e-08,
+      "loss": 3.4537,
+      "step": 8000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1249347988237544e-08,
+      "loss": 3.5076,
+      "step": 8001
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.099229469040011e-08,
+      "loss": 3.5128,
+      "step": 8002
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0736041539876094e-08,
+      "loss": 3.5574,
+      "step": 8003
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0480588578010735e-08,
+      "loss": 3.4596,
+      "step": 8004
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.02259358460233e-08,
+      "loss": 3.5843,
+      "step": 8005
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9972083385002023e-08,
+      "loss": 3.5654,
+      "step": 8006
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.971903123590692e-08,
+      "loss": 3.4566,
+      "step": 8007
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9466779439568116e-08,
+      "loss": 3.576,
+      "step": 8008
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9215328036686925e-08,
+      "loss": 3.477,
+      "step": 8009
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.896467706783591e-08,
+      "loss": 3.5915,
+      "step": 8010
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.871482657345826e-08,
+      "loss": 3.4645,
+      "step": 8011
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.846577659386785e-08,
+      "loss": 3.5764,
+      "step": 8012
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.821752716924976e-08,
+      "loss": 3.6414,
+      "step": 8013
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.79700783396586e-08,
+      "loss": 3.5705,
+      "step": 8014
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.772343014502189e-08,
+      "loss": 3.6825,
+      "step": 8015
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.747758262513612e-08,
+      "loss": 3.477,
+      "step": 8016
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7232535819670124e-08,
+      "loss": 3.5117,
+      "step": 8017
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6988289768162265e-08,
+      "loss": 3.5246,
+      "step": 8018
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6744844510022134e-08,
+      "loss": 3.5229,
+      "step": 8019
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.650220008452998e-08,
+      "loss": 3.532,
+      "step": 8020
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.626035653083837e-08,
+      "loss": 3.4748,
+      "step": 8021
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.601931388796776e-08,
+      "loss": 3.5072,
+      "step": 8022
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.577907219481147e-08,
+      "loss": 3.5007,
+      "step": 8023
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.553963149013295e-08,
+      "loss": 3.5023,
+      "step": 8024
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.530099181256685e-08,
+      "loss": 3.4316,
+      "step": 8025
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.506315320061793e-08,
+      "loss": 3.4622,
+      "step": 8026
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.48261156926627e-08,
+      "loss": 3.4583,
+      "step": 8027
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.458987932694668e-08,
+      "loss": 3.6517,
+      "step": 8028
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.435444414158773e-08,
+      "loss": 3.5503,
+      "step": 8029
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.411981017457322e-08,
+      "loss": 3.3776,
+      "step": 8030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.388597746376287e-08,
+      "loss": 3.6206,
+      "step": 8031
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3652946046885404e-08,
+      "loss": 3.6303,
+      "step": 8032
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.342071596154184e-08,
+      "loss": 3.6167,
+      "step": 8033
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3189287245201654e-08,
+      "loss": 3.5426,
+      "step": 8034
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2958659935207196e-08,
+      "loss": 3.3219,
+      "step": 8035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.272883406877092e-08,
+      "loss": 3.5659,
+      "step": 8036
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.249980968297595e-08,
+      "loss": 3.4155,
+      "step": 8037
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.227158681477438e-08,
+      "loss": 3.5052,
+      "step": 8038
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2044165500992316e-08,
+      "loss": 3.5312,
+      "step": 8039
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1817545778323744e-08,
+      "loss": 3.5839,
+      "step": 8040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.159172768333496e-08,
+      "loss": 3.5521,
+      "step": 8041
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.136671125246127e-08,
+      "loss": 3.5026,
+      "step": 8042
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.114249652201029e-08,
+      "loss": 3.4223,
+      "step": 8043
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.09190835281592e-08,
+      "loss": 3.4737,
+      "step": 8044
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0696472306956935e-08,
+      "loss": 3.498,
+      "step": 8045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.047466289432144e-08,
+      "loss": 3.5617,
+      "step": 8046
