diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,26803 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "global_step": 4460,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0000000000000004e-08,
+      "loss": 10.529,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 10.5167,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-08,
+      "loss": 10.5158,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 10.5187,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5000000000000002e-07,
+      "loss": 10.497,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8e-07,
+      "loss": 10.4678,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1e-07,
+      "loss": 10.512,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 10.4455,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7e-07,
+      "loss": 10.4465,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 10.4156,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2999999999999996e-07,
+      "loss": 10.4298,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6e-07,
+      "loss": 10.393,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8999999999999997e-07,
+      "loss": 10.371,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2e-07,
+      "loss": 10.3616,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5e-07,
+      "loss": 10.3207,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 10.3288,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.100000000000001e-07,
+      "loss": 10.3079,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4e-07,
+      "loss": 10.2617,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.7e-07,
+      "loss": 10.2496,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 10.2294,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.3e-07,
+      "loss": 10.2105,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.599999999999999e-07,
+      "loss": 10.1808,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.9e-07,
+      "loss": 10.1141,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.2e-07,
+      "loss": 10.1473,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.5e-07,
+      "loss": 10.1055,
+      "step": 25
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.799999999999999e-07,
+      "loss": 10.1081,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.1e-07,
+      "loss": 10.0573,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.4e-07,
+      "loss": 10.0225,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.7e-07,
+      "loss": 9.9779,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9e-07,
+      "loss": 9.9896,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.3e-07,
+      "loss": 9.9245,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 9.8712,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9e-07,
+      "loss": 9.8643,
+      "step": 33
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0200000000000002e-06,
+      "loss": 9.8699,
+      "step": 34
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.0500000000000001e-06,
+      "loss": 9.8367,
+      "step": 35
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.08e-06,
+      "loss": 9.8433,
+      "step": 36
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.11e-06,
+      "loss": 9.7988,
+      "step": 37
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.14e-06,
+      "loss": 9.7289,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.17e-06,
+      "loss": 9.7412,
+      "step": 39
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 9.6646,
+      "step": 40
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.23e-06,
+      "loss": 9.6418,
+      "step": 41
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.26e-06,
+      "loss": 9.6116,
+      "step": 42
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.29e-06,
+      "loss": 9.6084,
+      "step": 43
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.3199999999999999e-06,
+      "loss": 9.7013,
+      "step": 44
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.35e-06,
+      "loss": 9.6684,
+      "step": 45
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.38e-06,
+      "loss": 9.6288,
+      "step": 46
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.41e-06,
+      "loss": 9.6658,
+      "step": 47
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.44e-06,
+      "loss": 9.5059,
+      "step": 48
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.4700000000000001e-06,
+      "loss": 9.4185,
+      "step": 49
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5e-06,
+      "loss": 9.2665,
+      "step": 50
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.53e-06,
+      "loss": 9.5826,
+      "step": 51
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.5599999999999999e-06,
+      "loss": 9.5248,
+      "step": 52
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.59e-06,
+      "loss": 9.5629,
+      "step": 53
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.62e-06,
+      "loss": 9.6173,
+      "step": 54
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.65e-06,
+      "loss": 9.5804,
+      "step": 55
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.68e-06,
+      "loss": 9.5053,
+      "step": 56
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.7100000000000001e-06,
+      "loss": 9.5027,
+      "step": 57
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.74e-06,
+      "loss": 9.5158,
+      "step": 58
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.77e-06,
+      "loss": 9.4397,
+      "step": 59
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8e-06,
+      "loss": 9.4931,
+      "step": 60
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.83e-06,
+      "loss": 9.451,
+      "step": 61
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.86e-06,
+      "loss": 9.4017,
+      "step": 62
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.8900000000000001e-06,
+      "loss": 9.4151,
+      "step": 63
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 9.4236,
+      "step": 64
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.95e-06,
+      "loss": 9.4092,
+      "step": 65
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98e-06,
+      "loss": 9.4781,
+      "step": 66
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0100000000000002e-06,
+      "loss": 9.4191,
+      "step": 67
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.0400000000000004e-06,
+      "loss": 9.4503,
+      "step": 68
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.07e-06,
+      "loss": 9.29,
+      "step": 69
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 9.3118,
+      "step": 70
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.13e-06,
+      "loss": 9.3452,
+      "step": 71
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.16e-06,
+      "loss": 9.3762,
+      "step": 72
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1899999999999998e-06,
+      "loss": 9.2777,
+      "step": 73
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.22e-06,
+      "loss": 9.3375,
+      "step": 74
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.25e-06,
+      "loss": 9.1992,
+      "step": 75
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.28e-06,
+      "loss": 9.3094,
+      "step": 76
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.31e-06,
+      "loss": 9.2139,
+      "step": 77
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.34e-06,
+      "loss": 9.1552,
+      "step": 78
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.37e-06,
+      "loss": 9.1974,
+      "step": 79
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 9.2261,
+      "step": 80
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.43e-06,
+      "loss": 9.113,
+      "step": 81
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.46e-06,
+      "loss": 9.1176,
+      "step": 82
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.4900000000000003e-06,
+      "loss": 9.2032,
+      "step": 83
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.52e-06,
+      "loss": 9.138,
+      "step": 84
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.55e-06,
+      "loss": 9.2007,
+      "step": 85
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.58e-06,
+      "loss": 9.1338,
+      "step": 86
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.61e-06,
+      "loss": 9.1411,
+      "step": 87
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.6399999999999997e-06,
+      "loss": 9.0539,
+      "step": 88
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.67e-06,
+      "loss": 9.0898,
+      "step": 89
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7e-06,
+      "loss": 9.0746,
+      "step": 90
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.73e-06,
+      "loss": 9.1864,
+      "step": 91
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.76e-06,
+      "loss": 9.0876,
+      "step": 92
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.79e-06,
+      "loss": 9.0136,
+      "step": 93
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.82e-06,
+      "loss": 9.0275,
+      "step": 94
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8500000000000002e-06,
+      "loss": 9.1147,
+      "step": 95
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.88e-06,
+      "loss": 9.013,
+      "step": 96
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.91e-06,
+      "loss": 8.9371,
+      "step": 97
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 9.0037,
+      "step": 98
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9700000000000004e-06,
+      "loss": 8.7562,
+      "step": 99
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3e-06,
+      "loss": 8.5471,
+      "step": 100
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.0300000000000002e-06,
+      "loss": 9.1415,
+      "step": 101
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.06e-06,
+      "loss": 9.1779,
+      "step": 102
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.09e-06,
+      "loss": 9.1485,
+      "step": 103
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.1199999999999998e-06,
+      "loss": 9.0309,
+      "step": 104
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.15e-06,
+      "loss": 8.9754,
+      "step": 105
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.18e-06,
+      "loss": 9.0324,
+      "step": 106
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.21e-06,
+      "loss": 8.9699,
+      "step": 107
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.24e-06,
+      "loss": 9.0339,
+      "step": 108
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.27e-06,
+      "loss": 8.947,
+      "step": 109
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.3e-06,
+      "loss": 8.9981,
+      "step": 110
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.3300000000000003e-06,
+      "loss": 9.0209,
+      "step": 111
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.36e-06,
+      "loss": 8.934,
+      "step": 112
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.39e-06,
+      "loss": 8.8992,
+      "step": 113
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.4200000000000003e-06,
+      "loss": 8.937,
+      "step": 114
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.4500000000000004e-06,
+      "loss": 8.9259,
+      "step": 115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.48e-06,
+      "loss": 8.8993,
+      "step": 116
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.5100000000000003e-06,
+      "loss": 8.966,
+      "step": 117
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.54e-06,
+      "loss": 8.9703,
+      "step": 118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.57e-06,
+      "loss": 8.9275,
+      "step": 119
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.6e-06,
+      "loss": 8.9634,
+      "step": 120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.63e-06,
+      "loss": 8.8947,
+      "step": 121
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.66e-06,
+      "loss": 8.9257,
+      "step": 122
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.6900000000000002e-06,
+      "loss": 8.8275,
+      "step": 123
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.72e-06,
+      "loss": 8.8836,
+      "step": 124
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.75e-06,
+      "loss": 8.9333,
+      "step": 125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 8.9088,
+      "step": 126
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.81e-06,
+      "loss": 9.0233,
+      "step": 127
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 8.9532,
+      "step": 128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.87e-06,
+      "loss": 8.9071,
+      "step": 129
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.9e-06,
+      "loss": 8.8505,
+      "step": 130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.9300000000000005e-06,
+      "loss": 8.871,
+      "step": 131
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.96e-06,
+      "loss": 8.8608,
+      "step": 132
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.99e-06,
+      "loss": 8.8102,
+      "step": 133
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 8.8467,
+      "step": 134
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.05e-06,
+      "loss": 8.814,
+      "step": 135
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.080000000000001e-06,
+      "loss": 8.7656,
+      "step": 136
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.1100000000000005e-06,
+      "loss": 8.838,
+      "step": 137
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.14e-06,
+      "loss": 8.7502,
+      "step": 138
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.170000000000001e-06,
+      "loss": 8.7373,
+      "step": 139
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 8.7223,
+      "step": 140
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.229999999999999e-06,
+      "loss": 8.9108,
+      "step": 141
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.26e-06,
+      "loss": 8.696,
+      "step": 142
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.29e-06,
+      "loss": 8.7124,
+      "step": 143
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.32e-06,
+      "loss": 8.6667,
+      "step": 144
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.35e-06,
+      "loss": 8.5797,
+      "step": 145
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.3799999999999996e-06,
+      "loss": 8.5669,
+      "step": 146
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.41e-06,
+      "loss": 8.627,
+      "step": 147
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.44e-06,
+      "loss": 8.3217,
+      "step": 148
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.4699999999999996e-06,
+      "loss": 8.4527,
+      "step": 149
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.5e-06,
+      "loss": 8.1771,
+      "step": 150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.53e-06,
+      "loss": 8.8076,
+      "step": 151
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.56e-06,
+      "loss": 8.7071,
+      "step": 152
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.59e-06,
+      "loss": 8.7251,
+      "step": 153
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.62e-06,
+      "loss": 8.6946,
+      "step": 154
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.65e-06,
+      "loss": 8.6911,
+      "step": 155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.68e-06,
+      "loss": 8.7156,
+      "step": 156
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.71e-06,
+      "loss": 8.7567,
+      "step": 157
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.74e-06,
+      "loss": 8.7413,
+      "step": 158
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.77e-06,
+      "loss": 8.6944,
+      "step": 159
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 8.6725,
+      "step": 160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.83e-06,
+      "loss": 8.6318,
+      "step": 161
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.86e-06,
+      "loss": 8.6436,
+      "step": 162
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.890000000000001e-06,
+      "loss": 8.6433,
+      "step": 163
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.92e-06,
+      "loss": 8.6602,
+      "step": 164
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.95e-06,
+      "loss": 8.6162,
+      "step": 165
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 8.6242,
+      "step": 166
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.01e-06,
+      "loss": 8.6139,
+      "step": 167
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.04e-06,
+      "loss": 8.5709,
+      "step": 168
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.070000000000001e-06,
+      "loss": 8.5922,
+      "step": 169
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.1e-06,
+      "loss": 8.542,
+      "step": 170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.130000000000001e-06,
+      "loss": 8.6199,
+      "step": 171
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.16e-06,
+      "loss": 8.6007,
+      "step": 172
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.1899999999999994e-06,
+      "loss": 8.6201,
+      "step": 173
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.22e-06,
+      "loss": 8.594,
+      "step": 174
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.25e-06,
+      "loss": 8.4997,
+      "step": 175
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.279999999999999e-06,
+      "loss": 8.4324,
+      "step": 176
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.31e-06,
+      "loss": 8.4921,
+      "step": 177
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.34e-06,
+      "loss": 8.4763,
+      "step": 178
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.37e-06,
+      "loss": 8.5424,
+      "step": 179
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.4e-06,
+      "loss": 8.4919,
+      "step": 180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.43e-06,
+      "loss": 8.5249,
+      "step": 181
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.46e-06,
+      "loss": 8.4566,
+      "step": 182
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.49e-06,
+      "loss": 8.4394,
+      "step": 183
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.52e-06,
+      "loss": 8.5215,
+      "step": 184
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.55e-06,
+      "loss": 8.3932,
+      "step": 185
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.58e-06,
+      "loss": 8.4914,
+      "step": 186
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.6100000000000005e-06,
+      "loss": 8.4904,
+      "step": 187
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.64e-06,
+      "loss": 8.3994,
+      "step": 188
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.67e-06,
+      "loss": 8.4209,
+      "step": 189
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.7000000000000005e-06,
+      "loss": 8.402,
+      "step": 190
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.73e-06,
+      "loss": 8.3921,
+      "step": 191
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.76e-06,
+      "loss": 8.478,
+      "step": 192
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.7900000000000005e-06,
+      "loss": 8.3737,
+      "step": 193
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.82e-06,
+      "loss": 8.4118,
+      "step": 194
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.850000000000001e-06,
+      "loss": 8.1417,
+      "step": 195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 8.4181,
+      "step": 196
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.91e-06,
+      "loss": 8.1943,
+      "step": 197
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.940000000000001e-06,
+      "loss": 8.1377,
+      "step": 198
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9700000000000004e-06,
+      "loss": 7.9296,
+      "step": 199
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 6e-06,
+      "loss": 7.9374,
+      "step": 200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.030000000000001e-06,
+      "loss": 8.428,
+      "step": 201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 8.5026,
+      "step": 202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.090000000000001e-06,
+      "loss": 8.4035,
+      "step": 203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.12e-06,
+      "loss": 8.3529,
+      "step": 204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.1499999999999996e-06,
+      "loss": 8.3053,
+      "step": 205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.18e-06,
+      "loss": 8.3312,
+      "step": 206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.21e-06,
+      "loss": 8.3801,
+      "step": 207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.2399999999999995e-06,
+      "loss": 8.2363,
+      "step": 208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 6.27e-06,
+      "loss": 8.283,
+      "step": 209
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.3e-06,
+      "loss": 8.2906,
+      "step": 210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.3299999999999995e-06,
+      "loss": 8.2369,
+      "step": 211
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.36e-06,
+      "loss": 8.2395,
+      "step": 212
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.39e-06,
+      "loss": 8.317,
+      "step": 213
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.42e-06,
+      "loss": 8.2515,
+      "step": 214
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.45e-06,
+      "loss": 8.3046,
+      "step": 215
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.48e-06,
+      "loss": 8.2814,
+      "step": 216
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.51e-06,
+      "loss": 8.2115,
+      "step": 217
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 6.54e-06,
+      "loss": 8.1977,
+      "step": 218
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.57e-06,
+      "loss": 8.1126,
+      "step": 219
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.6e-06,
+      "loss": 8.2439,
+      "step": 220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.63e-06,
+      "loss": 8.1836,
+      "step": 221
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 8.1651,
+      "step": 222
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.69e-06,
+      "loss": 8.2667,
+      "step": 223
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.72e-06,
+      "loss": 8.2517,
+      "step": 224
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.750000000000001e-06,
+      "loss": 8.1514,
+      "step": 225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.78e-06,
+      "loss": 8.2232,
+      "step": 226
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 6.81e-06,
+      "loss": 8.1607,
+      "step": 227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.840000000000001e-06,
+      "loss": 8.1334,
+      "step": 228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.87e-06,
+      "loss": 8.1493,
+      "step": 229
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.900000000000001e-06,
+      "loss": 8.1795,
+      "step": 230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.9300000000000006e-06,
+      "loss": 8.105,
+      "step": 231
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.96e-06,
+      "loss": 8.1961,
+      "step": 232
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 6.990000000000001e-06,
+      "loss": 8.1514,
+      "step": 233
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 8.0986,
+      "step": 234
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.049999999999999e-06,
+      "loss": 8.1813,
+      "step": 235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 7.08e-06,
+      "loss": 7.9642,
+      "step": 236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.11e-06,
+      "loss": 8.0265,
+      "step": 237
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.14e-06,
+      "loss": 8.1702,
+      "step": 238
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.17e-06,
+      "loss": 8.011,
+      "step": 239
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.2e-06,
+      "loss": 8.0396,
+      "step": 240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.23e-06,
+      "loss": 8.0753,
+      "step": 241
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.26e-06,
+      "loss": 8.0009,
+      "step": 242
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.29e-06,
+      "loss": 7.9315,
+      "step": 243
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.32e-06,
+      "loss": 7.8408,
+      "step": 244
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 7.35e-06,
+      "loss": 7.9122,
+      "step": 245
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 7.7945,
+      "step": 246
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.41e-06,
+      "loss": 7.952,
+      "step": 247
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.44e-06,
+      "loss": 7.9202,
+      "step": 248
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.4700000000000005e-06,
+      "loss": 7.6324,
+      "step": 249
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.5e-06,
+      "loss": 7.5338,
+      "step": 250
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.53e-06,
+      "loss": 8.182,
+      "step": 251
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 8.1326,
+      "step": 252
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.59e-06,
+      "loss": 8.0491,
+      "step": 253
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 7.62e-06,
+      "loss": 8.1417,
+      "step": 254
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.65e-06,
+      "loss": 7.9578,
+      "step": 255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 7.9728,
+      "step": 256
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.71e-06,
+      "loss": 7.9388,
+      "step": 257
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.74e-06,
+      "loss": 7.9501,
+      "step": 258
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.77e-06,
+      "loss": 7.8949,
+      "step": 259
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.8e-06,
+      "loss": 7.9416,
+      "step": 260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.830000000000001e-06,
+      "loss": 7.9403,
+      "step": 261
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 7.8516,
+      "step": 262
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 7.89e-06,
+      "loss": 7.934,
+      "step": 263
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.92e-06,
+      "loss": 7.9562,
+      "step": 264
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.95e-06,
+      "loss": 7.9109,
+      "step": 265
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.98e-06,
+      "loss": 7.9547,
+      "step": 266
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.010000000000001e-06,
+      "loss": 7.9642,
+      "step": 267
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 7.9138,
+      "step": 268
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.07e-06,
+      "loss": 7.8365,
+      "step": 269
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.1e-06,
+      "loss": 7.8481,
+      "step": 270
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.13e-06,
+      "loss": 7.798,
+      "step": 271
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.160000000000001e-06,
+      "loss": 7.7801,
+      "step": 272
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.190000000000001e-06,
+      "loss": 7.7473,
+      "step": 273
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 7.6878,
+      "step": 274
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.25e-06,
+      "loss": 7.8519,
+      "step": 275
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.28e-06,
+      "loss": 7.7494,
+      "step": 276
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.310000000000002e-06,
+      "loss": 7.679,
+      "step": 277
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.340000000000001e-06,
+      "loss": 7.7505,
+      "step": 278
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.370000000000001e-06,
+      "loss": 7.8027,
+      "step": 279
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 7.8078,
+      "step": 280
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.43e-06,
+      "loss": 7.6369,
+      "step": 281
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.459999999999999e-06,
+      "loss": 7.7605,
+      "step": 282
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.49e-06,
+      "loss": 7.6024,
+      "step": 283
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.52e-06,
+      "loss": 7.7511,
+      "step": 284
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.55e-06,
+      "loss": 7.6924,
+      "step": 285
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.58e-06,
+      "loss": 7.617,
+      "step": 286
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.609999999999999e-06,
+      "loss": 7.8157,
+      "step": 287
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.64e-06,
+      "loss": 7.6566,
+      "step": 288
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.67e-06,
+      "loss": 7.7395,
+      "step": 289
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.7e-06,
+      "loss": 7.7573,
+      "step": 290
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.73e-06,
+      "loss": 7.7634,
+      "step": 291
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.759999999999999e-06,
+      "loss": 7.5908,
+      "step": 292
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.79e-06,
+      "loss": 7.5535,
+      "step": 293
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.82e-06,
+      "loss": 7.528,
+      "step": 294
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.85e-06,
+      "loss": 7.4976,
+      "step": 295
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.88e-06,
+      "loss": 7.4942,
+      "step": 296
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.91e-06,
+      "loss": 7.473,
+      "step": 297
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.939999999999999e-06,
+      "loss": 7.3699,
+      "step": 298
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.97e-06,
+      "loss": 7.3102,
+      "step": 299
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9e-06,
+      "loss": 7.1699,