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.025365532604241e-08,
+      "loss": 3.5995,
+      "step": 8047
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.003344963778021e-08,
+      "loss": 3.5842,
+      "step": 8048
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9814045865065864e-08,
+      "loss": 3.5372,
+      "step": 8049
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9595444043299392e-08,
+      "loss": 3.4238,
+      "step": 8050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9377644207753686e-08,
+      "loss": 3.5023,
+      "step": 8051
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.916064639357119e-08,
+      "loss": 3.4989,
+      "step": 8052
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8944450635764453e-08,
+      "loss": 3.4669,
+      "step": 8053
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8729056969217794e-08,
+      "loss": 3.4855,
+      "step": 8054
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8514465428685634e-08,
+      "loss": 3.6261,
+      "step": 8055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8300676048791942e-08,
+      "loss": 3.5481,
+      "step": 8056
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8087688864033014e-08,
+      "loss": 3.4653,
+      "step": 8057
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7875503908774136e-08,
+      "loss": 3.4933,
+      "step": 8058
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.766412121725237e-08,
+      "loss": 3.5526,
+      "step": 8059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.745354082357432e-08,
+      "loss": 3.4779,
+      "step": 8060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7243762761717806e-08,
+      "loss": 3.7541,
+      "step": 8061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7034787065531308e-08,
+      "loss": 3.5429,
+      "step": 8062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6826613768732856e-08,
+      "loss": 3.4666,
+      "step": 8063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6619242904912246e-08,
+      "loss": 3.5467,
+      "step": 8064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.641267450752938e-08,
+      "loss": 3.5313,
+      "step": 8065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.620690860991426e-08,
+      "loss": 3.5465,
+      "step": 8066
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6001945245267558e-08,
+      "loss": 3.4411,
+      "step": 8067
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.579778444666059e-08,
+      "loss": 3.5746,
+      "step": 8068
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5594426247035896e-08,
+      "loss": 3.5488,
+      "step": 8069
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5391870679204455e-08,
+      "loss": 3.468,
+      "step": 8070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.519011777585012e-08,
+      "loss": 3.5078,
+      "step": 8071
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.498916756952574e-08,
+      "loss": 3.4948,
+      "step": 8072
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4789020092655936e-08,
+      "loss": 3.4871,
+      "step": 8073
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4589675377533205e-08,
+      "loss": 3.5067,
+      "step": 8074
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4391133456324044e-08,
+      "loss": 3.4497,
+      "step": 8075
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.419339436106283e-08,
+      "loss": 3.5198,
+      "step": 8076
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.39964581236557e-08,
+      "loss": 3.4987,
+      "step": 8077
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.380032477587779e-08,
+      "loss": 3.5431,
+      "step": 8078
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3604994349376576e-08,
+      "loss": 3.5863,
+      "step": 8079
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3410466875668503e-08,
+      "loss": 3.5243,
+      "step": 8080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.32167423861418e-08,
+      "loss": 3.5693,
+      "step": 8081
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3023820912053685e-08,
+      "loss": 3.5566,
+      "step": 8082
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2831702484532035e-08,
+      "loss": 3.4873,
+      "step": 8083
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.264038713457706e-08,
+      "loss": 3.4664,
+      "step": 8084
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2449874893056832e-08,
+      "loss": 3.5257,
+      "step": 8085
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.226016579071122e-08,
+      "loss": 3.4652,
+      "step": 8086
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2071259858150175e-08,
+      "loss": 3.5895,
+      "step": 8087
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.188315712585376e-08,
+      "loss": 3.5521,
+      "step": 8088
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.169585762417381e-08,
+      "loss": 3.4693,
+      "step": 8089
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1509361383330597e-08,
+      "loss": 3.4641,
+      "step": 8090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1323668433416154e-08,
+      "loss": 3.4729,
+      "step": 8091
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1138778804392078e-08,
+      "loss": 3.3823,
+      "step": 8092
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.095469252609117e-08,