+      "step": 300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.03e-06,
+      "loss": 7.7308,
+      "step": 301
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.06e-06,
+      "loss": 7.7359,
+      "step": 302
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.09e-06,
+      "loss": 7.6904,
+      "step": 303
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.12e-06,
+      "loss": 7.7346,
+      "step": 304
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.15e-06,
+      "loss": 7.5395,
+      "step": 305
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.18e-06,
+      "loss": 7.599,
+      "step": 306
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.21e-06,
+      "loss": 7.6035,
+      "step": 307
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.24e-06,
+      "loss": 7.6335,
+      "step": 308
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.27e-06,
+      "loss": 7.5536,
+      "step": 309
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.3e-06,
+      "loss": 7.5669,
+      "step": 310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.33e-06,
+      "loss": 7.6518,
+      "step": 311
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.36e-06,
+      "loss": 7.6117,
+      "step": 312
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.39e-06,
+      "loss": 7.5847,
+      "step": 313
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.42e-06,
+      "loss": 7.545,
+      "step": 314
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.450000000000001e-06,
+      "loss": 7.4631,
+      "step": 315
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.48e-06,
+      "loss": 7.5538,
+      "step": 316
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.51e-06,
+      "loss": 7.5308,
+      "step": 317
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.54e-06,
+      "loss": 7.4779,
+      "step": 318
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.57e-06,
+      "loss": 7.419,
+      "step": 319
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 7.5887,
+      "step": 320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.630000000000001e-06,
+      "loss": 7.4839,
+      "step": 321
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.66e-06,
+      "loss": 7.4614,
+      "step": 322
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.69e-06,
+      "loss": 7.4525,
+      "step": 323
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.72e-06,
+      "loss": 7.4914,
+      "step": 324
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.75e-06,
+      "loss": 7.4453,
+      "step": 325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 7.4815,
+      "step": 326
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.810000000000001e-06,
+      "loss": 7.3857,
+      "step": 327
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.84e-06,
+      "loss": 7.4469,
+      "step": 328
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.87e-06,
+      "loss": 7.4547,
+      "step": 329
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.9e-06,
+      "loss": 7.4825,
+      "step": 330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.930000000000001e-06,
+      "loss": 7.4766,
+      "step": 331
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 7.2181,
+      "step": 332
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.990000000000001e-06,
+      "loss": 7.4452,
+      "step": 333
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.002e-05,
+      "loss": 7.4228,
+      "step": 334
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.005e-05,
+      "loss": 7.4751,
+      "step": 335
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.008e-05,
+      "loss": 7.3075,
+      "step": 336
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0110000000000001e-05,
+      "loss": 7.241,
+      "step": 337
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0140000000000001e-05,
+      "loss": 7.4351,
+      "step": 338
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0170000000000001e-05,
+      "loss": 7.4326,
+      "step": 339
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.02e-05,
+      "loss": 7.3597,
+      "step": 340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.023e-05,
+      "loss": 7.4116,
+      "step": 341
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0260000000000002e-05,
+      "loss": 7.3383,
+      "step": 342
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.0290000000000001e-05,
+      "loss": 7.3001,
+      "step": 343
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.032e-05,
+      "loss": 7.3052,
+      "step": 344
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.035e-05,
+      "loss": 7.1747,
+      "step": 345
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0379999999999999e-05,
+      "loss": 7.2559,
+      "step": 346
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.041e-05,
+      "loss": 7.1649,
+      "step": 347
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.044e-05,
+      "loss": 7.1853,
+      "step": 348
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.047e-05,
+      "loss": 7.0696,
+      "step": 349
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.05e-05,
+      "loss": 6.9193,
+      "step": 350
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0529999999999999e-05,
+      "loss": 7.4138,
+      "step": 351
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.0559999999999999e-05,
+      "loss": 7.4342,
+      "step": 352
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.059e-05,
+      "loss": 7.3867,
+      "step": 353
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.062e-05,
+      "loss": 7.3788,
+      "step": 354
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.065e-05,
+      "loss": 7.4602,
+      "step": 355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.068e-05,
+      "loss": 7.283,
+      "step": 356
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.0709999999999999e-05,
+      "loss": 7.4469,
+      "step": 357
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.074e-05,
+      "loss": 7.2593,
+      "step": 358
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.077e-05,
+      "loss": 7.3028,
+      "step": 359
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.08e-05,
+      "loss": 7.1992,
+      "step": 360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.083e-05,
+      "loss": 7.24,
+      "step": 361
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.086e-05,
+      "loss": 7.3108,
+      "step": 362
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.089e-05,
+      "loss": 7.265,
+      "step": 363
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.092e-05,
+      "loss": 7.2879,
+      "step": 364
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.095e-05,
+      "loss": 7.2659,
+      "step": 365
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.098e-05,
+      "loss": 7.1942,
+      "step": 366
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.101e-05,
+      "loss": 7.2203,
+      "step": 367
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.104e-05,
+      "loss": 7.1853,
+      "step": 368
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.107e-05,
+      "loss": 7.1243,
+      "step": 369
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.11e-05,
+      "loss": 7.3274,
+      "step": 370
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.113e-05,
+      "loss": 7.2143,
+      "step": 371
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.116e-05,
+      "loss": 7.3603,
+      "step": 372
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.119e-05,
+      "loss": 7.2227,
+      "step": 373
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1220000000000001e-05,
+      "loss": 7.2911,
+      "step": 374
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.125e-05,
+      "loss": 7.1161,
+      "step": 375
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.128e-05,
+      "loss": 7.2371,
+      "step": 376
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.131e-05,
+      "loss": 7.1319,
+      "step": 377
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.134e-05,
+      "loss": 7.162,
+      "step": 378
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.137e-05,
+      "loss": 7.1768,
+      "step": 379
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 7.2221,
+      "step": 380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.143e-05,
+      "loss": 7.2262,
+      "step": 381
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.146e-05,
+      "loss": 7.1323,
+      "step": 382
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.149e-05,
+      "loss": 7.2706,
+      "step": 383
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.152e-05,
+      "loss": 7.2399,
+      "step": 384
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1550000000000001e-05,
+      "loss": 7.123,
+      "step": 385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 7.1734,
+      "step": 386
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.161e-05,
+      "loss": 7.1634,
+      "step": 387
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.164e-05,
+      "loss": 7.1187,
+      "step": 388
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.167e-05,
+      "loss": 7.1166,
+      "step": 389
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1700000000000001e-05,
+      "loss": 7.3409,
+      "step": 390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1730000000000001e-05,
+      "loss": 7.1993,
+      "step": 391
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1760000000000001e-05,
+      "loss": 7.1809,
+      "step": 392
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.179e-05,
+      "loss": 7.0819,
+      "step": 393
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.182e-05,
+      "loss": 7.2385,
+      "step": 394
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.185e-05,
+      "loss": 7.0688,
+      "step": 395
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1880000000000001e-05,
+      "loss": 6.8675,
+      "step": 396
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1910000000000001e-05,
+      "loss": 6.7992,
+      "step": 397
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1940000000000001e-05,
+      "loss": 6.9666,
+      "step": 398
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.197e-05,
+      "loss": 6.8934,
+      "step": 399
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2e-05,
+      "loss": 6.6587,
+      "step": 400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2030000000000002e-05,
+      "loss": 7.3765,
+      "step": 401
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2060000000000001e-05,
+      "loss": 7.3654,
+      "step": 402
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2090000000000001e-05,
+      "loss": 7.3227,
+      "step": 403
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2120000000000001e-05,
+      "loss": 7.28,
+      "step": 404
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.215e-05,
+      "loss": 7.3649,
+      "step": 405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2180000000000002e-05,
+      "loss": 7.2365,
+      "step": 406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.221e-05,
+      "loss": 7.1794,
+      "step": 407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.224e-05,
+      "loss": 7.1244,
+      "step": 408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.227e-05,
+      "loss": 7.2877,
+      "step": 409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2299999999999999e-05,
+      "loss": 7.1793,
+      "step": 410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2329999999999999e-05,
+      "loss": 7.1637,
+      "step": 411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.236e-05,
+      "loss": 7.1168,
+      "step": 412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.239e-05,
+      "loss": 7.0768,
+      "step": 413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.242e-05,
+      "loss": 7.0132,
+      "step": 414
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.245e-05,
+      "loss": 7.0574,
+      "step": 415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2479999999999999e-05,
+      "loss": 7.1586,
+      "step": 416
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.251e-05,
+      "loss": 7.0684,
+      "step": 417
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.254e-05,
+      "loss": 7.1115,
+      "step": 418
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.257e-05,
+      "loss": 7.1108,
+      "step": 419
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.26e-05,
+      "loss": 7.0989,
+      "step": 420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.263e-05,
+      "loss": 7.1333,
+      "step": 421
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.2659999999999999e-05,
+      "loss": 7.1599,
+      "step": 422
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.269e-05,
+      "loss": 7.1185,
+      "step": 423
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.272e-05,
+      "loss": 7.0918,
+      "step": 424
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.275e-05,
+      "loss": 7.1031,
+      "step": 425
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.278e-05,
+      "loss": 7.2346,
+      "step": 426
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.281e-05,
+      "loss": 7.0122,
+      "step": 427
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.284e-05,
+      "loss": 7.2465,
+      "step": 428
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.287e-05,
+      "loss": 7.0599,
+      "step": 429
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.29e-05,
+      "loss": 7.0457,
+      "step": 430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.293e-05,
+      "loss": 7.2033,
+      "step": 431
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.296e-05,
+      "loss": 7.133,
+      "step": 432
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2990000000000001e-05,
+      "loss": 6.8479,
+      "step": 433
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.302e-05,
+      "loss": 7.0118,
+      "step": 434
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.305e-05,
+      "loss": 6.9634,
+      "step": 435
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.308e-05,
+      "loss": 7.1219,
+      "step": 436
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.311e-05,
+      "loss": 7.0447,
+      "step": 437
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.314e-05,
+      "loss": 7.0948,
+      "step": 438
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.3170000000000001e-05,
+      "loss": 6.9062,
+      "step": 439
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.32e-05,
+      "loss": 7.0378,
+      "step": 440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.323e-05,
+      "loss": 7.0822,
+      "step": 441
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.326e-05,
+      "loss": 7.0011,
+      "step": 442
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.329e-05,
+      "loss": 6.9484,
+      "step": 443
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3320000000000001e-05,
+      "loss": 7.1245,
+      "step": 444
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3350000000000001e-05,
+      "loss": 6.8149,
+      "step": 445
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.338e-05,
+      "loss": 6.8453,
+      "step": 446
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.341e-05,
+      "loss": 6.8721,
+      "step": 447
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.344e-05,
+      "loss": 6.8137,
+      "step": 448
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3470000000000001e-05,
+      "loss": 6.6156,
+      "step": 449
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 6.5602,
+      "step": 450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3530000000000001e-05,
+      "loss": 7.1587,
+      "step": 451
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.356e-05,
+      "loss": 7.2656,
+      "step": 452
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.359e-05,
+      "loss": 7.2567,
+      "step": 453
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.362e-05,
+      "loss": 7.0606,
+      "step": 454
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3650000000000001e-05,
+      "loss": 7.1304,
+      "step": 455
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3680000000000001e-05,
+      "loss": 7.1503,
+      "step": 456
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.3710000000000001e-05,
+      "loss": 6.9522,
+      "step": 457
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.374e-05,
+      "loss": 7.0655,
+      "step": 458
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.377e-05,
+      "loss": 7.0146,
+      "step": 459
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 7.1067,
+      "step": 460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3830000000000001e-05,
+      "loss": 6.9196,
+      "step": 461
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3860000000000001e-05,
+      "loss": 7.0795,
+      "step": 462
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.389e-05,
+      "loss": 6.9209,
+      "step": 463
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.392e-05,
+      "loss": 6.9123,
+      "step": 464
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.395e-05,
+      "loss": 6.9922,
+      "step": 465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.3980000000000002e-05,
+      "loss": 6.948,
+      "step": 466
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4010000000000001e-05,
+      "loss": 6.9436,
+      "step": 467
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4040000000000001e-05,
+      "loss": 7.0157,
+      "step": 468
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4069999999999999e-05,
+      "loss": 6.9262,
+      "step": 469
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4099999999999999e-05,
+      "loss": 6.9941,
+      "step": 470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.413e-05,
+      "loss": 6.9391,
+      "step": 471
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.416e-05,
+      "loss": 6.9376,
+      "step": 472
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.419e-05,
+      "loss": 6.9565,
+      "step": 473
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.422e-05,
+      "loss": 6.8788,
+      "step": 474
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4249999999999999e-05,
+      "loss": 7.0099,
+      "step": 475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.428e-05,
+      "loss": 7.004,
+      "step": 476
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.431e-05,
+      "loss": 7.0881,
+      "step": 477
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.434e-05,
+      "loss": 6.8615,
+      "step": 478
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.437e-05,
+      "loss": 6.8601,
+      "step": 479
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.44e-05,
+      "loss": 6.8863,
+      "step": 480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4429999999999999e-05,
+      "loss": 6.8749,
+      "step": 481
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.446e-05,
+      "loss": 6.8717,
+      "step": 482
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.449e-05,
+      "loss": 6.9278,
+      "step": 483
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.452e-05,
+      "loss": 6.9361,
+      "step": 484
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.455e-05,
+      "loss": 6.9849,
+      "step": 485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.458e-05,
+      "loss": 6.8833,
+      "step": 486
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.461e-05,
+      "loss": 6.948,
+      "step": 487
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.464e-05,
+      "loss": 6.8478,
+      "step": 488
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.467e-05,
+      "loss": 7.02,
+      "step": 489
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.47e-05,
+      "loss": 7.0518,
+      "step": 490
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.473e-05,
+      "loss": 6.8551,
+      "step": 491
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.4760000000000001e-05,
+      "loss": 6.8822,
+      "step": 492
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.479e-05,
+      "loss": 6.897,
+      "step": 493
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.482e-05,
+      "loss": 6.8555,
+      "step": 494
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.485e-05,
+      "loss": 6.8925,
+      "step": 495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.488e-05,
+      "loss": 6.703,
+      "step": 496
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.491e-05,
+      "loss": 6.432,
+      "step": 497
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.4940000000000001e-05,
+      "loss": 6.9492,
+      "step": 498
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.497e-05,
+      "loss": 6.7646,
+      "step": 499
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5e-05,
+      "loss": 6.3888,
+      "step": 500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.503e-05,
+      "loss": 7.5657,
+      "step": 501
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.506e-05,
+      "loss": 7.2214,
+      "step": 502
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.5090000000000001e-05,
+      "loss": 7.2656,
+      "step": 503
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5120000000000001e-05,
+      "loss": 7.2593,
+      "step": 504
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.515e-05,
+      "loss": 7.1604,
+      "step": 505
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.518e-05,
+      "loss": 7.0984,
+      "step": 506
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.521e-05,
+      "loss": 7.044,
+      "step": 507
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.524e-05,
+      "loss": 6.9858,
+      "step": 508
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.527e-05,
+      "loss": 6.9989,
+      "step": 509
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.53e-05,
+      "loss": 6.9709,
+      "step": 510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.533e-05,
+      "loss": 6.9102,
+      "step": 511
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.5360000000000002e-05,
+      "loss": 6.9431,
+      "step": 512
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.539e-05,
+      "loss": 7.0596,
+      "step": 513
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.542e-05,
+      "loss": 6.9836,
+      "step": 514
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.545e-05,
+      "loss": 6.9814,
+      "step": 515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.548e-05,
+      "loss": 6.7806,
+      "step": 516
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5510000000000002e-05,
+      "loss": 6.9839,
+      "step": 517
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.554e-05,
+      "loss": 6.8484,
+      "step": 518
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.5570000000000002e-05,
+      "loss": 7.0011,
+      "step": 519
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.56e-05,
+      "loss": 6.9772,
+      "step": 520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.563e-05,
+      "loss": 6.8379,
+      "step": 521
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5660000000000003e-05,
+      "loss": 6.8115,
+      "step": 522
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.569e-05,
+      "loss": 6.7369,
+      "step": 523
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5720000000000002e-05,
+      "loss": 6.805,
+      "step": 524
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.575e-05,
+      "loss": 6.9387,
+      "step": 525
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.578e-05,
+      "loss": 6.7562,
+      "step": 526
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5810000000000003e-05,
+      "loss": 6.8463,
+      "step": 527
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.584e-05,
+      "loss": 6.8984,
+      "step": 528
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.5870000000000002e-05,
+      "loss": 6.9348,
+      "step": 529
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.59e-05,
+      "loss": 6.9595,
+      "step": 530
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.593e-05,
+      "loss": 6.833,
+      "step": 531
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.596e-05,
+      "loss": 6.7737,
+      "step": 532
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.599e-05,
+      "loss": 6.768,
+      "step": 533
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6020000000000002e-05,
+      "loss": 6.7689,
+      "step": 534
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.605e-05,
+      "loss": 6.7882,
+      "step": 535
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6080000000000002e-05,
+      "loss": 6.932,
+      "step": 536
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.611e-05,
+      "loss": 6.876,
+      "step": 537
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.614e-05,
+      "loss": 6.9184,
+      "step": 538
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6170000000000003e-05,
+      "loss": 6.901,
+      "step": 539
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.62e-05,
+      "loss": 6.8328,
+      "step": 540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6230000000000002e-05,
+      "loss": 6.9365,
+      "step": 541
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.626e-05,
+      "loss": 6.7989,
+      "step": 542
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.629e-05,
+      "loss": 6.6952,
+      "step": 543
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6320000000000003e-05,
+      "loss": 6.6093,
+      "step": 544
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.635e-05,
+      "loss": 6.658,
+      "step": 545
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6380000000000002e-05,
+      "loss": 6.6633,
+      "step": 546
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.641e-05,
+      "loss": 6.8938,
+      "step": 547
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6440000000000002e-05,
+      "loss": 6.6474,
+      "step": 548
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6470000000000003e-05,
+      "loss": 6.1851,
+      "step": 549
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.65e-05,
+      "loss": 6.5158,
+      "step": 550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6530000000000003e-05,
+      "loss": 7.3829,
+      "step": 551
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.656e-05,
+      "loss": 7.51,
+      "step": 552
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6590000000000002e-05,
+      "loss": 7.1924,
+      "step": 553
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6620000000000004e-05,
+      "loss": 7.0141,
+      "step": 554
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.665e-05,
+      "loss": 6.9854,
+      "step": 555
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6680000000000003e-05,
+      "loss": 6.9506,
+      "step": 556
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.671e-05,
+      "loss": 6.8861,
+      "step": 557
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6740000000000002e-05,
+      "loss": 6.9081,
+      "step": 558
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.677e-05,
+      "loss": 6.8343,
+      "step": 559
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 6.7617,
+      "step": 560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6830000000000003e-05,
+      "loss": 6.9083,
+      "step": 561
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.686e-05,
+      "loss": 6.7417,
+      "step": 562
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.689e-05,
+      "loss": 6.7845,
+      "step": 563
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.6919999999999997e-05,
+      "loss": 6.8019,
+      "step": 564
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.695e-05,
+      "loss": 6.9395,
+      "step": 565
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.698e-05,
+      "loss": 6.6751,
+      "step": 566
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7009999999999998e-05,
+      "loss": 6.8468,
+      "step": 567
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.704e-05,
+      "loss": 6.647,
+      "step": 568
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7069999999999998e-05,
+      "loss": 6.6706,
+      "step": 569
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.71e-05,
+      "loss": 6.8234,
+      "step": 570
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.713e-05,
+      "loss": 6.7205,
+      "step": 571
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.716e-05,
+      "loss": 6.8656,
+      "step": 572
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.719e-05,
+      "loss": 6.6907,
+      "step": 573
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7219999999999998e-05,
+      "loss": 6.5996,
+      "step": 574
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.725e-05,
+      "loss": 6.8011,
+      "step": 575
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.728e-05,
+      "loss": 6.6263,
+      "step": 576
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.731e-05,
+      "loss": 6.7285,
+      "step": 577
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.734e-05,
+      "loss": 6.7957,
+      "step": 578
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7369999999999998e-05,
+      "loss": 6.6663,
+      "step": 579
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.74e-05,
+      "loss": 6.8978,
+      "step": 580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.743e-05,
+      "loss": 6.866,
+      "step": 581
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.746e-05,
+      "loss": 6.7711,
+      "step": 582
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.749e-05,
+      "loss": 6.6718,
+      "step": 583
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7519999999999998e-05,
+      "loss": 6.7416,
+      "step": 584
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.755e-05,
+      "loss": 6.8254,
+      "step": 585
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.758e-05,
+      "loss": 6.8156,
+      "step": 586
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.761e-05,
+      "loss": 6.7815,
+      "step": 587
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.764e-05,
+      "loss": 6.7313,
+      "step": 588
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.767e-05,
+      "loss": 6.6486,
+      "step": 589
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.77e-05,
+      "loss": 6.6887,
+      "step": 590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7729999999999998e-05,
+      "loss": 6.7206,
+      "step": 591
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.776e-05,
+      "loss": 6.7452,
+      "step": 592
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.779e-05,
+      "loss": 6.8119,
+      "step": 593
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.782e-05,
+      "loss": 6.5288,
+      "step": 594
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.785e-05,
+      "loss": 6.6024,
+      "step": 595
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7879999999999998e-05,
+      "loss": 6.4717,
+      "step": 596
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.791e-05,
+      "loss": 6.4726,
+      "step": 597
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.794e-05,
+      "loss": 6.4531,
+      "step": 598
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.797e-05,
+      "loss": 6.3246,
+      "step": 599
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8e-05,
+      "loss": 6.0595,
+      "step": 600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.803e-05,
+      "loss": 7.2845,
+      "step": 601
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.806e-05,
+      "loss": 7.0945,
+      "step": 602
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.809e-05,
+      "loss": 7.1077,
+      "step": 603
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.812e-05,
+      "loss": 6.9909,
+      "step": 604
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.815e-05,
+      "loss": 6.9545,
+      "step": 605
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.818e-05,
+      "loss": 6.6772,
+      "step": 606
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.821e-05,
+      "loss": 6.8011,
+      "step": 607
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.824e-05,
+      "loss": 6.8573,
+      "step": 608
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.827e-05,
+      "loss": 6.7732,
+      "step": 609
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.83e-05,
+      "loss": 6.73,
+      "step": 610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.833e-05,
+      "loss": 6.7131,
+      "step": 611
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.836e-05,
+      "loss": 6.7685,
+      "step": 612
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8390000000000002e-05,
+      "loss": 6.6396,
+      "step": 613
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.842e-05,
+      "loss": 6.7297,
+      "step": 614
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.845e-05,
+      "loss": 6.9083,
+      "step": 615
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.848e-05,
+      "loss": 6.6015,
+      "step": 616
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.851e-05,
+      "loss": 6.7134,
+      "step": 617
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.854e-05,
+      "loss": 6.5543,
+      "step": 618
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.857e-05,
+      "loss": 6.6502,
+      "step": 619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.86e-05,
+      "loss": 6.7256,
+      "step": 620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.863e-05,
+      "loss": 6.7804,
+      "step": 621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.866e-05,
+      "loss": 6.6864,
+      "step": 622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.869e-05,
+      "loss": 6.7672,
+      "step": 623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.872e-05,
+      "loss": 6.6197,
+      "step": 624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 6.8546,
+      "step": 625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.878e-05,
+      "loss": 6.6596,
+      "step": 626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.881e-05,
+      "loss": 6.8104,
+      "step": 627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.884e-05,
+      "loss": 6.7612,
+      "step": 628
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.887e-05,
+      "loss": 6.714,
+      "step": 629
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8900000000000002e-05,
+      "loss": 6.919,
+      "step": 630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.893e-05,
+      "loss": 6.6736,
+      "step": 631
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.896e-05,
+      "loss": 6.5336,
+      "step": 632
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.899e-05,
+      "loss": 6.7544,
+      "step": 633
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.902e-05,
+      "loss": 6.6841,
+      "step": 634
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9050000000000002e-05,
+      "loss": 6.7949,
+      "step": 635
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.908e-05,
+      "loss": 6.7632,
+      "step": 636
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.911e-05,
+      "loss": 6.7453,
+      "step": 637
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.914e-05,
+      "loss": 6.5481,
+      "step": 638
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.917e-05,
+      "loss": 6.8688,
+      "step": 639
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 6.5927,
+      "step": 640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.923e-05,
+      "loss": 6.6564,
+      "step": 641
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9260000000000002e-05,
+      "loss": 6.9019,
+      "step": 642
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.929e-05,
+      "loss": 6.7165,
+      "step": 643
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932e-05,
+      "loss": 6.4323,
+      "step": 644
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.935e-05,
+      "loss": 6.5196,
+      "step": 645
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.938e-05,
+      "loss": 6.7127,
+      "step": 646
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9410000000000002e-05,
+      "loss": 6.3939,
+      "step": 647
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.944e-05,
+      "loss": 6.2812,
+      "step": 648
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.947e-05,
+      "loss": 6.5294,
+      "step": 649
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.95e-05,
+      "loss": 6.3023,
+      "step": 650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.953e-05,
+      "loss": 7.1731,
+      "step": 651
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9560000000000002e-05,
+      "loss": 7.1681,
+      "step": 652
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.959e-05,
+      "loss": 7.0888,
+      "step": 653
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9620000000000002e-05,
+      "loss": 6.9006,
+      "step": 654
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.965e-05,
+      "loss": 6.8562,
+      "step": 655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.968e-05,
+      "loss": 6.8262,
+      "step": 656
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9710000000000003e-05,
+      "loss": 6.6168,
+      "step": 657
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.974e-05,
+      "loss": 6.9157,
+      "step": 658
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9770000000000002e-05,
+      "loss": 6.6025,
+      "step": 659
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.98e-05,
+      "loss": 6.6174,
+      "step": 660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.983e-05,
+      "loss": 6.6463,
+      "step": 661
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9860000000000003e-05,
+      "loss": 6.7119,
+      "step": 662
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.989e-05,
+      "loss": 6.7824,
+      "step": 663
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9920000000000002e-05,
+      "loss": 6.5796,
+      "step": 664
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.995e-05,
+      "loss": 6.6199,
+      "step": 665
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 6.7515,
+      "step": 666
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.0010000000000003e-05,
+      "loss": 6.6792,
+      "step": 667
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.004e-05,
+      "loss": 6.6072,
+      "step": 668
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.0070000000000003e-05,
+      "loss": 6.7619,
+      "step": 669
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.01e-05,
+      "loss": 6.645,
+      "step": 670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.0130000000000002e-05,
+      "loss": 6.5365,
+      "step": 671
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.016e-05,
+      "loss": 6.6618,
+      "step": 672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.019e-05,
+      "loss": 6.6541,
+      "step": 673
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.0220000000000003e-05,
+      "loss": 6.5649,
+      "step": 674
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.025e-05,
+      "loss": 6.6856,
+      "step": 675
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.0280000000000002e-05,
+      "loss": 6.7348,
+      "step": 676
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.031e-05,
+      "loss": 6.6344,
+      "step": 677
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.0340000000000002e-05,
+      "loss": 6.4282,
+      "step": 678
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.0370000000000003e-05,
+      "loss": 6.6915,
+      "step": 679
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.04e-05,
+      "loss": 6.5709,
+      "step": 680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.0430000000000003e-05,
+      "loss": 6.807,
+      "step": 681
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.046e-05,
+      "loss": 6.6091,
+      "step": 682
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.0490000000000002e-05,
+      "loss": 6.6238,
+      "step": 683
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.0520000000000003e-05,
+      "loss": 6.7503,
+      "step": 684
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.055e-05,
+      "loss": 6.7269,
+      "step": 685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.0580000000000003e-05,
+      "loss": 6.5173,
+      "step": 686
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.061e-05,
+      "loss": 6.5405,
+      "step": 687
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.064e-05,
+      "loss": 6.7792,
+      "step": 688
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.067e-05,
+      "loss": 6.6311,
+      "step": 689
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.07e-05,
+      "loss": 6.7035,
+      "step": 690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.073e-05,
+      "loss": 6.5084,
+      "step": 691
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0759999999999998e-05,
+      "loss": 6.5663,
+      "step": 692
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.079e-05,
+      "loss": 6.3822,
+      "step": 693
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.082e-05,
+      "loss": 6.5583,
+      "step": 694
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.085e-05,
+      "loss": 6.534,
+      "step": 695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.088e-05,
+      "loss": 6.5662,
+      "step": 696
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.0909999999999998e-05,
+      "loss": 6.4185,
+      "step": 697
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.094e-05,
+      "loss": 6.205,
+      "step": 698
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.097e-05,
+      "loss": 6.3609,
+      "step": 699
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.1e-05,
+      "loss": 5.9583,
+      "step": 700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.103e-05,
+      "loss": 6.8742,
+      "step": 701
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1059999999999998e-05,
+      "loss": 6.8557,
+      "step": 702
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.109e-05,
+      "loss": 6.7989,
+      "step": 703
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1119999999999998e-05,
+      "loss": 6.6631,
+      "step": 704
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.115e-05,
+      "loss": 6.7317,
+      "step": 705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.118e-05,
+      "loss": 6.8051,
+      "step": 706
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.121e-05,
+      "loss": 6.7518,
+      "step": 707
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.124e-05,
+      "loss": 6.8761,
+      "step": 708
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1269999999999998e-05,
+      "loss": 6.6564,
+      "step": 709
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.13e-05,
+      "loss": 6.5897,
+      "step": 710
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.133e-05,
+      "loss": 6.6477,
+      "step": 711
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.136e-05,
+      "loss": 6.4958,
+      "step": 712
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.139e-05,
+      "loss": 6.5248,
+      "step": 713
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1419999999999998e-05,
+      "loss": 6.6637,
+      "step": 714
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.145e-05,
+      "loss": 6.7085,
+      "step": 715
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.148e-05,
+      "loss": 6.6819,
+      "step": 716
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.151e-05,
+      "loss": 6.9019,
+      "step": 717
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.154e-05,
+      "loss": 6.6394,
+      "step": 718
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.157e-05,
+      "loss": 6.439,
+      "step": 719
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.16e-05,
+      "loss": 6.5588,
+      "step": 720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.163e-05,
+      "loss": 6.5887,
+      "step": 721
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.166e-05,
+      "loss": 6.5381,
+      "step": 722
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.169e-05,
+      "loss": 6.5145,
+      "step": 723
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.172e-05,
+      "loss": 6.6321,
+      "step": 724
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.175e-05,
+      "loss": 6.616,
+      "step": 725
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.178e-05,
+      "loss": 6.7522,
+      "step": 726
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.181e-05,
+      "loss": 6.6116,
+      "step": 727
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.184e-05,
+      "loss": 6.5015,
+      "step": 728
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.187e-05,
+      "loss": 6.5366,
+      "step": 729
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.19e-05,
+      "loss": 6.6674,
+      "step": 730
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.193e-05,
+      "loss": 6.8586,
+      "step": 731
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.196e-05,
+      "loss": 6.7016,
+      "step": 732
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.199e-05,
+      "loss": 6.59,
+      "step": 733
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.202e-05,
+      "loss": 6.7739,
+      "step": 734
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.205e-05,
+      "loss": 6.5316,
+      "step": 735
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.208e-05,
+      "loss": 6.6708,
+      "step": 736
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.211e-05,
+      "loss": 6.6138,
+      "step": 737
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.214e-05,
+      "loss": 6.6088,
+      "step": 738
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.217e-05,
+      "loss": 6.4944,
+      "step": 739
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.22e-05,
+      "loss": 6.619,
+      "step": 740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.223e-05,
+      "loss": 6.6547,
+      "step": 741
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.226e-05,
+      "loss": 6.5947,
+      "step": 742
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.2290000000000002e-05,
+      "loss": 6.46,
+      "step": 743
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.232e-05,
+      "loss": 6.5316,
+      "step": 744
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.235e-05,
+      "loss": 6.4311,
+      "step": 745
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.238e-05,
+      "loss": 6.7098,
+      "step": 746
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.241e-05,
+      "loss": 6.3111,
+      "step": 747
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2440000000000002e-05,
+      "loss": 6.3249,
+      "step": 748
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.247e-05,
+      "loss": 6.3004,
+      "step": 749
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.25e-05,
+      "loss": 6.1039,
+      "step": 750
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.253e-05,
+      "loss": 6.8155,
+      "step": 751
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.256e-05,
+      "loss": 6.9188,
+      "step": 752
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.2590000000000002e-05,
+      "loss": 6.8613,
+      "step": 753
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.262e-05,
+      "loss": 6.75,
+      "step": 754
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.265e-05,
+      "loss": 6.6995,
+      "step": 755
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.268e-05,
+      "loss": 6.8017,
+      "step": 756
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.271e-05,
+      "loss": 6.5982,
+      "step": 757
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.274e-05,
+      "loss": 6.5564,
+      "step": 758
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.277e-05,
+      "loss": 6.6461,
+      "step": 759
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 6.5323,
+      "step": 760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.283e-05,
+      "loss": 6.6537,
+      "step": 761
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.286e-05,
+      "loss": 6.5489,
+      "step": 762
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.289e-05,
+      "loss": 6.4215,
+      "step": 763
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.292e-05,
+      "loss": 6.4027,
+      "step": 764
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2950000000000002e-05,
+      "loss": 6.5636,
+      "step": 765
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.298e-05,
+      "loss": 6.4929,
+      "step": 766
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.301e-05,
+      "loss": 6.527,
+      "step": 767
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.304e-05,
+      "loss": 6.5393,
+      "step": 768
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.307e-05,
+      "loss": 6.6195,
+      "step": 769
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3100000000000002e-05,
+      "loss": 6.4657,
+      "step": 770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.313e-05,
+      "loss": 6.6625,
+      "step": 771
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3160000000000002e-05,
+      "loss": 6.7045,
+      "step": 772
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.319e-05,
+      "loss": 6.6015,
+      "step": 773
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.322e-05,
+      "loss": 6.4125,
+      "step": 774
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3250000000000003e-05,
+      "loss": 6.3341,
+      "step": 775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.328e-05,
+      "loss": 6.4739,
+      "step": 776
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3310000000000002e-05,
+      "loss": 6.3741,
+      "step": 777
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.334e-05,
+      "loss": 6.631,
+      "step": 778
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.337e-05,
+      "loss": 6.5112,
+      "step": 779
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 6.5395,
+      "step": 780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.343e-05,
+      "loss": 6.5417,
+      "step": 781
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3460000000000002e-05,
+      "loss": 6.4377,
+      "step": 782
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.349e-05,
+      "loss": 6.5858,
+      "step": 783
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3520000000000002e-05,
+      "loss": 6.4455,
+      "step": 784
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3550000000000003e-05,
+      "loss": 6.5975,
+      "step": 785
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.358e-05,
+      "loss": 6.54,
+      "step": 786
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3610000000000003e-05,
+      "loss": 6.5395,
+      "step": 787
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.364e-05,
+      "loss": 6.5964,
+      "step": 788
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3670000000000002e-05,
+      "loss": 6.497,
+      "step": 789
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.37e-05,
+      "loss": 6.451,
+      "step": 790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.373e-05,
+      "loss": 6.406,
+      "step": 791
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3760000000000003e-05,
+      "loss": 6.3319,
+      "step": 792
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.379e-05,
+      "loss": 6.4752,
+      "step": 793
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3820000000000002e-05,
+      "loss": 6.3672,
+      "step": 794
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.385e-05,
+      "loss": 6.5543,
+      "step": 795
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3880000000000002e-05,
+      "loss": 6.428,
+      "step": 796
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3910000000000003e-05,
+      "loss": 6.4266,
+      "step": 797
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.394e-05,
+      "loss": 6.3823,
+      "step": 798
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3970000000000003e-05,
+      "loss": 6.6254,
+      "step": 799
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4e-05,
+      "loss": 5.9803,
+      "step": 800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4030000000000002e-05,
+      "loss": 6.7023,
+      "step": 801
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4060000000000003e-05,
+      "loss": 6.6912,
+      "step": 802
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.409e-05,
+      "loss": 6.7892,
+      "step": 803
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4120000000000003e-05,
+      "loss": 6.7318,
+      "step": 804
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.415e-05,
+      "loss": 6.3851,
+      "step": 805
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4180000000000002e-05,
+      "loss": 6.6873,
+      "step": 806
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4210000000000004e-05,
+      "loss": 6.5969,
+      "step": 807
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4240000000000002e-05,
+      "loss": 6.4192,
+      "step": 808
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4270000000000003e-05,
+      "loss": 6.538,
+      "step": 809
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.43e-05,
+      "loss": 6.495,
+      "step": 810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4330000000000003e-05,
+      "loss": 6.6086,
+      "step": 811
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4360000000000004e-05,
+      "loss": 6.6438,
+      "step": 812
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.439e-05,
+      "loss": 6.4558,
+      "step": 813
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.442e-05,
+      "loss": 6.4896,
+      "step": 814
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4449999999999998e-05,
+      "loss": 6.5433,
+      "step": 815
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.448e-05,
+      "loss": 6.436,
+      "step": 816
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.4509999999999997e-05,
+      "loss": 6.6424,
+      "step": 817
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.454e-05,
+      "loss": 6.5036,
+      "step": 818
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.457e-05,
+      "loss": 6.4059,
+      "step": 819
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.4599999999999998e-05,
+      "loss": 6.561,
+      "step": 820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.463e-05,
+      "loss": 6.5829,
+      "step": 821
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.4659999999999998e-05,
+      "loss": 6.4619,
+      "step": 822
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.469e-05,
+      "loss": 6.5829,
+      "step": 823
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.472e-05,
+      "loss": 6.4608,
+      "step": 824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.475e-05,
+      "loss": 6.7263,
+      "step": 825
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.478e-05,
+      "loss": 6.6508,
+      "step": 826
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4809999999999998e-05,
+      "loss": 6.4835,
+      "step": 827
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.484e-05,
+      "loss": 6.3986,
+      "step": 828
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.487e-05,
+      "loss": 6.4318,
+      "step": 829
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.49e-05,
+      "loss": 6.5081,
+      "step": 830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.493e-05,
+      "loss": 6.4048,
+      "step": 831
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4959999999999998e-05,
+      "loss": 6.4423,
+      "step": 832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.499e-05,
+      "loss": 6.4915,
+      "step": 833
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.502e-05,
+      "loss": 6.4619,
+      "step": 834
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.505e-05,
+      "loss": 6.4911,
+      "step": 835
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.508e-05,
+      "loss": 6.5725,
+      "step": 836
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.511e-05,
+      "loss": 6.4884,
+      "step": 837
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.514e-05,
+      "loss": 6.6047,
+      "step": 838
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.517e-05,
+      "loss": 6.3531,
+      "step": 839
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.52e-05,
+      "loss": 6.366,
+      "step": 840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.523e-05,
+      "loss": 6.6571,
+      "step": 841
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.526e-05,
+      "loss": 6.395,
+      "step": 842
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.529e-05,
+      "loss": 6.6624,
+      "step": 843
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5319999999999998e-05,
+      "loss": 6.4617,
+      "step": 844
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.535e-05,
+      "loss": 6.4631,
+      "step": 845
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.538e-05,
+      "loss": 6.3443,
+      "step": 846
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.541e-05,
+      "loss": 6.1924,
+      "step": 847
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.544e-05,
+      "loss": 6.4759,
+      "step": 848
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.547e-05,
+      "loss": 6.2675,
+      "step": 849
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.55e-05,
+      "loss": 5.8545,
+      "step": 850
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.553e-05,
+      "loss": 6.8176,
+      "step": 851
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.556e-05,
+      "loss": 6.8206,
+      "step": 852
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.559e-05,
+      "loss": 6.956,
+      "step": 853
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.562e-05,
+      "loss": 6.8262,
+      "step": 854
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.565e-05,
+      "loss": 6.6278,
+      "step": 855
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.568e-05,
+      "loss": 6.7081,
+      "step": 856
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.571e-05,
+      "loss": 6.702,
+      "step": 857
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.574e-05,
+      "loss": 6.6024,
+      "step": 858
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.577e-05,
+      "loss": 6.3996,
+      "step": 859
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.58e-05,
+      "loss": 6.5073,
+      "step": 860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5830000000000002e-05,
+      "loss": 6.692,
+      "step": 861
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.586e-05,
+      "loss": 6.5992,
+      "step": 862
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.589e-05,
+      "loss": 6.5074,
+      "step": 863
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.592e-05,
+      "loss": 6.6726,
+      "step": 864
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.595e-05,
+      "loss": 6.5746,
+      "step": 865
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5980000000000002e-05,
+      "loss": 6.4291,
+      "step": 866
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.601e-05,
+      "loss": 6.5884,
+      "step": 867
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.604e-05,
+      "loss": 6.3909,
+      "step": 868
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.607e-05,
+      "loss": 6.5409,
+      "step": 869
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.61e-05,
+      "loss": 6.4559,
+      "step": 870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6130000000000002e-05,
+      "loss": 6.4033,
+      "step": 871
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.616e-05,
+      "loss": 6.4989,
+      "step": 872
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.619e-05,
+      "loss": 6.4252,
+      "step": 873
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.622e-05,
+      "loss": 6.5631,
+      "step": 874
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.625e-05,