+      "loss": 3.542,
+      "step": 8093
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.077140962821578e-08,
+      "loss": 3.553,
+      "step": 8094
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.058893014033947e-08,
+      "loss": 3.3847,
+      "step": 8095
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0407254091905358e-08,
+      "loss": 3.3879,
+      "step": 8096
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.022638151222722e-08,
+      "loss": 3.5516,
+      "step": 8097
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.004631243048949e-08,
+      "loss": 3.4204,
+      "step": 8098
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9867046875747255e-08,
+      "loss": 3.6088,
+      "step": 8099
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9688584876924043e-08,
+      "loss": 3.6225,
+      "step": 8100
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9510926462816825e-08,
+      "loss": 3.461,
+      "step": 8101
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9334071662089337e-08,
+      "loss": 3.4989,
+      "step": 8102
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9158020503279308e-08,
+      "loss": 3.5948,
+      "step": 8103
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.898277301479179e-08,
+      "loss": 3.5369,
+      "step": 8104
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8808329224904166e-08,
+      "loss": 3.4995,
+      "step": 8105
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8634689161762254e-08,
+      "loss": 3.4738,
+      "step": 8106
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8461852853384198e-08,
+      "loss": 3.5302,
+      "step": 8107
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.828982032765769e-08,
+      "loss": 3.5878,
+      "step": 8108
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.811859161233942e-08,
+      "loss": 3.5979,
+      "step": 8109
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7948166735058947e-08,
+      "loss": 3.5846,
+      "step": 8110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7778545723313724e-08,
+      "loss": 3.5886,
+      "step": 8111
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7609728604472966e-08,
+      "loss": 3.4926,
+      "step": 8112
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7441715405775438e-08,
+      "loss": 3.6066,
+      "step": 8113
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7274506154330573e-08,
+      "loss": 3.4197,
+      "step": 8114
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.710810087711845e-08,
+      "loss": 3.6282,
+      "step": 8115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6942499600987595e-08,
+      "loss": 3.6414,
+      "step": 8116
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6777702352659965e-08,
+      "loss": 3.3906,
+      "step": 8117
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.661370915872429e-08,
+      "loss": 3.3442,
+      "step": 8118
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.645052004564218e-08,
+      "loss": 3.4905,
+      "step": 8119
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.62881350397448e-08,
+      "loss": 3.4471,
+      "step": 8120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6126554167232854e-08,
+      "loss": 3.6166,
+      "step": 8121
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5965777454177154e-08,
+      "loss": 3.6438,
+      "step": 8122
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.580580492652084e-08,
+      "loss": 3.5056,
+      "step": 8123
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.564663661007493e-08,
+      "loss": 3.5951,
+      "step": 8124
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5488272530521653e-08,
+      "loss": 3.5485,
+      "step": 8125
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5330712713413908e-08,
+      "loss": 3.425,
+      "step": 8126
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5173957184173583e-08,
+      "loss": 3.5073,
+      "step": 8127
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5018005968093775e-08,
+      "loss": 3.5365,
+      "step": 8128
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4862859090338244e-08,
+      "loss": 3.613,
+      "step": 8129
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4708516575939746e-08,
+      "loss": 3.4945,
+      "step": 8130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4554978449801693e-08,
+      "loss": 3.5276,
+      "step": 8131
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4402244736697602e-08,
+      "loss": 3.5542,
+      "step": 8132
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4250315461272201e-08,
+      "loss": 3.5893,
+      "step": 8133
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4099190648039218e-08,
+      "loss": 3.6315,
+      "step": 8134
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3948870321383034e-08,
+      "loss": 3.5321,
+      "step": 8135
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3799354505557582e-08,
+      "loss": 3.5514,
+      "step": 8136
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3650643224688565e-08,
+      "loss": 3.5477,
+      "step": 8137
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.350273650277012e-08,
+      "loss": 3.4529,
+      "step": 8138
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3355634363668158e-08,