+      "loss": 6.6596,
+      "step": 875
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.628e-05,
+      "loss": 6.3188,
+      "step": 876
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.631e-05,
+      "loss": 6.3207,
+      "step": 877
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6340000000000002e-05,
+      "loss": 6.4421,
+      "step": 878
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.637e-05,
+      "loss": 6.4229,
+      "step": 879
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.64e-05,
+      "loss": 6.631,
+      "step": 880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.643e-05,
+      "loss": 6.2131,
+      "step": 881
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.646e-05,
+      "loss": 6.4829,
+      "step": 882
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6490000000000002e-05,
+      "loss": 6.3101,
+      "step": 883
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.652e-05,
+      "loss": 6.4484,
+      "step": 884
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.655e-05,
+      "loss": 6.454,
+      "step": 885
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.658e-05,
+      "loss": 6.453,
+      "step": 886
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.661e-05,
+      "loss": 6.179,
+      "step": 887
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6640000000000002e-05,
+      "loss": 6.3183,
+      "step": 888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.667e-05,
+      "loss": 6.2793,
+      "step": 889
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6700000000000002e-05,
+      "loss": 6.1758,
+      "step": 890
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.673e-05,
+      "loss": 6.0577,
+      "step": 891
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.676e-05,
+      "loss": 5.5931,
+      "step": 892
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6790000000000003e-05,
+      "loss": 6.8214,
+      "step": 893
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.682e-05,
+      "loss": 6.9,
+      "step": 894
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6850000000000002e-05,
+      "loss": 6.4872,
+      "step": 895
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.688e-05,
+      "loss": 6.5326,
+      "step": 896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.691e-05,
+      "loss": 6.5399,
+      "step": 897
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.6940000000000003e-05,
+      "loss": 6.6006,
+      "step": 898
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.697e-05,
+      "loss": 6.4644,
+      "step": 899
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 6.5087,
+      "step": 900
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.703e-05,
+      "loss": 6.5824,
+      "step": 901
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7060000000000002e-05,
+      "loss": 6.5484,
+      "step": 902
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.709e-05,
+      "loss": 6.2505,
+      "step": 903
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.712e-05,
+      "loss": 6.4793,
+      "step": 904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.7150000000000003e-05,
+      "loss": 6.3513,
+      "step": 905
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.718e-05,
+      "loss": 6.3483,
+      "step": 906
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7210000000000002e-05,
+      "loss": 6.6038,
+      "step": 907
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.724e-05,
+      "loss": 6.6282,
+      "step": 908
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.727e-05,
+      "loss": 6.4172,
+      "step": 909
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7300000000000003e-05,
+      "loss": 6.2283,
+      "step": 910
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.733e-05,
+      "loss": 6.3461,
+      "step": 911
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7360000000000002e-05,
+      "loss": 6.3779,
+      "step": 912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.739e-05,
+      "loss": 6.4478,
+      "step": 913
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7420000000000002e-05,
+      "loss": 6.5228,
+      "step": 914
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.7450000000000003e-05,
+      "loss": 6.2169,
+      "step": 915
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.748e-05,
+      "loss": 6.3876,
+      "step": 916
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.7510000000000003e-05,
+      "loss": 6.3348,
+      "step": 917
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.754e-05,
+      "loss": 6.4445,
+      "step": 918
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.7570000000000002e-05,
+      "loss": 6.4775,
+      "step": 919
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 6.2965,
+      "step": 920
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.763e-05,
+      "loss": 6.2906,
+      "step": 921
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.7660000000000003e-05,
+      "loss": 6.5401,
+      "step": 922
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.769e-05,
+      "loss": 6.4425,
+      "step": 923
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7720000000000002e-05,
+      "loss": 6.4596,
+      "step": 924
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7750000000000004e-05,
+      "loss": 6.4175,
+      "step": 925
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.778e-05,
+      "loss": 6.379,
+      "step": 926
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7810000000000003e-05,
+      "loss": 6.4415,
+      "step": 927
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.784e-05,
+      "loss": 6.3839,
+      "step": 928
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7870000000000003e-05,
+      "loss": 6.4452,
+      "step": 929
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.79e-05,
+      "loss": 6.5531,
+      "step": 930
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7930000000000002e-05,
+      "loss": 6.3241,
+      "step": 931
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7960000000000003e-05,
+      "loss": 6.2351,
+      "step": 932
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.799e-05,
+      "loss": 6.3819,
+      "step": 933
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.8020000000000003e-05,
+      "loss": 6.4854,
+      "step": 934
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.805e-05,
+      "loss": 6.2732,
+      "step": 935
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.8080000000000002e-05,
+      "loss": 6.2256,
+      "step": 936
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.8110000000000004e-05,
+      "loss": 6.3417,
+      "step": 937
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.8139999999999998e-05,
+      "loss": 6.1134,
+      "step": 938
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.817e-05,
+      "loss": 5.9921,
+      "step": 939
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 6.2065,
+      "step": 940
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.823e-05,
+      "loss": 5.9176,
+      "step": 941
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.826e-05,
+      "loss": 6.2175,
+      "step": 942
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.829e-05,
+      "loss": 6.7487,
+      "step": 943
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.832e-05,
+      "loss": 6.5763,
+      "step": 944
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.8349999999999998e-05,
+      "loss": 6.828,
+      "step": 945
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.838e-05,
+      "loss": 6.5777,
+      "step": 946
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.841e-05,
+      "loss": 6.7417,
+      "step": 947
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.844e-05,
+      "loss": 6.6563,
+      "step": 948
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.847e-05,
+      "loss": 6.5255,
+      "step": 949
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 6.6186,
+      "step": 950
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.853e-05,
+      "loss": 6.2116,
+      "step": 951
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.856e-05,
+      "loss": 6.5128,
+      "step": 952
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.859e-05,
+      "loss": 6.4767,
+      "step": 953
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.862e-05,
+      "loss": 6.3433,
+      "step": 954
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.865e-05,
+      "loss": 6.6377,
+      "step": 955
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.868e-05,
+      "loss": 6.4633,
+      "step": 956
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.871e-05,
+      "loss": 6.438,
+      "step": 957
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.874e-05,
+      "loss": 6.4368,
+      "step": 958
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.877e-05,
+      "loss": 6.5752,
+      "step": 959
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.88e-05,
+      "loss": 6.344,
+      "step": 960
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.883e-05,
+      "loss": 6.3879,
+      "step": 961
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.8859999999999998e-05,
+      "loss": 6.4671,
+      "step": 962
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.889e-05,
+      "loss": 6.5592,
+      "step": 963
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.892e-05,
+      "loss": 6.4091,
+      "step": 964
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.895e-05,
+      "loss": 6.3488,
+      "step": 965
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.898e-05,
+      "loss": 6.4542,
+      "step": 966
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.901e-05,
+      "loss": 6.3449,
+      "step": 967
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.904e-05,
+      "loss": 6.266,
+      "step": 968
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.907e-05,
+      "loss": 6.5133,
+      "step": 969
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.91e-05,
+      "loss": 6.3458,
+      "step": 970
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.913e-05,
+      "loss": 6.386,
+      "step": 971
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.916e-05,
+      "loss": 6.39,
+      "step": 972
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.919e-05,
+      "loss": 6.474,
+      "step": 973
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.922e-05,
+      "loss": 6.4531,
+      "step": 974
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.925e-05,
+      "loss": 6.3094,
+      "step": 975
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.928e-05,
+      "loss": 6.3801,
+      "step": 976
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.931e-05,
+      "loss": 6.5444,
+      "step": 977
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.934e-05,
+      "loss": 6.2127,
+      "step": 978
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.9370000000000002e-05,
+      "loss": 6.3919,
+      "step": 979
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.94e-05,
+      "loss": 6.3265,
+      "step": 980
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.943e-05,
+      "loss": 6.2449,
+      "step": 981
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.946e-05,
+      "loss": 6.5581,
+      "step": 982
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.949e-05,
+      "loss": 6.6469,
+      "step": 983
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.9520000000000002e-05,
+      "loss": 6.3359,
+      "step": 984
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.955e-05,
+      "loss": 6.1187,
+      "step": 985
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.958e-05,
+      "loss": 6.3318,
+      "step": 986
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.961e-05,
+      "loss": 6.2047,
+      "step": 987
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.964e-05,
+      "loss": 6.0997,
+      "step": 988
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.967e-05,
+      "loss": 5.976,
+      "step": 989
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.97e-05,
+      "loss": 6.3181,
+      "step": 990
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.973e-05,
+      "loss": 5.7822,
+      "step": 991
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.976e-05,
+      "loss": 5.6529,
+      "step": 992
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.979e-05,
+      "loss": 6.4569,
+      "step": 993
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.982e-05,
+      "loss": 6.5987,
+      "step": 994
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.985e-05,
+      "loss": 6.5572,
+      "step": 995
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.9880000000000002e-05,
+      "loss": 6.5939,
+      "step": 996
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.991e-05,
+      "loss": 6.3698,
+      "step": 997
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.994e-05,
+      "loss": 6.6087,
+      "step": 998
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.997e-05,
+      "loss": 6.4409,
+      "step": 999
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3e-05,
+      "loss": 6.2052,
+      "step": 1000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.9991329479768787e-05,
+      "loss": 6.3962,
+      "step": 1001
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.9982658959537573e-05,
+      "loss": 6.4514,
+      "step": 1002
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.997398843930636e-05,
+      "loss": 6.349,
+      "step": 1003
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.9965317919075145e-05,
+      "loss": 6.4812,
+      "step": 1004
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.9956647398843934e-05,
+      "loss": 6.549,
+      "step": 1005
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.9947976878612717e-05,
+      "loss": 6.3698,
+      "step": 1006
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.9939306358381503e-05,
+      "loss": 6.3286,
+      "step": 1007
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.993063583815029e-05,
+      "loss": 6.3825,
+      "step": 1008
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.9921965317919075e-05,
+      "loss": 6.3354,
+      "step": 1009
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.991329479768786e-05,
+      "loss": 6.4124,
+      "step": 1010
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.9904624277456647e-05,
+      "loss": 6.4267,
+      "step": 1011
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.9895953757225433e-05,
+      "loss": 6.4854,
+      "step": 1012
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.9887283236994223e-05,
+      "loss": 6.2181,
+      "step": 1013
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.987861271676301e-05,
+      "loss": 6.4243,
+      "step": 1014
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.9869942196531795e-05,
+      "loss": 6.2302,
+      "step": 1015
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.9861271676300577e-05,
+      "loss": 6.3434,
+      "step": 1016
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.9852601156069363e-05,
+      "loss": 6.3277,
+      "step": 1017
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.984393063583815e-05,
+      "loss": 6.4775,
+      "step": 1018
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.9835260115606935e-05,
+      "loss": 6.3528,
+      "step": 1019
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.982658959537572e-05,
+      "loss": 6.5094,
+      "step": 1020
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.981791907514451e-05,
+      "loss": 6.3824,
+      "step": 1021
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.9809248554913297e-05,
+      "loss": 6.3059,
+      "step": 1022
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.9800578034682083e-05,
+      "loss": 6.3117,
+      "step": 1023
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.979190751445087e-05,
+      "loss": 6.4707,
+      "step": 1024
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.9783236994219655e-05,
+      "loss": 6.1623,
+      "step": 1025
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.977456647398844e-05,
+      "loss": 6.2285,
+      "step": 1026
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.9765895953757224e-05,
+      "loss": 6.5116,
+      "step": 1027
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.975722543352601e-05,
+      "loss": 6.1918,
+      "step": 1028
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.97485549132948e-05,
+      "loss": 6.0719,
+      "step": 1029
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.9739884393063585e-05,
+      "loss": 6.4236,
+      "step": 1030
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.973121387283237e-05,
+      "loss": 6.1679,
+      "step": 1031
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.9722543352601157e-05,
+      "loss": 6.214,
+      "step": 1032
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.9713872832369943e-05,
+      "loss": 6.2791,
+      "step": 1033
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.970520231213873e-05,
+      "loss": 6.3302,
+      "step": 1034
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.9696531791907515e-05,
+      "loss": 6.4467,
+      "step": 1035
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.96878612716763e-05,
+      "loss": 6.3552,
+      "step": 1036
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.9679190751445087e-05,
+      "loss": 6.2603,
+      "step": 1037
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.9670520231213873e-05,
+      "loss": 6.3424,
+      "step": 1038
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.966184971098266e-05,
+      "loss": 6.0977,
+      "step": 1039
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.9653179190751446e-05,
+      "loss": 6.2095,
+      "step": 1040
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.964450867052023e-05,
+      "loss": 6.0595,
+      "step": 1041
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.9635838150289018e-05,
+      "loss": 5.7647,
+      "step": 1042
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.9627167630057804e-05,
+      "loss": 6.5883,
+      "step": 1043
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.961849710982659e-05,
+      "loss": 6.5554,
+      "step": 1044
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.960982658959538e-05,
+      "loss": 6.5449,
+      "step": 1045
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.9601156069364165e-05,
+      "loss": 6.4869,
+      "step": 1046
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.9592485549132948e-05,
+      "loss": 6.5141,
+      "step": 1047
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.9583815028901734e-05,
+      "loss": 6.3538,
+      "step": 1048
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.957514450867052e-05,
+      "loss": 6.4335,
+      "step": 1049
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.9566473988439306e-05,
+      "loss": 6.378,
+      "step": 1050
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.9557803468208092e-05,
+      "loss": 6.2721,
+      "step": 1051
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.9549132947976878e-05,
+      "loss": 6.5694,
+      "step": 1052
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.9540462427745667e-05,
+      "loss": 6.4083,
+      "step": 1053
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.9531791907514453e-05,
+      "loss": 6.5434,
+      "step": 1054
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.952312138728324e-05,
+      "loss": 6.3229,
+      "step": 1055
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.9514450867052025e-05,
+      "loss": 6.2929,
+      "step": 1056
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.950578034682081e-05,
+      "loss": 6.4637,
+      "step": 1057
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.9497109826589594e-05,
+      "loss": 6.5493,
+      "step": 1058
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.948843930635838e-05,
+      "loss": 6.2688,
+      "step": 1059
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.9479768786127166e-05,
+      "loss": 6.2804,
+      "step": 1060
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.9471098265895956e-05,
+      "loss": 6.3573,
+      "step": 1061
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.946242774566474e-05,
+      "loss": 6.3813,
+      "step": 1062
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.9453757225433528e-05,
+      "loss": 6.2867,
+      "step": 1063
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.9445086705202314e-05,
+      "loss": 6.3384,
+      "step": 1064
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.94364161849711e-05,
+      "loss": 6.4041,
+      "step": 1065
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9427745664739886e-05,
+      "loss": 6.3254,
+      "step": 1066
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9419075144508672e-05,
+      "loss": 6.1958,
+      "step": 1067
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9410404624277454e-05,
+      "loss": 6.2452,
+      "step": 1068
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9401734104046244e-05,
+      "loss": 6.5093,
+      "step": 1069
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.939306358381503e-05,
+      "loss": 6.3606,
+      "step": 1070
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9384393063583816e-05,
+      "loss": 6.3113,
+      "step": 1071
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9375722543352602e-05,
+      "loss": 6.4034,
+      "step": 1072
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9367052023121388e-05,
+      "loss": 6.2713,
+      "step": 1073
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9358381502890174e-05,
+      "loss": 6.1548,
+      "step": 1074
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.934971098265896e-05,
+      "loss": 6.2458,
+      "step": 1075
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9341040462427746e-05,
+      "loss": 6.2965,
+      "step": 1076
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9332369942196536e-05,
+      "loss": 6.1909,
+      "step": 1077
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9323699421965318e-05,
+      "loss": 6.3795,
+      "step": 1078
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9315028901734104e-05,
+      "loss": 6.3967,
+      "step": 1079
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.930635838150289e-05,
+      "loss": 6.3126,
+      "step": 1080
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9297687861271676e-05,
+      "loss": 6.2782,
+      "step": 1081
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9289017341040462e-05,
+      "loss": 6.3584,
+      "step": 1082
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.928034682080925e-05,
+      "loss": 6.34,
+      "step": 1083
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9271676300578034e-05,
+      "loss": 6.2096,
+      "step": 1084
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9263005780346824e-05,
+      "loss": 6.2644,
+      "step": 1085
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.925433526011561e-05,
+      "loss": 6.1852,
+      "step": 1086
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9245664739884396e-05,
+      "loss": 6.2131,
+      "step": 1087
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9236994219653182e-05,
+      "loss": 6.324,
+      "step": 1088
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9228323699421965e-05,
+      "loss": 6.1627,
+      "step": 1089
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.921965317919075e-05,
+      "loss": 6.0056,
+      "step": 1090
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9210982658959537e-05,
+      "loss": 5.9909,
+      "step": 1091
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9202312138728323e-05,
+      "loss": 5.7031,
+      "step": 1092
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9193641618497112e-05,
+      "loss": 6.6192,
+      "step": 1093
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9184971098265898e-05,
+      "loss": 6.3992,
+      "step": 1094
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9176300578034684e-05,
+      "loss": 6.5089,
+      "step": 1095
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.916763005780347e-05,
+      "loss": 6.5741,
+      "step": 1096
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9158959537572256e-05,
+      "loss": 6.5924,
+      "step": 1097
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9150289017341042e-05,
+      "loss": 6.4528,
+      "step": 1098
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9141618497109825e-05,
+      "loss": 6.5165,
+      "step": 1099
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.913294797687861e-05,
+      "loss": 6.5939,
+      "step": 1100
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9124277456647397e-05,
+      "loss": 6.3692,
+      "step": 1101
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9115606936416186e-05,
+      "loss": 6.2597,
+      "step": 1102
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9106936416184972e-05,
+      "loss": 6.1679,
+      "step": 1103
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.909826589595376e-05,
+      "loss": 6.4271,
+      "step": 1104
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9089595375722544e-05,
+      "loss": 6.3509,
+      "step": 1105
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.908092485549133e-05,
+      "loss": 6.313,
+      "step": 1106
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9072254335260117e-05,
+      "loss": 6.3069,
+      "step": 1107
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9063583815028903e-05,
+      "loss": 6.169,
+      "step": 1108
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.905491329479769e-05,
+      "loss": 6.2338,
+      "step": 1109
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9046242774566475e-05,
+      "loss": 6.2431,
+      "step": 1110
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.903757225433526e-05,
+      "loss": 6.3758,
+      "step": 1111
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9028901734104047e-05,
+      "loss": 6.2712,
+      "step": 1112
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9020231213872833e-05,
+      "loss": 6.3099,
+      "step": 1113
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.901156069364162e-05,
+      "loss": 6.336,
+      "step": 1114
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9002890173410405e-05,
+      "loss": 6.3504,
+      "step": 1115
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.899421965317919e-05,
+      "loss": 6.1575,
+      "step": 1116
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.8985549132947977e-05,
+      "loss": 6.2228,
+      "step": 1117
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.8976878612716766e-05,
+      "loss": 6.4242,
+      "step": 1118
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.8968208092485552e-05,
+      "loss": 6.267,
+      "step": 1119
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8959537572254335e-05,
+      "loss": 6.2855,
+      "step": 1120
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.895086705202312e-05,
+      "loss": 6.3305,
+      "step": 1121
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8942196531791907e-05,
+      "loss": 6.3657,
+      "step": 1122
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8933526011560693e-05,
+      "loss": 6.2415,
+      "step": 1123
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.892485549132948e-05,
+      "loss": 6.3322,
+      "step": 1124
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8916184971098265e-05,
+      "loss": 6.3352,
+      "step": 1125
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8907514450867055e-05,
+      "loss": 6.4829,
+      "step": 1126
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.889884393063584e-05,
+      "loss": 6.2653,
+      "step": 1127
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8890173410404627e-05,
+      "loss": 6.2235,
+      "step": 1128
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8881502890173413e-05,
+      "loss": 6.3938,
+      "step": 1129
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8872832369942195e-05,
+      "loss": 6.1937,
+      "step": 1130
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.886416184971098e-05,
+      "loss": 6.3012,
+      "step": 1131
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8855491329479767e-05,
+      "loss": 6.1637,
+      "step": 1132
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8846820809248553e-05,
+      "loss": 6.3318,
+      "step": 1133
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8838150289017343e-05,
+      "loss": 6.447,
+      "step": 1134
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.882947976878613e-05,
+      "loss": 5.7901,
+      "step": 1135
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8820809248554915e-05,
+      "loss": 5.981,
+      "step": 1136
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.88121387283237e-05,
+      "loss": 6.2895,
+      "step": 1137
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8803468208092487e-05,
+      "loss": 5.8951,
+      "step": 1138
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8794797687861273e-05,
+      "loss": 5.9531,
+      "step": 1139
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.878612716763006e-05,
+      "loss": 5.9991,
+      "step": 1140
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8777456647398842e-05,
+      "loss": 6.1306,
+      "step": 1141
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.876878612716763e-05,
+      "loss": 6.0785,
+      "step": 1142
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8760115606936417e-05,
+      "loss": 6.5917,
+      "step": 1143
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8751445086705203e-05,
+      "loss": 6.3198,
+      "step": 1144
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.874277456647399e-05,
+      "loss": 6.3736,
+      "step": 1145
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8734104046242775e-05,
+      "loss": 6.5297,
+      "step": 1146
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.872543352601156e-05,
+      "loss": 6.3577,