+      "loss": 3.478,
+      "step": 8139
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3209336831117025e-08,
+      "loss": 3.5859,
+      "step": 8140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3063843928722841e-08,
+      "loss": 3.4464,
+      "step": 8141
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2919155679960716e-08,
+      "loss": 3.4847,
+      "step": 8142
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2775272108176972e-08,
+      "loss": 3.4749,
+      "step": 8143
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2632193236586932e-08,
+      "loss": 3.4795,
+      "step": 8144
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.248991908827768e-08,
+      "loss": 3.5252,
+      "step": 8145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.234844968620419e-08,
+      "loss": 3.5268,
+      "step": 8146
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2207785053193755e-08,
+      "loss": 3.5292,
+      "step": 8147
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2067925211942665e-08,
+      "loss": 3.5732,
+      "step": 8148
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1928870185017872e-08,
+      "loss": 3.5611,
+      "step": 8149
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1790619994855867e-08,
+      "loss": 3.6125,
+      "step": 8150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1653174663763811e-08,
+      "loss": 3.5282,
+      "step": 8151
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1516534213918406e-08,
+      "loss": 3.5681,
+      "step": 8152
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1380698667367574e-08,
+      "loss": 3.5621,
+      "step": 8153
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.124566804602878e-08,
+      "loss": 3.4641,
+      "step": 8154
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1111442371688486e-08,
+      "loss": 3.4526,
+      "step": 8155
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0978021666005479e-08,
+      "loss": 3.4767,
+      "step": 8156
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0845405950506982e-08,
+      "loss": 3.4162,
+      "step": 8157
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.071359524659088e-08,
+      "loss": 3.5516,
+      "step": 8158
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.058258957552516e-08,
+      "loss": 3.6013,
+      "step": 8159
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0452388958447912e-08,
+      "loss": 3.6261,
+      "step": 8160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0322993416366778e-08,
+      "loss": 3.5775,
+      "step": 8161
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.019440297016061e-08,
+      "loss": 3.5235,
+      "step": 8162
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.006661764057837e-08,
+      "loss": 3.3685,
+      "step": 8163
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.939637448236894e-09,
+      "loss": 3.43,
+      "step": 8164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.813462413626462e-09,
+      "loss": 3.5452,
+      "step": 8165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.688092557104123e-09,
+      "loss": 3.4781,
+      "step": 8166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.563527898899804e-09,
+      "loss": 3.5787,
+      "step": 8167
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.439768459111876e-09,
+      "loss": 3.5834,
+      "step": 8168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.316814257708807e-09,
+      "loss": 3.5467,
+      "step": 8169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.194665314530283e-09,
+      "loss": 3.4874,
+      "step": 8170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.073321649284982e-09,
+      "loss": 3.5389,
+      "step": 8171
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.952783281551691e-09,
+      "loss": 3.4976,
+      "step": 8172
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.833050230779849e-09,
+      "loss": 3.3651,
+      "step": 8173
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.714122516288447e-09,
+      "loss": 3.4474,
+      "step": 8174
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.596000157267136e-09,
+      "loss": 3.4159,
+      "step": 8175
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.478683172775115e-09,
+      "loss": 3.4611,
+      "step": 8176
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.362171581741685e-09,
+      "loss": 3.6631,
+      "step": 8177
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.246465402966252e-09,
+      "loss": 3.4988,
+      "step": 8178
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.131564655117774e-09,
+      "loss": 3.5414,
+      "step": 8179
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.017469356736974e-09,
+      "loss": 3.5236,
+      "step": 8180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.904179526231902e-09,
+      "loss": 3.5105,
+      "step": 8181
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.791695181883496e-09,
+      "loss": 3.4678,
+      "step": 8182
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.680016341840567e-09,
+      "loss": 3.4681,
+      "step": 8183
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.569143024122594e-09,
+      "loss": 3.4339,
+      "step": 8184
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.459075246619707e-09,
+      "loss": 3.5889,
+      "step": 8185
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.34981302709159e-09,
+      "loss": 3.3805,
+      "step": 8186
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.24135638316803e-09,
+      "loss": 3.4242,
+      "step": 8187
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.133705332348362e-09,
+      "loss": 3.3609,
+      "step": 8188
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.02685989200258e-09,
+      "loss": 3.4985,
+      "step": 8189
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.920820079370227e-09,
+      "loss": 3.5966,
+      "step": 8190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.815585911560952e-09,
+      "loss": 3.4842,
+      "step": 8191
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.711157405555612e-09,
+      "loss": 3.4698,
+      "step": 8192
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.607534578202401e-09,
+      "loss": 3.4905,
+      "step": 8193
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.504717446222386e-09,
+      "loss": 3.4167,
+      "step": 8194
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.402706026204519e-09,
+      "loss": 3.613,
+      "step": 8195
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.3015003346095225e-09,
+      "loss": 3.4214,
+      "step": 8196
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.201100387766001e-09,
+      "loss": 3.5878,
+      "step": 8197
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.10150620187433e-09,
+      "loss": 3.467,
+      "step": 8198
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.002717793004431e-09,
+      "loss": 3.5728,
+      "step": 8199
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.904735177095222e-09,
+      "loss": 3.4795,
+      "step": 8200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.807558369957389e-09,
+      "loss": 3.4747,
+      "step": 8201
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.711187387270056e-09,
+      "loss": 3.425,
+      "step": 8202
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.615622244583563e-09,
+      "loss": 3.5683,
+      "step": 8203
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.520862957316686e-09,
+      "loss": 3.5563,
+      "step": 8204
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.426909540759973e-09,
+      "loss": 3.5234,
+      "step": 8205
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.333762010071852e-09,
+      "loss": 3.5842,
+      "step": 8206
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.24142038028308e-09,
+      "loss": 3.5387,
+      "step": 8207
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.149884666292293e-09,
+      "loss": 3.5241,
+      "step": 8208
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.059154882869899e-09,
+      "loss": 3.4862,
+      "step": 8209
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.969231044654188e-09,
+      "loss": 3.4625,
+      "step": 8210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8801131661557756e-09,
+      "loss": 3.5921,
+      "step": 8211
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.79180126175316e-09,
+      "loss": 3.5119,
+      "step": 8212
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.704295345696053e-09,
+      "loss": 3.6174,
+      "step": 8213
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.617595432103716e-09,
+      "loss": 3.4827,
+      "step": 8214
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5317015349649605e-09,
+      "loss": 3.4679,
+      "step": 8215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.44661366813981e-09,
+      "loss": 3.5302,
+      "step": 8216
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.362331845356726e-09,
+      "loss": 3.5904,
+      "step": 8217
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.278856080214833e-09,
+      "loss": 3.4853,
+      "step": 8218
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.196186386183909e-09,
+      "loss": 3.4779,
+      "step": 8219
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1143227766016206e-09,
+      "loss": 3.5432,
+      "step": 8220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0332652646774e-09,
+      "loss": 3.5572,
+      "step": 8221
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.953013863490784e-09,
+      "loss": 3.5974,
+      "step": 8222
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.873568585989196e-09,
+      "loss": 3.5653,
+      "step": 8223
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.794929444992379e-09,
+      "loss": 3.6379,
+      "step": 8224
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.717096453188518e-09,
+      "loss": 3.5254,
+      "step": 8225
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6400696231364552e-09,
+      "loss": 3.5275,
+      "step": 8226
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5638489672640273e-09,
+      "loss": 3.433,
+      "step": 8227
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4884344978702857e-09,
+      "loss": 3.4952,
+      "step": 8228
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.413826227123274e-09,
+      "loss": 3.555,
+      "step": 8229
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3400241670605847e-09,
+      "loss": 3.5768,
+      "step": 8230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2670283295915815e-09,
+      "loss": 3.4382,
+      "step": 8231
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.194838726492955e-09,