+      "step": 1147
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8716763005780347e-05,
+      "loss": 6.4607,
+      "step": 1148
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8708092485549133e-05,
+      "loss": 6.3468,
+      "step": 1149
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8699421965317923e-05,
+      "loss": 6.3519,
+      "step": 1150
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8690751445086705e-05,
+      "loss": 6.2017,
+      "step": 1151
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.868208092485549e-05,
+      "loss": 6.2174,
+      "step": 1152
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8673410404624278e-05,
+      "loss": 6.3174,
+      "step": 1153
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8664739884393064e-05,
+      "loss": 6.4212,
+      "step": 1154
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.865606936416185e-05,
+      "loss": 6.4011,
+      "step": 1155
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8647398843930636e-05,
+      "loss": 6.2459,
+      "step": 1156
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.863872832369942e-05,
+      "loss": 6.2751,
+      "step": 1157
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.863005780346821e-05,
+      "loss": 6.5429,
+      "step": 1158
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8621387283236997e-05,
+      "loss": 6.3178,
+      "step": 1159
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8612716763005783e-05,
+      "loss": 6.1803,
+      "step": 1160
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.860404624277457e-05,
+      "loss": 6.3823,
+      "step": 1161
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8595375722543352e-05,
+      "loss": 6.2735,
+      "step": 1162
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8586705202312138e-05,
+      "loss": 6.2115,
+      "step": 1163
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8578034682080924e-05,
+      "loss": 6.117,
+      "step": 1164
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.856936416184971e-05,
+      "loss": 6.2632,
+      "step": 1165
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.85606936416185e-05,
+      "loss": 6.0609,
+      "step": 1166
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8552023121387285e-05,
+      "loss": 6.5572,
+      "step": 1167
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.854335260115607e-05,
+      "loss": 6.3156,
+      "step": 1168
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8534682080924857e-05,
+      "loss": 6.5028,
+      "step": 1169
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8526011560693643e-05,
+      "loss": 6.4658,
+      "step": 1170
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.851734104046243e-05,
+      "loss": 6.3226,
+      "step": 1171
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8508670520231212e-05,
+      "loss": 6.3967,
+      "step": 1172
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 6.2672,
+      "step": 1173
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8491329479768788e-05,
+      "loss": 6.2762,
+      "step": 1174
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8482658959537574e-05,
+      "loss": 6.2904,
+      "step": 1175
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.847398843930636e-05,
+      "loss": 6.3865,
+      "step": 1176
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8465317919075146e-05,
+      "loss": 6.3683,
+      "step": 1177
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8456647398843932e-05,
+      "loss": 6.1805,
+      "step": 1178
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8447976878612718e-05,
+      "loss": 6.3661,
+      "step": 1179
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8439306358381504e-05,
+      "loss": 6.2054,
+      "step": 1180
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.843063583815029e-05,
+      "loss": 6.2389,
+      "step": 1181
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.8421965317919076e-05,
+      "loss": 6.3837,
+      "step": 1182
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.8413294797687862e-05,
+      "loss": 6.4403,
+      "step": 1183
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.8404624277456648e-05,
+      "loss": 6.3659,
+      "step": 1184
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.8395953757225434e-05,
+      "loss": 6.0583,
+      "step": 1185
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.838728323699422e-05,
+      "loss": 6.0025,
+      "step": 1186
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.8378612716763006e-05,
+      "loss": 5.9863,
+      "step": 1187
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.8369942196531792e-05,
+      "loss": 6.1624,
+      "step": 1188
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.8361271676300578e-05,
+      "loss": 5.7929,
+      "step": 1189
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.8352601156069368e-05,
+      "loss": 5.7544,
+      "step": 1190
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.8343930635838154e-05,
+      "loss": 5.9114,
+      "step": 1191
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.833526011560694e-05,
+      "loss": 5.5662,
+      "step": 1192
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.8326589595375722e-05,
+      "loss": 6.707,
+      "step": 1193
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.8317919075144508e-05,
+      "loss": 6.6331,
+      "step": 1194
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.8309248554913294e-05,
+      "loss": 6.5638,
+      "step": 1195
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.830057803468208e-05,
+      "loss": 6.6016,
+      "step": 1196
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.8291907514450866e-05,
+      "loss": 6.5589,
+      "step": 1197
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.8283236994219652e-05,
+      "loss": 6.3828,
+      "step": 1198
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.8274566473988442e-05,
+      "loss": 6.4907,
+      "step": 1199
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.8265895953757228e-05,
+      "loss": 6.5001,
+      "step": 1200
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.8257225433526014e-05,
+      "loss": 6.3519,
+      "step": 1201
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.82485549132948e-05,
+      "loss": 6.3831,
+      "step": 1202
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.8239884393063583e-05,
+      "loss": 6.2969,
+      "step": 1203
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.823121387283237e-05,
+      "loss": 6.2844,
+      "step": 1204
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.8222543352601155e-05,
+      "loss": 6.2959,
+      "step": 1205
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.821387283236994e-05,
+      "loss": 6.2394,
+      "step": 1206
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.820520231213873e-05,
+      "loss": 6.2634,
+      "step": 1207
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.8196531791907516e-05,
+      "loss": 6.278,
+      "step": 1208
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.8187861271676302e-05,
+      "loss": 6.1984,
+      "step": 1209
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.8179190751445088e-05,
+      "loss": 6.3644,
+      "step": 1210
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.8170520231213874e-05,
+      "loss": 6.4176,
+      "step": 1211
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.816184971098266e-05,
+      "loss": 6.3551,
+      "step": 1212
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.8153179190751446e-05,
+      "loss": 6.423,
+      "step": 1213
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.814450867052023e-05,
+      "loss": 6.2585,
+      "step": 1214
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.813583815028902e-05,
+      "loss": 6.4339,
+      "step": 1215
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.8127167630057804e-05,
+      "loss": 6.1676,
+      "step": 1216
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.811849710982659e-05,
+      "loss": 6.2461,
+      "step": 1217
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.8109826589595376e-05,
+      "loss": 6.3715,
+      "step": 1218
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.8101156069364163e-05,
+      "loss": 6.2221,
+      "step": 1219
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.809248554913295e-05,
+      "loss": 6.2326,
+      "step": 1220
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.8083815028901735e-05,
+      "loss": 6.3205,
+      "step": 1221
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.807514450867052e-05,
+      "loss": 6.2443,
+      "step": 1222
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.806647398843931e-05,
+      "loss": 6.1953,
+      "step": 1223
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.8057803468208093e-05,
+      "loss": 6.3148,
+      "step": 1224
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.804913294797688e-05,
+      "loss": 6.3438,
+      "step": 1225
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.8040462427745665e-05,
+      "loss": 6.4073,
+      "step": 1226
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.803179190751445e-05,
+      "loss": 6.3325,
+      "step": 1227
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.8023121387283237e-05,
+      "loss": 6.2955,
+      "step": 1228
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.8014450867052023e-05,
+      "loss": 6.3203,
+      "step": 1229
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.800578034682081e-05,
+      "loss": 6.0468,
+      "step": 1230
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.79971098265896e-05,
+      "loss": 6.1508,
+      "step": 1231
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7988439306358384e-05,
+      "loss": 6.3438,
+      "step": 1232
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.797976878612717e-05,
+      "loss": 6.2987,
+      "step": 1233
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7971098265895953e-05,
+      "loss": 6.3372,
+      "step": 1234
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.796242774566474e-05,
+      "loss": 6.2596,
+      "step": 1235
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.7953757225433525e-05,
+      "loss": 6.0764,
+      "step": 1236
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.794508670520231e-05,
+      "loss": 6.0871,
+      "step": 1237
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.7936416184971097e-05,
+      "loss": 6.0701,
+      "step": 1238
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.7927745664739887e-05,
+      "loss": 5.9842,
+      "step": 1239
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.7919075144508673e-05,
+      "loss": 5.7866,
+      "step": 1240
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.791040462427746e-05,
+      "loss": 5.7193,
+      "step": 1241
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.7901734104046245e-05,
+      "loss": 5.8465,
+      "step": 1242
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.789306358381503e-05,
+      "loss": 6.6361,
+      "step": 1243
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.7884393063583817e-05,
+      "loss": 6.3786,
+      "step": 1244
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.78757225433526e-05,
+      "loss": 6.5653,
+      "step": 1245
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.7867052023121385e-05,
+      "loss": 6.2996,
+      "step": 1246
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.7858381502890175e-05,
+      "loss": 6.3742,
+      "step": 1247
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.784971098265896e-05,
+      "loss": 6.3763,
+      "step": 1248
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.7841040462427747e-05,
+      "loss": 6.3122,
+      "step": 1249
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.7832369942196533e-05,
+      "loss": 6.324,
+      "step": 1250
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.782369942196532e-05,
+      "loss": 6.2267,
+      "step": 1251
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.7815028901734105e-05,
+      "loss": 6.0222,
+      "step": 1252
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.780635838150289e-05,
+      "loss": 5.9934,
+      "step": 1253
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.7797687861271677e-05,
+      "loss": 6.3977,
+      "step": 1254
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.7789017341040463e-05,
+      "loss": 6.3024,
+      "step": 1255
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.778034682080925e-05,
+      "loss": 6.1206,
+      "step": 1256
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.7771676300578035e-05,
+      "loss": 6.214,
+      "step": 1257
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.776300578034682e-05,
+      "loss": 6.2634,
+      "step": 1258
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.7754335260115607e-05,
+      "loss": 6.2009,
+      "step": 1259
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.7745664739884393e-05,
+      "loss": 6.2189,
+      "step": 1260
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.773699421965318e-05,
+      "loss": 6.2113,
+      "step": 1261
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.7728323699421965e-05,
+      "loss": 6.3698,
+      "step": 1262
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.7719653179190755e-05,
+      "loss": 6.0034,
+      "step": 1263
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.771098265895954e-05,
+      "loss": 6.17,
+      "step": 1264
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.7702312138728323e-05,
+      "loss": 6.1709,
+      "step": 1265
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.769364161849711e-05,
+      "loss": 6.1716,
+      "step": 1266
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.7684971098265896e-05,
+      "loss": 6.3823,
+      "step": 1267
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.767630057803468e-05,
+      "loss": 6.1881,
+      "step": 1268
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.7667630057803468e-05,
+      "loss": 6.1436,
+      "step": 1269
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.7658959537572254e-05,
+      "loss": 6.2631,
+      "step": 1270
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.7650289017341043e-05,
+      "loss": 6.0763,
+      "step": 1271
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.764161849710983e-05,
+      "loss": 6.1757,
+      "step": 1272
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.7632947976878615e-05,
+      "loss": 6.1577,
+      "step": 1273
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.76242774566474e-05,
+      "loss": 6.1243,
+      "step": 1274
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.7615606936416187e-05,
+      "loss": 6.1965,
+      "step": 1275
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.760693641618497e-05,
+      "loss": 6.1083,
+      "step": 1276
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.7598265895953756e-05,
+      "loss": 6.4865,
+      "step": 1277
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.7589595375722542e-05,
+      "loss": 6.1774,
+      "step": 1278
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.758092485549133e-05,
+      "loss": 6.341,
+      "step": 1279
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.7572254335260117e-05,
+      "loss": 6.1177,
+      "step": 1280
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.7563583815028903e-05,
+      "loss": 6.1652,
+      "step": 1281
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.755491329479769e-05,
+      "loss": 6.1179,
+      "step": 1282
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.7546242774566475e-05,
+      "loss": 6.2109,
+      "step": 1283
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.753757225433526e-05,
+      "loss": 6.1062,
+      "step": 1284
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.7528901734104048e-05,
+      "loss": 6.2062,
+      "step": 1285
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.752023121387283e-05,
+      "loss": 6.2357,
+      "step": 1286
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.751156069364162e-05,
+      "loss": 5.909,
+      "step": 1287
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.7502890173410406e-05,
+      "loss": 6.1103,
+      "step": 1288
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.749421965317919e-05,
+      "loss": 6.215,
+      "step": 1289
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.7485549132947978e-05,
+      "loss": 5.9143,
+      "step": 1290
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.7476878612716764e-05,
+      "loss": 6.1953,
+      "step": 1291
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.746820809248555e-05,
+      "loss": 6.0269,
+      "step": 1292
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.7459537572254336e-05,
+      "loss": 6.4823,
+      "step": 1293
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.7450867052023122e-05,
+      "loss": 6.5392,
+      "step": 1294
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.7442196531791908e-05,
+      "loss": 6.4642,
+      "step": 1295
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.7433526011560697e-05,
+      "loss": 6.3856,
+      "step": 1296
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.742485549132948e-05,
+      "loss": 6.2614,
+      "step": 1297
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.7416184971098266e-05,
+      "loss": 6.3641,
+      "step": 1298
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.7407514450867052e-05,
+      "loss": 6.2064,
+      "step": 1299
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.7398843930635838e-05,
+      "loss": 6.4005,
+      "step": 1300
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.7390173410404624e-05,
+      "loss": 6.3837,
+      "step": 1301
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.738150289017341e-05,
+      "loss": 6.2013,
+      "step": 1302
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.7372832369942196e-05,
+      "loss": 6.1745,
+      "step": 1303
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.7364161849710986e-05,
+      "loss": 6.2257,
+      "step": 1304
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.735549132947977e-05,
+      "loss": 6.2837,
+      "step": 1305
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.7346820809248558e-05,
+      "loss": 6.2289,
+      "step": 1306
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.733815028901734e-05,
+      "loss": 6.068,
+      "step": 1307
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.7329479768786126e-05,
+      "loss": 6.332,
+      "step": 1308
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.7320809248554912e-05,
+      "loss": 6.1209,
+      "step": 1309
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.73121387283237e-05,
+      "loss": 6.195,
+      "step": 1310
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.7303468208092484e-05,
+      "loss": 6.2316,
+      "step": 1311
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.7294797687861274e-05,
+      "loss": 6.3516,
+      "step": 1312
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.728612716763006e-05,
+      "loss": 6.0744,
+      "step": 1313
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.7277456647398846e-05,
+      "loss": 6.2351,
+      "step": 1314
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.7268786127167632e-05,
+      "loss": 6.2587,
+      "step": 1315
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.7260115606936418e-05,
+      "loss": 6.1306,
+      "step": 1316
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.72514450867052e-05,
+      "loss": 6.2312,
+      "step": 1317
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.7242774566473987e-05,
+      "loss": 6.3282,
+      "step": 1318
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.7234104046242773e-05,
+      "loss": 6.3005,
+      "step": 1319
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.7225433526011562e-05,
+      "loss": 6.309,
+      "step": 1320
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.7216763005780348e-05,
+      "loss": 6.162,
+      "step": 1321
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.7208092485549134e-05,
+      "loss": 6.3355,
+      "step": 1322
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.719942196531792e-05,
+      "loss": 6.2424,
+      "step": 1323
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.7190751445086706e-05,
+      "loss": 6.1945,
+      "step": 1324
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.7182080924855492e-05,
+      "loss": 6.3358,
+      "step": 1325
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.717341040462428e-05,
+      "loss": 6.2052,
+      "step": 1326
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.7164739884393064e-05,
+      "loss": 6.3334,
+      "step": 1327
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.715606936416185e-05,
+      "loss": 6.2859,
+      "step": 1328
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.7147398843930636e-05,
+      "loss": 6.4278,
+      "step": 1329
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.7138728323699422e-05,
+      "loss": 6.244,
+      "step": 1330
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.713005780346821e-05,
+      "loss": 6.4237,
+      "step": 1331
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.7121387283236995e-05,
+      "loss": 6.2479,
+      "step": 1332
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.711271676300578e-05,
+      "loss": 6.1284,
+      "step": 1333
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.7104046242774567e-05,
+      "loss": 6.2508,
+      "step": 1334
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.7095375722543353e-05,
+      "loss": 6.1688,
+      "step": 1335
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.7086705202312142e-05,
+      "loss": 6.1574,
+      "step": 1336
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.7078034682080928e-05,
+      "loss": 6.1217,
+      "step": 1337
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.706936416184971e-05,
+      "loss": 6.0566,
+      "step": 1338
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.7060693641618497e-05,
+      "loss": 5.902,
+      "step": 1339
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.7052023121387283e-05,
+      "loss": 6.0428,
+      "step": 1340
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.704335260115607e-05,
+      "loss": 5.8236,
+      "step": 1341
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.7034682080924855e-05,
+      "loss": 5.3147,
+      "step": 1342
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.702601156069364e-05,
+      "loss": 6.7084,
+      "step": 1343
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.701734104046243e-05,
+      "loss": 6.2346,
+      "step": 1344
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.7008670520231216e-05,
+      "loss": 6.3949,
+      "step": 1345
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 6.2309,
+      "step": 1346
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.699132947976879e-05,
+      "loss": 6.3097,
+      "step": 1347
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.6982658959537574e-05,
+      "loss": 6.3885,
+      "step": 1348
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.6973988439306357e-05,
+      "loss": 6.2782,
+      "step": 1349
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.6965317919075143e-05,
+      "loss": 6.2067,
+      "step": 1350
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.695664739884393e-05,
+      "loss": 6.272,
+      "step": 1351
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.694797687861272e-05,
+      "loss": 6.2171,
+      "step": 1352
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.6939306358381505e-05,
+      "loss": 6.1971,
+      "step": 1353
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.693063583815029e-05,
+      "loss": 6.2717,
+      "step": 1354
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.6921965317919077e-05,
+      "loss": 6.2004,
+      "step": 1355
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.6913294797687863e-05,
+      "loss": 6.2507,
+      "step": 1356
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.690462427745665e-05,
+      "loss": 6.2065,
+      "step": 1357
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.6895953757225435e-05,
+      "loss": 6.3333,
+      "step": 1358
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.6887283236994217e-05,
+      "loss": 6.1322,
+      "step": 1359
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.6878612716763007e-05,
+      "loss": 5.988,
+      "step": 1360
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.6869942196531793e-05,
+      "loss": 5.9064,
+      "step": 1361
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.686127167630058e-05,
+      "loss": 6.3742,
+      "step": 1362
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.6852601156069365e-05,
+      "loss": 6.1021,
+      "step": 1363
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.684393063583815e-05,
+      "loss": 6.2267,
+      "step": 1364
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.6835260115606937e-05,
+      "loss": 6.2283,
+      "step": 1365
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.6826589595375723e-05,
+      "loss": 6.3929,
+      "step": 1366
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.681791907514451e-05,
+      "loss": 6.1938,
+      "step": 1367
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.68092485549133e-05,
+      "loss": 6.2273,
+      "step": 1368
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.680057803468208e-05,
+      "loss": 6.1166,
+      "step": 1369
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6791907514450867e-05,
+      "loss": 6.0771,
+      "step": 1370
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6783236994219653e-05,
+      "loss": 6.2733,
+      "step": 1371
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.677456647398844e-05,
+      "loss": 6.2502,
+      "step": 1372
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6765895953757225e-05,
+      "loss": 6.162,
+      "step": 1373
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.675722543352601e-05,
+      "loss": 6.084,
+      "step": 1374
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6748554913294797e-05,
+      "loss": 6.2742,
+      "step": 1375
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6739884393063587e-05,
+      "loss": 6.2197,
+      "step": 1376
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.6731213872832373e-05,
+      "loss": 5.9182,
+      "step": 1377
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.672254335260116e-05,
+      "loss": 6.0963,
+      "step": 1378
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.6713872832369945e-05,
+      "loss": 6.3411,
+      "step": 1379
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.6705202312138728e-05,
+      "loss": 6.3459,
+      "step": 1380
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.6696531791907514e-05,
+      "loss": 6.1801,
+      "step": 1381
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.66878612716763e-05,
+      "loss": 6.2793,
+      "step": 1382
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.6679190751445086e-05,
+      "loss": 5.9925,
+      "step": 1383
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.6670520231213875e-05,
+      "loss": 6.1056,
+      "step": 1384
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.666184971098266e-05,
+      "loss": 6.2376,
+      "step": 1385
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.6653179190751447e-05,
+      "loss": 5.9738,
+      "step": 1386
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.6644508670520233e-05,
+      "loss": 6.1732,
+      "step": 1387
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.663583815028902e-05,
+      "loss": 6.2517,
+      "step": 1388
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.6627167630057805e-05,
+      "loss": 5.6835,
+      "step": 1389
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.6618497109826588e-05,
+      "loss": 5.8376,
+      "step": 1390
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.6609826589595374e-05,
+      "loss": 5.9564,
+      "step": 1391
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.660115606936416e-05,
+      "loss": 5.3964,
+      "step": 1392
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.659248554913295e-05,
+      "loss": 6.3731,
+      "step": 1393
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.6583815028901735e-05,
+      "loss": 6.5015,
+      "step": 1394
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.657514450867052e-05,
+      "loss": 6.3761,
+      "step": 1395
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.6566473988439307e-05,
+      "loss": 6.2261,
+      "step": 1396
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.6557803468208093e-05,
+      "loss": 6.0833,
+      "step": 1397
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.654913294797688e-05,
+      "loss": 6.2733,
+      "step": 1398
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.6540462427745666e-05,
+      "loss": 6.1612,
+      "step": 1399
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.653179190751445e-05,
+      "loss": 6.2686,
+      "step": 1400
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.6523121387283238e-05,
+      "loss": 6.2355,
+      "step": 1401
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.6514450867052024e-05,
+      "loss": 6.3069,
+      "step": 1402
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.650578034682081e-05,
+      "loss": 6.1415,
+      "step": 1403
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.6497109826589596e-05,
+      "loss": 6.3354,
+      "step": 1404
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.6488439306358382e-05,
+      "loss": 6.0942,
+      "step": 1405
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.6479768786127168e-05,
+      "loss": 6.2311,
+      "step": 1406
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.6471098265895954e-05,
+      "loss": 6.2892,
+      "step": 1407
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.646242774566474e-05,
+      "loss": 6.1027,
+      "step": 1408
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.645375722543353e-05,
+      "loss": 6.2497,
+      "step": 1409
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.6445086705202315e-05,
+      "loss": 6.2667,
+      "step": 1410
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.6436416184971098e-05,
+      "loss": 6.2227,
+      "step": 1411
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.6427745664739884e-05,
+      "loss": 6.2527,
+      "step": 1412
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.641907514450867e-05,
+      "loss": 6.1005,
+      "step": 1413
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.6410404624277456e-05,
+      "loss": 6.1341,
+      "step": 1414
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.6401734104046242e-05,
+      "loss": 6.2009,
+      "step": 1415
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.6393063583815028e-05,
+      "loss": 5.9807,
+      "step": 1416
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.6384393063583818e-05,
+      "loss": 6.0298,
+      "step": 1417
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.6375722543352604e-05,
+      "loss": 6.0276,
+      "step": 1418
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.636705202312139e-05,
+      "loss": 6.0892,
+      "step": 1419
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.6358381502890176e-05,
+      "loss": 6.368,
+      "step": 1420
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.634971098265896e-05,
+      "loss": 6.2816,
+      "step": 1421
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.6341040462427744e-05,
+      "loss": 6.3493,
+      "step": 1422
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.633236994219653e-05,
+      "loss": 6.1856,
+      "step": 1423
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.6323699421965316e-05,
+      "loss": 6.0905,
+      "step": 1424
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.6315028901734106e-05,
+      "loss": 6.043,
+      "step": 1425
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.6306358381502892e-05,
+      "loss": 5.9551,
+      "step": 1426
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.6297687861271678e-05,
+      "loss": 6.0545,
+      "step": 1427
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.6289017341040464e-05,
+      "loss": 6.1289,
+      "step": 1428
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.628034682080925e-05,
+      "loss": 6.0019,
+      "step": 1429
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.6271676300578036e-05,
+      "loss": 6.1464,
+      "step": 1430
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.6263005780346822e-05,
+      "loss": 6.1611,
+      "step": 1431
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.6254335260115605e-05,
+      "loss": 6.2996,
+      "step": 1432
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.6245664739884394e-05,
+      "loss": 6.2925,
+      "step": 1433
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.623699421965318e-05,
+      "loss": 5.942,
+      "step": 1434
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.6228323699421966e-05,
+      "loss": 6.1876,
+      "step": 1435
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.6219653179190752e-05,
+      "loss": 6.2594,
+      "step": 1436
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.6210982658959538e-05,
+      "loss": 6.1294,
+      "step": 1437
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.6202312138728324e-05,
+      "loss": 5.7786,
+      "step": 1438
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.619364161849711e-05,
+      "loss": 5.8714,
+      "step": 1439
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.6184971098265896e-05,
+      "loss": 5.8425,
+      "step": 1440
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.6176300578034686e-05,
+      "loss": 5.5191,
+      "step": 1441
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.616763005780347e-05,
+      "loss": 5.7357,
+      "step": 1442
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.6158959537572254e-05,
+      "loss": 6.2819,
+      "step": 1443
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.615028901734104e-05,
+      "loss": 6.2697,
+      "step": 1444
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.6141618497109827e-05,
+      "loss": 6.4875,
+      "step": 1445
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.6132947976878613e-05,
+      "loss": 6.2219,
+      "step": 1446
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.61242774566474e-05,
+      "loss": 6.2837,
+      "step": 1447
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.6115606936416185e-05,
+      "loss": 6.1859,
+      "step": 1448
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.6106936416184974e-05,
+      "loss": 6.194,
+      "step": 1449
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.609826589595376e-05,
+      "loss": 6.2818,
+      "step": 1450
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6089595375722546e-05,
+      "loss": 6.2062,
+      "step": 1451
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.608092485549133e-05,
+      "loss": 6.359,
+      "step": 1452
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6072254335260115e-05,
+      "loss": 6.1438,
+      "step": 1453
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.60635838150289e-05,
+      "loss": 6.048,
+      "step": 1454
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6054913294797687e-05,
+      "loss": 6.265,
+      "step": 1455
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6046242774566473e-05,
+      "loss": 6.1652,
+      "step": 1456
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.6037572254335262e-05,
+      "loss": 6.1899,
+      "step": 1457
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.602890173410405e-05,
+      "loss": 6.0999,
+      "step": 1458
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6020231213872834e-05,
+      "loss": 6.2481,
+      "step": 1459
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.601156069364162e-05,
+      "loss": 6.1707,
+      "step": 1460
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.6002890173410406e-05,
+      "loss": 6.0573,
+      "step": 1461
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5994219653179192e-05,
+      "loss": 6.2444,
+      "step": 1462
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5985549132947975e-05,
+      "loss": 6.0833,
+      "step": 1463
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.597687861271676e-05,
+      "loss": 6.3467,
+      "step": 1464
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.596820809248555e-05,
+      "loss": 6.1711,
+      "step": 1465
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5959537572254337e-05,
+      "loss": 6.1688,
+      "step": 1466
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5950867052023123e-05,
+      "loss": 6.1478,
+      "step": 1467
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.594219653179191e-05,
+      "loss": 6.224,
+      "step": 1468
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5933526011560695e-05,
+      "loss": 6.1833,
+      "step": 1469
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.592485549132948e-05,
+      "loss": 6.0907,
+      "step": 1470
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5916184971098267e-05,
+      "loss": 5.9997,
+      "step": 1471
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5907514450867053e-05,
+      "loss": 6.2394,
+      "step": 1472
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.589884393063584e-05,
+      "loss": 6.0504,
+      "step": 1473
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5890173410404625e-05,
+      "loss": 5.9797,
+      "step": 1474
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.588150289017341e-05,
+      "loss": 5.956,
+      "step": 1475
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5872832369942197e-05,
+      "loss": 6.0492,
+      "step": 1476
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5864161849710983e-05,
+      "loss": 6.1948,
+      "step": 1477
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.585549132947977e-05,
+      "loss": 6.0333,
+      "step": 1478
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5846820809248555e-05,
+      "loss": 6.1764,
+      "step": 1479
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.583815028901734e-05,
+      "loss": 6.1419,
+      "step": 1480
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5829479768786127e-05,
+      "loss": 5.927,
+      "step": 1481
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5820809248554917e-05,
+      "loss": 5.9953,
+      "step": 1482
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5812138728323703e-05,
+      "loss": 5.9734,
+      "step": 1483
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5803468208092485e-05,
+      "loss": 6.172,
+      "step": 1484
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.579479768786127e-05,
+      "loss": 6.1226,
+      "step": 1485
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5786127167630057e-05,
+      "loss": 6.1146,
+      "step": 1486
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5777456647398843e-05,
+      "loss": 5.9554,
+      "step": 1487
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.576878612716763e-05,
+      "loss": 6.0643,
+      "step": 1488
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5760115606936415e-05,
+      "loss": 5.7116,
+      "step": 1489
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5751445086705205e-05,
+      "loss": 6.1036,
+      "step": 1490
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.574277456647399e-05,
+      "loss": 5.6689,
+      "step": 1491
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5734104046242777e-05,
+      "loss": 5.7351,
+      "step": 1492
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5725433526011563e-05,
+      "loss": 6.4811,
+      "step": 1493
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5716763005780346e-05,
+      "loss": 6.2961,
+      "step": 1494
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.570809248554913e-05,
+      "loss": 6.3228,
+      "step": 1495
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5699421965317918e-05,
+      "loss": 6.253,
+      "step": 1496
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5690751445086704e-05,
+      "loss": 6.353,
+      "step": 1497
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5682080924855493e-05,
+      "loss": 6.1662,
+      "step": 1498
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.567341040462428e-05,
+      "loss": 6.3658,
+      "step": 1499
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5664739884393065e-05,
+      "loss": 6.1577,
+      "step": 1500
+    },
+    {
+      "epoch": 1.68,
+      "eval_loss": 6.072827339172363,
+      "eval_runtime": 715.4795,
+      "eval_samples_per_second": 3.693,
+      "eval_steps_per_second": 0.463,
+      "eval_wer": 1.0588923943885549,
+      "step": 1500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.565606936416185e-05,
+      "loss": 6.2235,
+      "step": 1501
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5647398843930637e-05,
+      "loss": 6.1251,
+      "step": 1502
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.5638728323699423e-05,
+      "loss": 5.9952,
+      "step": 1503
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.5630057803468206e-05,
+      "loss": 6.1333,
+      "step": 1504
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.5621387283236992e-05,
+      "loss": 6.1492,
+      "step": 1505
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.561271676300578e-05,
+      "loss": 6.1435,
+      "step": 1506
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.5604046242774567e-05,
+      "loss": 5.9636,
+      "step": 1507
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.5595375722543353e-05,
+      "loss": 6.0902,
+      "step": 1508
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.558670520231214e-05,
+      "loss": 5.9837,
+      "step": 1509
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.5578034682080925e-05,
+      "loss": 5.9935,
+      "step": 1510
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.556936416184971e-05,
+      "loss": 6.0575,
+      "step": 1511
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.5560693641618498e-05,
+      "loss": 6.0852,
+      "step": 1512
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.5552023121387284e-05,
+      "loss": 6.1872,
+      "step": 1513
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.5543352601156073e-05,
+      "loss": 6.2197,
+      "step": 1514
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.5534682080924856e-05,
+      "loss": 6.1075,
+      "step": 1515
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.5526011560693642e-05,
+      "loss": 6.2198,
+      "step": 1516
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.5517341040462428e-05,
+      "loss": 6.0741,
+      "step": 1517
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.5508670520231214e-05,
+      "loss": 6.0675,
+      "step": 1518
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.55e-05,
+      "loss": 6.0507,
+      "step": 1519
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.5491329479768786e-05,
+      "loss": 6.0324,
+      "step": 1520
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.5482658959537572e-05,
+      "loss": 6.2876,
+      "step": 1521
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.547398843930636e-05,
+      "loss": 6.2967,
+      "step": 1522
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.5465317919075147e-05,
+      "loss": 5.9607,
+      "step": 1523
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.5456647398843933e-05,
+      "loss": 6.1573,
+      "step": 1524
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.5447976878612716e-05,
+      "loss": 6.2156,
+      "step": 1525
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.5439306358381502e-05,
+      "loss": 6.069,
+      "step": 1526
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.5430635838150288e-05,
+      "loss": 6.265,
+      "step": 1527
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.5421965317919074e-05,
+      "loss": 6.2058,
+      "step": 1528
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.541329479768786e-05,
+      "loss": 6.1662,
+      "step": 1529
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.540462427745665e-05,
+      "loss": 6.1574,
+      "step": 1530
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.5395953757225436e-05,
+      "loss": 6.0657,
+      "step": 1531
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.538728323699422e-05,
+      "loss": 6.1076,
+      "step": 1532
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.5378612716763008e-05,
+      "loss": 5.9635,
+      "step": 1533
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.5369942196531794e-05,
+      "loss": 6.0725,
+      "step": 1534
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.536127167630058e-05,
+      "loss": 5.9708,
+      "step": 1535
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.5352601156069362e-05,
+      "loss": 6.0953,
+      "step": 1536
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.534393063583815e-05,
+      "loss": 6.0642,
+      "step": 1537
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.5335260115606938e-05,
+      "loss": 5.8414,
+      "step": 1538
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.5326589595375724e-05,
+      "loss": 5.7643,
+      "step": 1539
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.531791907514451e-05,
+      "loss": 5.9545,
+      "step": 1540
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.5309248554913296e-05,
+      "loss": 5.7539,
+      "step": 1541
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.5300578034682082e-05,
+      "loss": 5.3834,
+      "step": 1542
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.5291907514450868e-05,
+      "loss": 6.3866,
+      "step": 1543
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.5283236994219654e-05,
+      "loss": 6.4387,
+      "step": 1544
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.527456647398844e-05,
+      "loss": 6.3733,
+      "step": 1545
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.5265895953757226e-05,
+      "loss": 6.3107,
+      "step": 1546
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.5257225433526012e-05,
+      "loss": 6.2223,
+      "step": 1547
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5248554913294798e-05,
+      "loss": 6.1739,
+      "step": 1548
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5239884393063584e-05,
+      "loss": 6.4111,
+      "step": 1549
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.523121387283237e-05,
+      "loss": 6.1965,
+      "step": 1550
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5222543352601156e-05,
+      "loss": 6.0858,
+      "step": 1551
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5213872832369942e-05,
+      "loss": 6.312,
+      "step": 1552
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.520520231213873e-05,
+      "loss": 6.2323,
+      "step": 1553
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5196531791907518e-05,
+      "loss": 5.965,
+      "step": 1554
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5187861271676304e-05,
+      "loss": 6.0858,
+      "step": 1555
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.5179190751445086e-05,
+      "loss": 6.0304,
+      "step": 1556
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5170520231213872e-05,
+      "loss": 6.0737,
+      "step": 1557
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.516184971098266e-05,
+      "loss": 6.132,
+      "step": 1558
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5153179190751445e-05,
+      "loss": 6.177,
+      "step": 1559
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.514450867052023e-05,
+      "loss": 6.2119,
+      "step": 1560
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5135838150289017e-05,
+      "loss": 6.0694,
+      "step": 1561
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5127167630057806e-05,
+      "loss": 6.2903,
+      "step": 1562
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5118497109826592e-05,
+      "loss": 6.2012,
+      "step": 1563
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5109826589595378e-05,
+      "loss": 6.1201,
+      "step": 1564
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.5101156069364164e-05,
+      "loss": 5.903,
+      "step": 1565
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.509248554913295e-05,
+      "loss": 6.3521,
+      "step": 1566
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.5083815028901733e-05,
+      "loss": 6.1776,
+      "step": 1567
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.507514450867052e-05,
+      "loss": 5.9289,
+      "step": 1568
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.5066473988439305e-05,
+      "loss": 6.0244,
+      "step": 1569
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.5057803468208094e-05,
+      "loss": 6.1044,
+      "step": 1570
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.504913294797688e-05,
+      "loss": 6.0686,
+      "step": 1571
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.5040462427745666e-05,
+      "loss": 6.226,
+      "step": 1572
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.5031791907514452e-05,
+      "loss": 6.1644,
+      "step": 1573
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.502312138728324e-05,
+      "loss": 5.9948,
+      "step": 1574
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.5014450867052024e-05,
+      "loss": 6.087,
+      "step": 1575
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.500578034682081e-05,
+      "loss": 6.1853,
+      "step": 1576
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4997109826589593e-05,
+      "loss": 6.0792,
+      "step": 1577
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.498843930635838e-05,
+      "loss": 6.0771,
+      "step": 1578
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.497976878612717e-05,
+      "loss": 6.0691,
+      "step": 1579
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4971098265895955e-05,
+      "loss": 6.1646,
+      "step": 1580
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.496242774566474e-05,
+      "loss": 6.0865,
+      "step": 1581
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4953757225433527e-05,
+      "loss": 6.1141,
+      "step": 1582
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4945086705202313e-05,
+      "loss": 6.0771,
+      "step": 1583
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.49364161849711e-05,
+      "loss": 6.0976,
+      "step": 1584
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4927745664739885e-05,
+      "loss": 5.9536,
+      "step": 1585
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.491907514450867e-05,
+      "loss": 6.0542,
+      "step": 1586
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4910404624277457e-05,
+      "loss": 6.1033,
+      "step": 1587
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4901734104046243e-05,
+      "loss": 5.7871,
+      "step": 1588
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.489306358381503e-05,
+      "loss": 5.987,
+      "step": 1589
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4884393063583815e-05,
+      "loss": 5.769,
+      "step": 1590
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.48757225433526e-05,
+      "loss": 5.8513,
+      "step": 1591
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4867052023121387e-05,
+      "loss": 5.2961,
+      "step": 1592
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4858381502890173e-05,
+      "loss": 6.3699,
+      "step": 1593
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.484971098265896e-05,
+      "loss": 6.3001,
+      "step": 1594
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.484104046242775e-05,
+      "loss": 6.2508,
+      "step": 1595
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4832369942196535e-05,
+      "loss": 6.21,
+      "step": 1596
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.482369942196532e-05,
+      "loss": 6.1527,
+      "step": 1597
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4815028901734103e-05,
+      "loss": 6.171,
+      "step": 1598
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.480635838150289e-05,
+      "loss": 6.1184,
+      "step": 1599
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4797687861271675e-05,
+      "loss": 6.3504,
+      "step": 1600
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.478901734104046e-05,
+      "loss": 6.2026,
+      "step": 1601
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4780346820809247e-05,
+      "loss": 6.1709,
+      "step": 1602
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4771676300578037e-05,
+      "loss": 6.141,
+      "step": 1603
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4763005780346823e-05,
+      "loss": 6.1761,
+      "step": 1604
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.475433526011561e-05,
+      "loss": 6.1882,
+      "step": 1605
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4745664739884395e-05,
+      "loss": 6.1344,
+      "step": 1606
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.473699421965318e-05,
+      "loss": 6.0855,
+      "step": 1607
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4728323699421964e-05,
+      "loss": 6.1108,
+      "step": 1608
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.471965317919075e-05,
+      "loss": 6.0804,
+      "step": 1609
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4710982658959536e-05,
+      "loss": 6.0005,
+      "step": 1610
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4702312138728325e-05,
+      "loss": 6.1721,
+      "step": 1611
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.469364161849711e-05,
+      "loss": 6.0424,
+      "step": 1612
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4684971098265897e-05,
+      "loss": 6.0632,
+      "step": 1613
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4676300578034683e-05,
+      "loss": 6.1969,
+      "step": 1614
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.466763005780347e-05,
+      "loss": 6.143,
+      "step": 1615
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4658959537572255e-05,
+      "loss": 6.1069,
+      "step": 1616
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.465028901734104e-05,
+      "loss": 6.1832,
+      "step": 1617
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4641618497109827e-05,
+      "loss": 6.3686,
+      "step": 1618
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4632947976878613e-05,
+      "loss": 6.0949,
+      "step": 1619
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.46242774566474e-05,
+      "loss": 5.9097,
+      "step": 1620
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4615606936416185e-05,
+      "loss": 6.1882,
+      "step": 1621
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.460693641618497e-05,
+      "loss": 5.9957,
+      "step": 1622
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4598265895953757e-05,
+      "loss": 6.1606,
+      "step": 1623
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4589595375722544e-05,
+      "loss": 5.9026,
+      "step": 1624
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.458092485549133e-05,
+      "loss": 5.9734,
+      "step": 1625
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4572254335260116e-05,
+      "loss": 5.9898,
+      "step": 1626
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4563583815028905e-05,
+      "loss": 6.0167,
+      "step": 1627
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.455491329479769e-05,
+      "loss": 6.2304,
+      "step": 1628
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4546242774566474e-05,
+      "loss": 6.1438,
+      "step": 1629
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.453757225433526e-05,
+      "loss": 6.1116,
+      "step": 1630
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4528901734104046e-05,
+      "loss": 5.8551,
+      "step": 1631
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4520231213872832e-05,
+      "loss": 6.0208,
+      "step": 1632
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4511560693641618e-05,
+      "loss": 6.2521,
+      "step": 1633
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4502890173410404e-05,
+      "loss": 6.2684,
+      "step": 1634
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4494219653179193e-05,
+      "loss": 6.0193,
+      "step": 1635
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.448554913294798e-05,
+      "loss": 5.909,
+      "step": 1636
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4476878612716765e-05,
+      "loss": 5.9076,
+      "step": 1637
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.446820809248555e-05,
+      "loss": 5.91,
+      "step": 1638
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4459537572254337e-05,
+      "loss": 5.7091,
+      "step": 1639
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.445086705202312e-05,
+      "loss": 5.7304,
+      "step": 1640
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4442196531791906e-05,
+      "loss": 5.8785,
+      "step": 1641
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4433526011560692e-05,
+      "loss": 5.242,
+      "step": 1642
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.442485549132948e-05,
+      "loss": 6.1853,
+      "step": 1643
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4416184971098268e-05,
+      "loss": 6.3568,
+      "step": 1644
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4407514450867054e-05,
+      "loss": 6.2157,
+      "step": 1645
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.439884393063584e-05,
+      "loss": 6.2217,
+      "step": 1646
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.4390173410404626e-05,
+      "loss": 6.1469,
+      "step": 1647
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.4381502890173412e-05,
+      "loss": 6.0103,
+      "step": 1648
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.4372832369942198e-05,
+      "loss": 6.1704,
+      "step": 1649
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.436416184971098e-05,
+      "loss": 6.3015,
+      "step": 1650
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.435549132947977e-05,
+      "loss": 6.0851,
+      "step": 1651
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.4346820809248556e-05,
+      "loss": 6.3174,
+      "step": 1652
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.4338150289017342e-05,
+      "loss": 6.0116,
+      "step": 1653
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.4329479768786128e-05,
+      "loss": 5.9518,
+      "step": 1654
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4320809248554914e-05,
+      "loss": 6.0891,
+      "step": 1655
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.43121387283237e-05,
+      "loss": 6.0793,
+      "step": 1656
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4303468208092486e-05,
+      "loss": 6.2418,
+      "step": 1657
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4294797687861272e-05,
+      "loss": 6.2892,
+      "step": 1658
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.428612716763006e-05,
+      "loss": 5.9556,
+      "step": 1659
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4277456647398844e-05,
+      "loss": 6.1276,
+      "step": 1660
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.426878612716763e-05,
+      "loss": 6.1602,
+      "step": 1661
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4260115606936416e-05,
+      "loss": 6.0259,
+      "step": 1662
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.4251445086705202e-05,
+      "loss": 6.022,
+      "step": 1663
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.4242774566473988e-05,
+      "loss": 6.0961,
+      "step": 1664
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.4234104046242774e-05,
+      "loss": 5.9328,
+      "step": 1665
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.422543352601156e-05,
+      "loss": 6.2524,
+      "step": 1666