+      "loss": 3.5414,
+      "step": 8232
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1234553694137193e-09,
+      "loss": 3.4757,
+      "step": 8233
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.052878269871329e-09,
+      "loss": 3.5084,
+      "step": 8234
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.983107439253896e-09,
+      "loss": 3.488,
+      "step": 8235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9141428888185275e-09,
+      "loss": 3.5378,
+      "step": 8236
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8459846296929885e-09,
+      "loss": 3.583,
+      "step": 8237
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.778632672875148e-09,
+      "loss": 3.4486,
+      "step": 8238
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.712087029232424e-09,
+      "loss": 3.4898,
+      "step": 8239
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.646347709501229e-09,
+      "loss": 3.6458,
+      "step": 8240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5814147242897438e-09,
+      "loss": 3.4174,
+      "step": 8241
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5172880840745873e-09,
+      "loss": 3.3906,
+      "step": 8242
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4539677992024834e-09,
+      "loss": 3.5661,
+      "step": 8243
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.391453879890815e-09,
+      "loss": 3.5734,
+      "step": 8244
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.329746336226513e-09,
+      "loss": 3.4414,
+      "step": 8245
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2688451781655017e-09,
+      "loss": 3.5719,
+      "step": 8246
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2087504155349217e-09,
+      "loss": 3.4916,
+      "step": 8247
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1494620580309046e-09,
+      "loss": 3.5262,
+      "step": 8248
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0909801152196875e-09,
+      "loss": 3.5015,
+      "step": 8249
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0333045965376108e-09,
+      "loss": 3.5768,
+      "step": 8250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9764355112911192e-09,
+      "loss": 3.5618,
+      "step": 8251
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.920372868655651e-09,
+      "loss": 3.5902,
+      "step": 8252
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8651166776773033e-09,
+      "loss": 3.55,
+      "step": 8253
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.810666947271167e-09,
+      "loss": 3.5455,
+      "step": 8254
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.757023686224102e-09,
+      "loss": 3.5436,
+      "step": 8255
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7041869031902968e-09,
+      "loss": 3.4541,
+      "step": 8256
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6521566066957096e-09,
+      "loss": 3.5004,
+      "step": 8257
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6009328051358464e-09,
+      "loss": 3.4838,
+      "step": 8258
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5505155067752076e-09,
+      "loss": 3.4746,
+      "step": 8259
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5009047197489523e-09,
+      "loss": 3.6522,
+      "step": 8260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4521004520617887e-09,
+      "loss": 3.4914,
+      "step": 8261
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4041027115885286e-09,
+      "loss": 3.4633,
+      "step": 8262
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3569115060735328e-09,
+      "loss": 3.475,
+      "step": 8263
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3105268431312657e-09,
+      "loss": 3.55,
+      "step": 8264
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.264948730246296e-09,
+      "loss": 3.5342,
+      "step": 8265
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.220177174772741e-09,
+      "loss": 3.4732,
+      "step": 8266
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1762121839342666e-09,
+      "loss": 3.5074,
+      "step": 8267
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1330537648246431e-09,
+      "loss": 3.5328,
+      "step": 8268
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0907019244077444e-09,
+      "loss": 3.5528,
+      "step": 8269
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0491566695175481e-09,
+      "loss": 3.5803,
+      "step": 8270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0084180068570259e-09,
+      "loss": 3.5807,
+      "step": 8271
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.684859429998083e-10,
+      "loss": 3.3919,
+      "step": 8272
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.293604843879645e-10,
+      "loss": 3.4856,
+      "step": 8273
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.910416373358876e-10,
+      "loss": 3.4929,
+      "step": 8274
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.535294080258549e-10,
+      "loss": 3.6297,
+      "step": 8275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.16823802510247e-10,
+      "loss": 3.4683,
+      "step": 8276
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.809248267121039e-10,
+      "loss": 3.5095,
+      "step": 8277
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.458324864234589e-10,
+      "loss": 3.5654,