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.421676300578035e-05,
+      "loss": 5.9299,
+      "step": 1667
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.4208092485549136e-05,
+      "loss": 6.1714,
+      "step": 1668
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.4199421965317922e-05,
+      "loss": 6.1323,
+      "step": 1669
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.4190751445086708e-05,
+      "loss": 6.0128,
+      "step": 1670
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.418208092485549e-05,
+      "loss": 6.0005,
+      "step": 1671
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.4173410404624277e-05,
+      "loss": 6.1021,
+      "step": 1672
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.4164739884393063e-05,
+      "loss": 6.2314,
+      "step": 1673
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.415606936416185e-05,
+      "loss": 6.0531,
+      "step": 1674
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.4147398843930635e-05,
+      "loss": 6.0675,
+      "step": 1675
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.4138728323699424e-05,
+      "loss": 6.1798,
+      "step": 1676
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.413005780346821e-05,
+      "loss": 6.0782,
+      "step": 1677
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.4121387283236996e-05,
+      "loss": 5.8234,
+      "step": 1678
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.4112716763005782e-05,
+      "loss": 6.2015,
+      "step": 1679
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.4104046242774568e-05,
+      "loss": 6.0358,
+      "step": 1680
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.409537572254335e-05,
+      "loss": 6.0631,
+      "step": 1681
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4086705202312137e-05,
+      "loss": 5.8634,
+      "step": 1682
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4078034682080923e-05,
+      "loss": 6.196,
+      "step": 1683
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4069364161849712e-05,
+      "loss": 5.8919,
+      "step": 1684
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.40606936416185e-05,
+      "loss": 5.7836,
+      "step": 1685
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4052023121387284e-05,
+      "loss": 5.87,
+      "step": 1686
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.404335260115607e-05,
+      "loss": 5.7961,
+      "step": 1687
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4034682080924856e-05,
+      "loss": 5.8297,
+      "step": 1688
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.4026011560693643e-05,
+      "loss": 5.8616,
+      "step": 1689
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.401734104046243e-05,
+      "loss": 5.8838,
+      "step": 1690
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.4008670520231215e-05,
+      "loss": 6.1637,
+      "step": 1691
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.4e-05,
+      "loss": 5.4243,
+      "step": 1692
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3991329479768787e-05,
+      "loss": 6.318,
+      "step": 1693
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3982658959537573e-05,
+      "loss": 6.2516,
+      "step": 1694
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.397398843930636e-05,
+      "loss": 6.2331,
+      "step": 1695
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3965317919075145e-05,
+      "loss": 6.2631,
+      "step": 1696
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.395664739884393e-05,
+      "loss": 6.1273,
+      "step": 1697
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3947976878612717e-05,
+      "loss": 6.1847,
+      "step": 1698
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3939306358381503e-05,
+      "loss": 6.1087,
+      "step": 1699
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3930635838150292e-05,
+      "loss": 6.0426,
+      "step": 1700
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3921965317919078e-05,
+      "loss": 6.0457,
+      "step": 1701
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.391329479768786e-05,
+      "loss": 6.0331,
+      "step": 1702
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3904624277456647e-05,
+      "loss": 6.1284,
+      "step": 1703
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3895953757225433e-05,
+      "loss": 6.1511,
+      "step": 1704
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.388728323699422e-05,
+      "loss": 6.0921,
+      "step": 1705
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3878612716763005e-05,
+      "loss": 5.8741,
+      "step": 1706
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.386994219653179e-05,
+      "loss": 6.0077,
+      "step": 1707
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.386127167630058e-05,
+      "loss": 6.0516,
+      "step": 1708
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3852601156069367e-05,
+      "loss": 6.1112,
+      "step": 1709
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3843930635838153e-05,
+      "loss": 6.0035,
+      "step": 1710
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.383526011560694e-05,
+      "loss": 5.9603,
+      "step": 1711
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.382658959537572e-05,
+      "loss": 6.0673,
+      "step": 1712
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3817919075144507e-05,
+      "loss": 6.0218,
+      "step": 1713
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3809248554913293e-05,
+      "loss": 5.9184,
+      "step": 1714
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.380057803468208e-05,
+      "loss": 6.0477,
+      "step": 1715
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.379190751445087e-05,
+      "loss": 6.1481,
+      "step": 1716
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3783236994219655e-05,
+      "loss": 6.2321,
+      "step": 1717
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.377456647398844e-05,
+      "loss": 6.0074,
+      "step": 1718
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3765895953757227e-05,
+      "loss": 5.9698,
+      "step": 1719
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3757225433526013e-05,
+      "loss": 5.8732,
+      "step": 1720
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.37485549132948e-05,
+      "loss": 6.1175,
+      "step": 1721
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3739884393063585e-05,
+      "loss": 6.0281,
+      "step": 1722
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3731213872832368e-05,
+      "loss": 5.9687,
+      "step": 1723
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3722543352601157e-05,
+      "loss": 6.0265,
+      "step": 1724
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3713872832369943e-05,
+      "loss": 6.0445,
+      "step": 1725
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.370520231213873e-05,
+      "loss": 5.8339,
+      "step": 1726
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3696531791907515e-05,
+      "loss": 6.0886,
+      "step": 1727
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.36878612716763e-05,
+      "loss": 6.0647,
+      "step": 1728
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3679190751445087e-05,
+      "loss": 5.8787,
+      "step": 1729
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3670520231213873e-05,
+      "loss": 6.0577,
+      "step": 1730
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.366184971098266e-05,
+      "loss": 5.8694,
+      "step": 1731
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.365317919075145e-05,
+      "loss": 6.2309,
+      "step": 1732
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.364450867052023e-05,
+      "loss": 5.7861,
+      "step": 1733
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3635838150289017e-05,
+      "loss": 5.9149,
+      "step": 1734
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3627167630057803e-05,
+      "loss": 5.8307,
+      "step": 1735
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.361849710982659e-05,
+      "loss": 6.0092,
+      "step": 1736
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3609826589595376e-05,
+      "loss": 5.8856,
+      "step": 1737
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.360115606936416e-05,
+      "loss": 5.8958,
+      "step": 1738
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3592485549132948e-05,
+      "loss": 5.8921,
+      "step": 1739
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3583815028901737e-05,
+      "loss": 5.7853,
+      "step": 1740
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3575144508670523e-05,
+      "loss": 5.6163,
+      "step": 1741
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.356647398843931e-05,
+      "loss": 5.3976,
+      "step": 1742
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3557803468208092e-05,
+      "loss": 6.1361,
+      "step": 1743
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3549132947976878e-05,
+      "loss": 6.0583,
+      "step": 1744
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3540462427745664e-05,
+      "loss": 6.306,
+      "step": 1745
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.353179190751445e-05,
+      "loss": 6.2121,
+      "step": 1746
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3523121387283236e-05,
+      "loss": 6.1305,
+      "step": 1747
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3514450867052025e-05,
+      "loss": 6.0678,
+      "step": 1748
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.350578034682081e-05,
+      "loss": 6.1038,
+      "step": 1749
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3497109826589597e-05,
+      "loss": 6.2775,
+      "step": 1750
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3488439306358383e-05,
+      "loss": 5.9493,
+      "step": 1751
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.347976878612717e-05,
+      "loss": 6.1138,
+      "step": 1752
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3471098265895955e-05,
+      "loss": 6.0905,
+      "step": 1753
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3462427745664738e-05,
+      "loss": 6.0448,
+      "step": 1754
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3453757225433524e-05,
+      "loss": 5.9364,
+      "step": 1755
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3445086705202314e-05,
+      "loss": 5.962,
+      "step": 1756
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.34364161849711e-05,
+      "loss": 6.1058,
+      "step": 1757
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3427745664739886e-05,
+      "loss": 6.0754,
+      "step": 1758
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.341907514450867e-05,
+      "loss": 6.0929,
+      "step": 1759
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3410404624277458e-05,
+      "loss": 5.8681,
+      "step": 1760
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3401734104046244e-05,
+      "loss": 6.177,
+      "step": 1761
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.339306358381503e-05,
+      "loss": 5.8874,
+      "step": 1762
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3384393063583816e-05,
+      "loss": 6.1065,
+      "step": 1763
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3375722543352602e-05,
+      "loss": 6.0586,
+      "step": 1764
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3367052023121388e-05,
+      "loss": 5.892,
+      "step": 1765
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3358381502890174e-05,
+      "loss": 6.0549,
+      "step": 1766
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.334971098265896e-05,
+      "loss": 5.9466,
+      "step": 1767
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3341040462427746e-05,
+      "loss": 6.0805,
+      "step": 1768
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3332369942196532e-05,
+      "loss": 5.8952,
+      "step": 1769
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3323699421965318e-05,
+      "loss": 6.1025,
+      "step": 1770
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3315028901734104e-05,
+      "loss": 5.9456,
+      "step": 1771
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.330635838150289e-05,
+      "loss": 5.9091,
+      "step": 1772
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.329768786127168e-05,
+      "loss": 5.9981,
+      "step": 1773
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3289017341040466e-05,
+      "loss": 6.0894,
+      "step": 1774
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3280346820809248e-05,
+      "loss": 5.9958,
+      "step": 1775
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3271676300578034e-05,
+      "loss": 5.9928,
+      "step": 1776
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.326300578034682e-05,
+      "loss": 6.086,
+      "step": 1777
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3254335260115606e-05,
+      "loss": 6.005,
+      "step": 1778
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3245664739884392e-05,
+      "loss": 5.9949,
+      "step": 1779
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.323699421965318e-05,
+      "loss": 5.7449,
+      "step": 1780
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3228323699421968e-05,
+      "loss": 5.7976,
+      "step": 1781
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3219653179190754e-05,
+      "loss": 6.0101,
+      "step": 1782
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.321098265895954e-05,
+      "loss": 5.7106,
+      "step": 1783
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3202312138728326e-05,
+      "loss": 5.0412,
+      "step": 1784
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.319364161849711e-05,
+      "loss": 6.1997,
+      "step": 1785
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3184971098265895e-05,
+      "loss": 5.8576,
+      "step": 1786
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.317630057803468e-05,
+      "loss": 6.0152,
+      "step": 1787
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3167630057803467e-05,
+      "loss": 6.0669,
+      "step": 1788
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.3158959537572256e-05,
+      "loss": 6.0345,
+      "step": 1789
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.3150289017341042e-05,
+      "loss": 5.9785,
+      "step": 1790
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.3141618497109828e-05,
+      "loss": 6.0689,
+      "step": 1791
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.3132947976878614e-05,
+      "loss": 6.0393,
+      "step": 1792
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.31242774566474e-05,
+      "loss": 5.9537,
+      "step": 1793
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.3115606936416186e-05,
+      "loss": 5.995,
+      "step": 1794
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.310693641618497e-05,
+      "loss": 6.0953,
+      "step": 1795
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.3098265895953755e-05,
+      "loss": 6.0639,
+      "step": 1796
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.3089595375722544e-05,
+      "loss": 5.994,
+      "step": 1797
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.308092485549133e-05,
+      "loss": 5.954,
+      "step": 1798
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.3072254335260116e-05,
+      "loss": 6.1009,
+      "step": 1799
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.3063583815028902e-05,
+      "loss": 5.9391,
+      "step": 1800
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.305491329479769e-05,
+      "loss": 6.0676,
+      "step": 1801
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.3046242774566475e-05,
+      "loss": 6.3286,
+      "step": 1802
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.303757225433526e-05,
+      "loss": 6.0184,
+      "step": 1803
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.3028901734104047e-05,
+      "loss": 5.9088,
+      "step": 1804
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.3020231213872836e-05,
+      "loss": 5.855,
+      "step": 1805
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.301156069364162e-05,
+      "loss": 5.9949,
+      "step": 1806
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.3002890173410405e-05,
+      "loss": 6.0598,
+      "step": 1807
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.299421965317919e-05,
+      "loss": 5.949,
+      "step": 1808
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2985549132947977e-05,
+      "loss": 5.9486,
+      "step": 1809
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2976878612716763e-05,
+      "loss": 5.9142,
+      "step": 1810
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.296820809248555e-05,
+      "loss": 5.8862,
+      "step": 1811
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2959537572254335e-05,
+      "loss": 5.964,
+      "step": 1812
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2950867052023124e-05,
+      "loss": 5.9529,
+      "step": 1813
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.294219653179191e-05,
+      "loss": 6.024,
+      "step": 1814
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2933526011560696e-05,
+      "loss": 6.08,
+      "step": 1815
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.292485549132948e-05,
+      "loss": 5.9283,
+      "step": 1816
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2916184971098265e-05,
+      "loss": 5.9323,
+      "step": 1817
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.290751445086705e-05,
+      "loss": 5.9051,
+      "step": 1818
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2898843930635837e-05,
+      "loss": 5.8555,
+      "step": 1819
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2890173410404623e-05,
+      "loss": 5.7074,
+      "step": 1820
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2881502890173413e-05,
+      "loss": 5.9101,
+      "step": 1821
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.28728323699422e-05,
+      "loss": 5.8857,
+      "step": 1822
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2864161849710985e-05,
+      "loss": 5.7911,
+      "step": 1823
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.285549132947977e-05,
+      "loss": 5.9378,
+      "step": 1824
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2846820809248557e-05,
+      "loss": 5.849,
+      "step": 1825
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2838150289017343e-05,
+      "loss": 6.0905,
+      "step": 1826
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2829479768786125e-05,
+      "loss": 5.8053,
+      "step": 1827
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.282080924855491e-05,
+      "loss": 5.9488,
+      "step": 1828
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.28121387283237e-05,
+      "loss": 5.822,
+      "step": 1829
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2803468208092487e-05,
+      "loss": 5.8367,
+      "step": 1830
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2794797687861273e-05,
+      "loss": 5.7213,
+      "step": 1831
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.278612716763006e-05,
+      "loss": 5.6425,
+      "step": 1832
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2777456647398845e-05,
+      "loss": 5.6403,
+      "step": 1833
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.276878612716763e-05,
+      "loss": 5.2444,
+      "step": 1834
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2760115606936417e-05,
+      "loss": 6.2691,
+      "step": 1835
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2751445086705203e-05,
+      "loss": 6.2306,
+      "step": 1836
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.274277456647399e-05,
+      "loss": 6.0027,
+      "step": 1837
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2734104046242775e-05,
+      "loss": 6.1211,
+      "step": 1838
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.272543352601156e-05,
+      "loss": 6.1946,
+      "step": 1839
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2716763005780347e-05,
+      "loss": 5.9293,
+      "step": 1840
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2708092485549133e-05,
+      "loss": 6.1424,
+      "step": 1841
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.269942196531792e-05,
+      "loss": 5.9558,
+      "step": 1842
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2690751445086705e-05,
+      "loss": 5.8676,
+      "step": 1843
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.268208092485549e-05,
+      "loss": 6.051,
+      "step": 1844
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.267341040462428e-05,
+      "loss": 6.0727,
+      "step": 1845
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2664739884393067e-05,
+      "loss": 5.9368,
+      "step": 1846
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.265606936416185e-05,
+      "loss": 5.9372,
+      "step": 1847
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2647398843930635e-05,
+      "loss": 5.887,
+      "step": 1848
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.263872832369942e-05,
+      "loss": 6.0,
+      "step": 1849
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2630057803468208e-05,
+      "loss": 5.8354,
+      "step": 1850
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2621387283236994e-05,
+      "loss": 6.0696,
+      "step": 1851
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.261271676300578e-05,
+      "loss": 5.833,
+      "step": 1852
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.260404624277457e-05,
+      "loss": 6.0862,
+      "step": 1853
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2595375722543355e-05,
+      "loss": 5.9928,
+      "step": 1854
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.258670520231214e-05,
+      "loss": 6.1273,
+      "step": 1855
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2578034682080927e-05,
+      "loss": 6.0471,
+      "step": 1856
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2569364161849713e-05,
+      "loss": 5.9945,
+      "step": 1857
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2560693641618496e-05,
+      "loss": 5.8566,
+      "step": 1858
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2552023121387282e-05,
+      "loss": 6.1048,
+      "step": 1859
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2543352601156068e-05,
+      "loss": 5.9595,
+      "step": 1860
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2534682080924857e-05,
+      "loss": 6.0401,
+      "step": 1861
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2526011560693643e-05,
+      "loss": 5.8838,
+      "step": 1862
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.251734104046243e-05,
+      "loss": 5.7783,
+      "step": 1863
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2508670520231215e-05,
+      "loss": 5.9749,
+      "step": 1864
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.25e-05,
+      "loss": 5.9778,
+      "step": 1865
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2491329479768787e-05,
+      "loss": 5.8208,
+      "step": 1866
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2482658959537573e-05,
+      "loss": 6.0445,
+      "step": 1867
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2473988439306356e-05,
+      "loss": 5.9748,
+      "step": 1868
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2465317919075142e-05,
+      "loss": 5.9954,
+      "step": 1869
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.245664739884393e-05,
+      "loss": 5.9181,
+      "step": 1870
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2447976878612718e-05,
+      "loss": 6.0676,
+      "step": 1871
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2439306358381504e-05,
+      "loss": 5.7806,
+      "step": 1872
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.243063583815029e-05,
+      "loss": 5.9643,
+      "step": 1873
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2421965317919076e-05,
+      "loss": 5.9513,
+      "step": 1874
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2413294797687862e-05,
+      "loss": 5.9618,
+      "step": 1875
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2404624277456648e-05,
+      "loss": 5.9681,
+      "step": 1876
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2395953757225434e-05,
+      "loss": 5.8339,
+      "step": 1877
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.238728323699422e-05,
+      "loss": 5.6443,
+      "step": 1878
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2378612716763006e-05,
+      "loss": 5.9384,
+      "step": 1879
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2369942196531792e-05,
+      "loss": 5.8022,
+      "step": 1880
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2361271676300578e-05,
+      "loss": 5.8219,
+      "step": 1881
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2352601156069364e-05,
+      "loss": 5.7292,
+      "step": 1882
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.234393063583815e-05,
+      "loss": 5.3261,
+      "step": 1883
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2335260115606936e-05,
+      "loss": 5.1694,
+      "step": 1884
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2326589595375722e-05,
+      "loss": 6.2365,
+      "step": 1885
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.231791907514451e-05,
+      "loss": 6.1719,
+      "step": 1886
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2309248554913298e-05,
+      "loss": 6.3344,
+      "step": 1887
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2300578034682084e-05,
+      "loss": 6.0927,
+      "step": 1888
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2291907514450866e-05,
+      "loss": 6.0633,
+      "step": 1889
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2283236994219652e-05,
+      "loss": 6.0745,
+      "step": 1890
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2274566473988438e-05,
+      "loss": 5.9517,
+      "step": 1891
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2265895953757224e-05,
+      "loss": 5.9697,
+      "step": 1892
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.225722543352601e-05,
+      "loss": 6.2258,
+      "step": 1893
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.22485549132948e-05,
+      "loss": 6.0948,
+      "step": 1894
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2239884393063586e-05,
+      "loss": 5.8789,
+      "step": 1895
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2231213872832372e-05,
+      "loss": 6.0556,
+      "step": 1896
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2222543352601158e-05,
+      "loss": 5.8369,
+      "step": 1897
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2213872832369944e-05,
+      "loss": 5.8581,
+      "step": 1898
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2205202312138727e-05,
+      "loss": 5.8789,
+      "step": 1899
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2196531791907513e-05,
+      "loss": 5.7218,
+      "step": 1900
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.21878612716763e-05,
+      "loss": 5.9668,
+      "step": 1901
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2179190751445088e-05,
+      "loss": 6.0625,
+      "step": 1902
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2170520231213874e-05,
+      "loss": 5.9506,
+      "step": 1903
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.216184971098266e-05,
+      "loss": 5.9871,
+      "step": 1904
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.2153179190751446e-05,
+      "loss": 5.8931,
+      "step": 1905
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.2144508670520232e-05,
+      "loss": 5.9675,
+      "step": 1906
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.2135838150289018e-05,
+      "loss": 5.9543,
+      "step": 1907
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.2127167630057804e-05,
+      "loss": 6.0368,
+      "step": 1908
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.211849710982659e-05,
+      "loss": 5.9505,
+      "step": 1909
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.2109826589595376e-05,
+      "loss": 5.981,
+      "step": 1910
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.2101156069364162e-05,
+      "loss": 5.9225,
+      "step": 1911
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.209248554913295e-05,
+      "loss": 6.0652,
+      "step": 1912
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.2083815028901734e-05,
+      "loss": 6.0566,
+      "step": 1913
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.207514450867052e-05,
+      "loss": 5.8935,
+      "step": 1914
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.2066473988439307e-05,
+      "loss": 5.8476,
+      "step": 1915
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.2057803468208093e-05,
+      "loss": 5.9579,
+      "step": 1916
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.204913294797688e-05,
+      "loss": 5.7938,
+      "step": 1917
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.2040462427745668e-05,
+      "loss": 5.9244,
+      "step": 1918
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.2031791907514454e-05,
+      "loss": 5.955,
+      "step": 1919
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.2023121387283237e-05,
+      "loss": 6.0376,
+      "step": 1920
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.2014450867052023e-05,
+      "loss": 5.7398,
+      "step": 1921
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.200578034682081e-05,
+      "loss": 5.8729,
+      "step": 1922
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.1997109826589595e-05,
+      "loss": 5.8118,
+      "step": 1923
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.198843930635838e-05,
+      "loss": 5.7015,
+      "step": 1924
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.1979768786127167e-05,
+      "loss": 6.0077,
+      "step": 1925
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.1971098265895956e-05,
+      "loss": 5.8205,
+      "step": 1926
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.1962427745664742e-05,
+      "loss": 5.6056,
+      "step": 1927
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.195375722543353e-05,