+      "step": 8278
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.115467873064497e-10,
+      "loss": 3.5742,
+      "step": 8279
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.780677348933173e-10,
+      "loss": 3.6072,
+      "step": 8280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.453953345864073e-10,
+      "loss": 3.5229,
+      "step": 8281
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.135295916565032e-10,
+      "loss": 3.544,
+      "step": 8282
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.824705112461582e-10,
+      "loss": 3.3632,
+      "step": 8283
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.522180983663639e-10,
+      "loss": 3.5183,
+      "step": 8284
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.227723578987709e-10,
+      "loss": 3.5085,
+      "step": 8285
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.941332945940236e-10,
+      "loss": 3.4826,
+      "step": 8286
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.663009130734253e-10,
+      "loss": 3.513,
+      "step": 8287
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.392752178278281e-10,
+      "loss": 3.5734,
+      "step": 8288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.13056213217633e-10,
+      "loss": 3.4905,
+      "step": 8289
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.876439034733448e-10,
+      "loss": 3.5668,
+      "step": 8290
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6303829269557224e-10,
+      "loss": 3.5184,
+      "step": 8291
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3923938485447283e-10,
+      "loss": 3.4883,
+      "step": 8292
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1624718378975297e-10,
+      "loss": 3.5609,
+      "step": 8293
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.940616932112228e-10,
+      "loss": 3.4687,
+      "step": 8294
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7268291669935167e-10,
+      "loss": 3.526,
+      "step": 8295
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5211085770249223e-10,
+      "loss": 3.6013,
+      "step": 8296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3234551954132155e-10,
+      "loss": 3.463,
+      "step": 8297
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.133869054038451e-10,
+      "loss": 3.4975,
+      "step": 8298
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9523501834928239e-10,
+      "loss": 3.5389,
+      "step": 8299
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7788986130695683e-10,
+      "loss": 3.4849,
+      "step": 8300
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6135143707518564e-10,
+      "loss": 3.5093,
+      "step": 8301
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.456197483229449e-10,
+      "loss": 3.4946,
+      "step": 8302
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3069479758764936e-10,
+      "loss": 3.6513,
+      "step": 8303
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1657658727848297e-10,
+      "loss": 3.5318,
+      "step": 8304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.032651196730683e-10,
+      "loss": 3.4732,
+      "step": 8305
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.076039691913174e-11,
+      "loss": 3.6217,
+      "step": 8306
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.906242103394857e-11,
+      "loss": 3.4463,
+      "step": 8307
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.817119390600813e-11,
+      "loss": 3.3884,
+      "step": 8308
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.8086717291683247e-11,
+      "loss": 3.4328,
+      "step": 8309
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8808992819116e-11,
+      "loss": 3.4468,
+      "step": 8310
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0338021984331896e-11,
+      "loss": 3.5195,
+      "step": 8311
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.267380615457061e-11,
+      "loss": 3.6545,
+      "step": 8312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5816346566620576e-11,
+      "loss": 3.5145,
+      "step": 8313
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9765644326819043e-11,
+      "loss": 3.5084,
+      "step": 8314
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4521700411052054e-11,
+      "loss": 3.5813,
+      "step": 8315
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0084515665864658e-11,
+      "loss": 3.6081,
+      "step": 8316
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.45409080735071e-12,
+      "loss": 3.579,
+      "step": 8317
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6304264205977434e-12,
+      "loss": 3.5335,
+      "step": 8318
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.613522961907421e-12,
+      "loss": 3.5328,
+      "step": 8319
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0338075657508913e-13,
+      "loss": 3.5833,
+      "step": 8320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 3.3968,
+      "step": 8321
+    },
+    {
+      "epoch": 1.0,
+      "step": 8321,
+      "total_flos": 1.0728709803317658e+18,
+      "train_loss": 3.564992360377624,
+      "train_runtime": 9434.0892,
+      "train_samples_per_second": 56.448,
+      "train_steps_per_second": 0.882
+    }
+  ],
+  "max_steps": 8321,
+  "num_train_epochs": 1,
+  "total_flos": 1.0728709803317658e+18,
+  "trial_name": null,
+  "trial_params": null
+}