+      "loss": 5.5818,
+      "step": 1928
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.1945086705202314e-05,
+      "loss": 5.5976,
+      "step": 1929
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.1936416184971097e-05,
+      "loss": 5.7121,
+      "step": 1930
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.1927745664739883e-05,
+      "loss": 5.5727,
+      "step": 1931
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.191907514450867e-05,
+      "loss": 5.3392,
+      "step": 1932
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.1910404624277455e-05,
+      "loss": 5.6283,
+      "step": 1933
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.1901734104046245e-05,
+      "loss": 5.4542,
+      "step": 1934
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.189306358381503e-05,
+      "loss": 6.3369,
+      "step": 1935
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.1884393063583817e-05,
+      "loss": 6.2307,
+      "step": 1936
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.1875722543352603e-05,
+      "loss": 6.0914,
+      "step": 1937
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.186705202312139e-05,
+      "loss": 5.9348,
+      "step": 1938
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.1858381502890175e-05,
+      "loss": 6.083,
+      "step": 1939
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.184971098265896e-05,
+      "loss": 6.2494,
+      "step": 1940
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.1841040462427743e-05,
+      "loss": 6.0797,
+      "step": 1941
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.1832369942196533e-05,
+      "loss": 5.9635,
+      "step": 1942
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.182369942196532e-05,
+      "loss": 5.9624,
+      "step": 1943
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.1815028901734105e-05,
+      "loss": 5.9166,
+      "step": 1944
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.180635838150289e-05,
+      "loss": 6.2142,
+      "step": 1945
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.1797687861271677e-05,
+      "loss": 5.8961,
+      "step": 1946
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.1789017341040463e-05,
+      "loss": 5.8969,
+      "step": 1947
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.178034682080925e-05,
+      "loss": 5.8371,
+      "step": 1948
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.1771676300578035e-05,
+      "loss": 6.1984,
+      "step": 1949
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.1763005780346824e-05,
+      "loss": 6.0205,
+      "step": 1950
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.1754335260115607e-05,
+      "loss": 5.8343,
+      "step": 1951
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.1745664739884393e-05,
+      "loss": 5.8176,
+      "step": 1952
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.173699421965318e-05,
+      "loss": 5.9144,
+      "step": 1953
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.1728323699421965e-05,
+      "loss": 6.1221,
+      "step": 1954
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.171965317919075e-05,
+      "loss": 5.9993,
+      "step": 1955
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.1710982658959537e-05,
+      "loss": 6.0096,
+      "step": 1956
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.1702312138728323e-05,
+      "loss": 5.9266,
+      "step": 1957
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.1693641618497113e-05,
+      "loss": 5.9057,
+      "step": 1958
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.16849710982659e-05,
+      "loss": 5.845,
+      "step": 1959
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.1676300578034685e-05,
+      "loss": 6.0251,
+      "step": 1960
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.166763005780347e-05,
+      "loss": 5.7998,
+      "step": 1961
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.1658959537572253e-05,
+      "loss": 5.7912,
+      "step": 1962
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.165028901734104e-05,
+      "loss": 5.8884,
+      "step": 1963
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.1641618497109826e-05,
+      "loss": 5.8037,
+      "step": 1964
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.163294797687861e-05,
+      "loss": 5.7983,
+      "step": 1965
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.1624277456647398e-05,
+      "loss": 5.8225,
+      "step": 1966
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.1615606936416187e-05,
+      "loss": 5.7649,
+      "step": 1967
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.1606936416184973e-05,
+      "loss": 5.9422,
+      "step": 1968
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.159826589595376e-05,
+      "loss": 5.988,
+      "step": 1969
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.1589595375722545e-05,
+      "loss": 5.7519,
+      "step": 1970
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.158092485549133e-05,
+      "loss": 5.8809,
+      "step": 1971
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.1572254335260114e-05,
+      "loss": 5.833,
+      "step": 1972
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.15635838150289e-05,
+      "loss": 5.7192,
+      "step": 1973
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.1554913294797686e-05,
+      "loss": 5.8725,
+      "step": 1974
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.1546242774566475e-05,
+      "loss": 5.8712,
+      "step": 1975
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.153757225433526e-05,
+      "loss": 5.7387,
+      "step": 1976
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.1528901734104047e-05,
+      "loss": 5.6256,
+      "step": 1977
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.1520231213872833e-05,
+      "loss": 5.8312,
+      "step": 1978
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.151156069364162e-05,
+      "loss": 5.4818,
+      "step": 1979
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.1502890173410405e-05,
+      "loss": 5.6519,
+      "step": 1980
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.149421965317919e-05,
+      "loss": 5.7099,
+      "step": 1981
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.1485549132947974e-05,
+      "loss": 5.6704,
+      "step": 1982
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.1476878612716764e-05,
+      "loss": 5.5509,
+      "step": 1983
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.146820809248555e-05,
+      "loss": 5.1604,
+      "step": 1984
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.1459537572254336e-05,
+      "loss": 6.2449,
+      "step": 1985
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.145086705202312e-05,
+      "loss": 6.2221,
+      "step": 1986
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.1442196531791908e-05,
+      "loss": 6.1787,
+      "step": 1987
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.1433526011560694e-05,
+      "loss": 5.9005,
+      "step": 1988
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.142485549132948e-05,
+      "loss": 6.0299,
+      "step": 1989
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.1416184971098266e-05,
+      "loss": 6.0828,
+      "step": 1990
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.1407514450867055e-05,
+      "loss": 6.065,
+      "step": 1991
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.139884393063584e-05,
+      "loss": 6.118,
+      "step": 1992
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.1390173410404624e-05,
+      "loss": 5.8771,
+      "step": 1993
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.138150289017341e-05,
+      "loss": 6.005,
+      "step": 1994
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.1372832369942196e-05,
+      "loss": 6.1019,
+      "step": 1995
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.1364161849710982e-05,
+      "loss": 5.818,
+      "step": 1996
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.1355491329479768e-05,
+      "loss": 5.8789,
+      "step": 1997
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.1346820809248554e-05,
+      "loss": 5.7809,
+      "step": 1998
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.1338150289017344e-05,
+      "loss": 6.0349,
+      "step": 1999
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.132947976878613e-05,
+      "loss": 5.8035,
+      "step": 2000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.1320809248554916e-05,
+      "loss": 5.8684,
+      "step": 2001
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.13121387283237e-05,
+      "loss": 5.7261,
+      "step": 2002
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.1303468208092484e-05,
+      "loss": 5.8487,
+      "step": 2003
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.129479768786127e-05,
+      "loss": 5.9747,
+      "step": 2004
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.1286127167630056e-05,
+      "loss": 5.8411,
+      "step": 2005
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.1277456647398842e-05,
+      "loss": 6.0872,
+      "step": 2006
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.1268786127167632e-05,
+      "loss": 5.8153,
+      "step": 2007
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.1260115606936418e-05,
+      "loss": 5.9209,
+      "step": 2008
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.1251445086705204e-05,
+      "loss": 5.958,
+      "step": 2009
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.124277456647399e-05,
+      "loss": 5.8944,
+      "step": 2010
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.1234104046242776e-05,
+      "loss": 5.9233,
+      "step": 2011
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.1225433526011562e-05,
+      "loss": 5.8158,
+      "step": 2012
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.1216763005780348e-05,
+      "loss": 5.8436,
+      "step": 2013
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.120809248554913e-05,
+      "loss": 5.9657,
+      "step": 2014
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.119942196531792e-05,
+      "loss": 5.9226,
+      "step": 2015
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.1190751445086706e-05,
+      "loss": 5.8653,
+      "step": 2016
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.1182080924855492e-05,
+      "loss": 5.984,
+      "step": 2017
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.1173410404624278e-05,
+      "loss": 5.8075,
+      "step": 2018
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.1164739884393064e-05,
+      "loss": 6.0144,
+      "step": 2019
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.115606936416185e-05,
+      "loss": 5.9897,
+      "step": 2020
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.1147398843930636e-05,
+      "loss": 5.8643,
+      "step": 2021
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.1138728323699422e-05,
+      "loss": 5.9281,
+      "step": 2022
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.1130057803468212e-05,
+      "loss": 5.8625,
+      "step": 2023
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.1121387283236994e-05,
+      "loss": 5.7755,
+      "step": 2024
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.111271676300578e-05,
+      "loss": 5.742,
+      "step": 2025
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.1104046242774566e-05,
+      "loss": 5.8053,
+      "step": 2026
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.1095375722543352e-05,
+      "loss": 5.8671,
+      "step": 2027
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.108670520231214e-05,
+      "loss": 5.6495,
+      "step": 2028
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.1078034682080925e-05,
+      "loss": 5.6395,
+      "step": 2029
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.106936416184971e-05,
+      "loss": 5.833,
+      "step": 2030
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.10606936416185e-05,
+      "loss": 5.3257,
+      "step": 2031
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.1052023121387286e-05,
+      "loss": 5.6651,
+      "step": 2032
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.1043352601156072e-05,
+      "loss": 5.3535,
+      "step": 2033
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.1034682080924855e-05,
+      "loss": 5.6291,
+      "step": 2034
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.102601156069364e-05,
+      "loss": 6.2038,
+      "step": 2035
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.1017341040462427e-05,
+      "loss": 6.1366,
+      "step": 2036
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.1008670520231213e-05,
+      "loss": 6.1731,
+      "step": 2037
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.1e-05,
+      "loss": 6.1088,
+      "step": 2038
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.0991329479768788e-05,
+      "loss": 5.9792,
+      "step": 2039
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.0982658959537574e-05,
+      "loss": 6.1875,
+      "step": 2040
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.097398843930636e-05,
+      "loss": 5.9442,
+      "step": 2041
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.0965317919075146e-05,
+      "loss": 6.1357,
+      "step": 2042
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.0956647398843932e-05,
+      "loss": 6.0647,
+      "step": 2043
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.094797687861272e-05,
+      "loss": 5.801,
+      "step": 2044
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.09393063583815e-05,
+      "loss": 5.9849,
+      "step": 2045
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.0930635838150287e-05,
+      "loss": 5.9166,
+      "step": 2046
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.0921965317919077e-05,
+      "loss": 6.1707,
+      "step": 2047
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.0913294797687863e-05,
+      "loss": 5.76,
+      "step": 2048
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.090462427745665e-05,
+      "loss": 6.1633,
+      "step": 2049
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.0895953757225435e-05,
+      "loss": 5.9143,
+      "step": 2050
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.088728323699422e-05,
+      "loss": 5.9914,
+      "step": 2051
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.0878612716763007e-05,
+      "loss": 5.8259,
+      "step": 2052
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.0869942196531793e-05,
+      "loss": 5.9472,
+      "step": 2053
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.086127167630058e-05,
+      "loss": 5.901,
+      "step": 2054
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.0852601156069365e-05,
+      "loss": 5.9352,
+      "step": 2055
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.084393063583815e-05,
+      "loss": 5.7859,
+      "step": 2056
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.0835260115606937e-05,
+      "loss": 5.8712,
+      "step": 2057
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.0826589595375723e-05,
+      "loss": 5.7092,
+      "step": 2058
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.081791907514451e-05,
+      "loss": 6.0655,
+      "step": 2059
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.0809248554913295e-05,
+      "loss": 5.8968,
+      "step": 2060
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.080057803468208e-05,
+      "loss": 6.1585,
+      "step": 2061
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.0791907514450867e-05,
+      "loss": 5.96,
+      "step": 2062
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.0783236994219653e-05,
+      "loss": 5.7832,
+      "step": 2063
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.0774566473988442e-05,
+      "loss": 6.1088,
+      "step": 2064
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.0765895953757225e-05,
+      "loss": 5.9136,
+      "step": 2065
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.075722543352601e-05,
+      "loss": 5.65,
+      "step": 2066
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.0748554913294797e-05,
+      "loss": 5.8189,
+      "step": 2067
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.0739884393063583e-05,
+      "loss": 5.7784,
+      "step": 2068
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.073121387283237e-05,
+      "loss": 5.921,
+      "step": 2069
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.0722543352601155e-05,
+      "loss": 5.8289,
+      "step": 2070
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.071387283236994e-05,
+      "loss": 5.7291,
+      "step": 2071
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.070520231213873e-05,
+      "loss": 5.8614,
+      "step": 2072
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.0696531791907517e-05,
+      "loss": 5.6563,
+      "step": 2073
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0687861271676303e-05,
+      "loss": 5.8697,
+      "step": 2074
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.067919075144509e-05,
+      "loss": 5.8316,
+      "step": 2075
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.067052023121387e-05,
+      "loss": 5.778,
+      "step": 2076
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0661849710982658e-05,
+      "loss": 5.7579,
+      "step": 2077
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0653179190751444e-05,
+      "loss": 5.8215,
+      "step": 2078
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.064450867052023e-05,
+      "loss": 5.4621,
+      "step": 2079
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.063583815028902e-05,
+      "loss": 5.7348,
+      "step": 2080
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.0627167630057805e-05,
+      "loss": 5.3838,
+      "step": 2081
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.061849710982659e-05,
+      "loss": 5.3373,
+      "step": 2082
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0609826589595377e-05,
+      "loss": 5.2815,
+      "step": 2083
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0601156069364163e-05,
+      "loss": 5.2655,
+      "step": 2084
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.059248554913295e-05,
+      "loss": 6.1351,
+      "step": 2085
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0583815028901732e-05,
+      "loss": 6.1697,
+      "step": 2086
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0575144508670518e-05,
+      "loss": 6.1794,
+      "step": 2087
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0566473988439307e-05,
+      "loss": 6.0535,
+      "step": 2088
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0557803468208093e-05,
+      "loss": 6.0463,
+      "step": 2089
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.054913294797688e-05,
+      "loss": 5.9763,
+      "step": 2090
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.0540462427745665e-05,
+      "loss": 6.0339,
+      "step": 2091
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.053179190751445e-05,
+      "loss": 5.9118,
+      "step": 2092
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.0523121387283237e-05,
+      "loss": 6.0155,
+      "step": 2093
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.0514450867052024e-05,
+      "loss": 6.2152,
+      "step": 2094
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.050578034682081e-05,
+      "loss": 5.9135,
+      "step": 2095
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.04971098265896e-05,
+      "loss": 6.0803,
+      "step": 2096
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.048843930635838e-05,
+      "loss": 6.2101,
+      "step": 2097
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.0479768786127168e-05,
+      "loss": 6.16,
+      "step": 2098
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.0471098265895954e-05,
+      "loss": 5.884,
+      "step": 2099
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.046242774566474e-05,
+      "loss": 5.9472,
+      "step": 2100
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.0453757225433526e-05,
+      "loss": 5.7303,
+      "step": 2101
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.0445086705202312e-05,
+      "loss": 6.076,
+      "step": 2102
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.0436416184971098e-05,
+      "loss": 6.0308,
+      "step": 2103
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.0427745664739887e-05,
+      "loss": 5.8384,
+      "step": 2104
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.0419075144508673e-05,
+      "loss": 6.0024,
+      "step": 2105
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.041040462427746e-05,
+      "loss": 5.9412,
+      "step": 2106
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.0401734104046242e-05,
+      "loss": 5.737,
+      "step": 2107
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.0393063583815028e-05,
+      "loss": 5.8987,
+      "step": 2108
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.0384393063583814e-05,
+      "loss": 5.867,
+      "step": 2109
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.03757225433526e-05,
+      "loss": 5.8394,
+      "step": 2110
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.0367052023121386e-05,
+      "loss": 5.9723,
+      "step": 2111
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.0358381502890176e-05,
+      "loss": 5.7589,
+      "step": 2112
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.034971098265896e-05,
+      "loss": 5.8837,
+      "step": 2113
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.0341040462427748e-05,
+      "loss": 5.9728,
+      "step": 2114
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.0332369942196534e-05,
+      "loss": 5.7801,
+      "step": 2115
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.032369942196532e-05,
+      "loss": 5.8347,
+      "step": 2116
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.0315028901734102e-05,
+      "loss": 5.6843,
+      "step": 2117
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.030635838150289e-05,
+      "loss": 5.8334,
+      "step": 2118
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0297687861271674e-05,
+      "loss": 5.8607,
+      "step": 2119
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0289017341040464e-05,
+      "loss": 5.9343,
+      "step": 2120
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.028034682080925e-05,
+      "loss": 5.7534,
+      "step": 2121
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0271676300578036e-05,
+      "loss": 5.873,
+      "step": 2122
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0263005780346822e-05,
+      "loss": 5.7988,
+      "step": 2123
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0254335260115608e-05,
+      "loss": 5.8986,
+      "step": 2124
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0245664739884394e-05,
+      "loss": 5.8339,
+      "step": 2125
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.023699421965318e-05,
+      "loss": 5.7647,
+      "step": 2126
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0228323699421966e-05,
+      "loss": 5.8488,
+      "step": 2127
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0219653179190752e-05,
+      "loss": 5.5449,
+      "step": 2128
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0210982658959538e-05,
+      "loss": 5.6942,
+      "step": 2129
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0202312138728324e-05,
+      "loss": 5.5734,
+      "step": 2130
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.019364161849711e-05,
+      "loss": 5.5049,
+      "step": 2131
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0184971098265896e-05,
+      "loss": 5.4605,
+      "step": 2132
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0176300578034682e-05,
+      "loss": 5.6038,
+      "step": 2133
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0167630057803468e-05,
+      "loss": 5.3695,
+      "step": 2134
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0158959537572254e-05,
+      "loss": 6.1349,
+      "step": 2135
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0150289017341044e-05,
+      "loss": 6.2053,
+      "step": 2136
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.014161849710983e-05,
+      "loss": 5.9756,
+      "step": 2137
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0132947976878612e-05,
+      "loss": 5.9766,
+      "step": 2138
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.01242774566474e-05,
+      "loss": 6.0201,
+      "step": 2139
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0115606936416184e-05,
+      "loss": 5.9466,
+      "step": 2140
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.010693641618497e-05,
+      "loss": 5.7465,
+      "step": 2141
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0098265895953757e-05,
+      "loss": 6.1457,
+      "step": 2142
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0089595375722543e-05,
+      "loss": 6.1158,
+      "step": 2143
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0080924855491332e-05,
+      "loss": 5.7221,
+      "step": 2144
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0072254335260118e-05,
+      "loss": 5.7866,
+      "step": 2145
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0063583815028904e-05,
+      "loss": 5.931,
+      "step": 2146
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.005491329479769e-05,
+      "loss": 5.9441,
+      "step": 2147
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0046242774566476e-05,
+      "loss": 5.9144,
+      "step": 2148
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.003757225433526e-05,
+      "loss": 5.9278,
+      "step": 2149
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0028901734104045e-05,
+      "loss": 6.0181,
+      "step": 2150
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.002023121387283e-05,
+      "loss": 5.9665,
+      "step": 2151
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.001156069364162e-05,
+      "loss": 6.0181,
+      "step": 2152
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 2.0002890173410406e-05,
+      "loss": 5.7878,
+      "step": 2153
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9994219653179192e-05,
+      "loss": 5.9854,
+      "step": 2154
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.998554913294798e-05,
+      "loss": 6.0488,
+      "step": 2155
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9976878612716764e-05,
+      "loss": 5.7768,
+      "step": 2156
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.996820809248555e-05,
+      "loss": 5.7394,
+      "step": 2157
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9959537572254336e-05,
+      "loss": 5.7361,
+      "step": 2158
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.995086705202312e-05,
+      "loss": 5.7292,
+      "step": 2159
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9942196531791905e-05,
+      "loss": 6.049,
+      "step": 2160
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9933526011560695e-05,
+      "loss": 5.8684,
+      "step": 2161
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.992485549132948e-05,
+      "loss": 5.6858,
+      "step": 2162
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9916184971098267e-05,
+      "loss": 5.8927,
+      "step": 2163
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9907514450867053e-05,
+      "loss": 5.8511,
+      "step": 2164
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.989884393063584e-05,
+      "loss": 5.9786,
+      "step": 2165
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9890173410404625e-05,
+      "loss": 5.7265,
+      "step": 2166
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.988150289017341e-05,
+      "loss": 5.8258,
+      "step": 2167
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9872832369942197e-05,
+      "loss": 5.8309,
+      "step": 2168
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9864161849710983e-05,
+      "loss": 5.7509,
+      "step": 2169
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.985549132947977e-05,
+      "loss": 5.8656,
+      "step": 2170
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9846820809248555e-05,
+      "loss": 6.0665,
+      "step": 2171
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.983815028901734e-05,
+      "loss": 5.8109,
+      "step": 2172
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9829479768786127e-05,
+      "loss": 5.8046,
+      "step": 2173
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9820809248554913e-05,
+      "loss": 5.7199,
+      "step": 2174
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.98121387283237e-05,
+      "loss": 5.7812,
+      "step": 2175
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9803468208092485e-05,
+      "loss": 5.7894,
+      "step": 2176
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9794797687861274e-05,
+      "loss": 6.0573,
+      "step": 2177
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.978612716763006e-05,
+      "loss": 6.0109,
+      "step": 2178
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.9777456647398847e-05,
+      "loss": 5.7073,
+      "step": 2179
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.976878612716763e-05,
+      "loss": 5.6908,
+      "step": 2180
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9760115606936415e-05,
+      "loss": 5.4984,
+      "step": 2181
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.97514450867052e-05,
+      "loss": 5.8105,
+      "step": 2182
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9742774566473987e-05,
+      "loss": 5.2189,
+      "step": 2183
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9734104046242773e-05,
+      "loss": 5.2114,
+      "step": 2184
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9725433526011563e-05,
+      "loss": 6.1406,
+      "step": 2185
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.971676300578035e-05,
+      "loss": 6.0496,
+      "step": 2186
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9708092485549135e-05,
+      "loss": 6.0855,
+      "step": 2187
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.969942196531792e-05,
+      "loss": 6.0236,
+      "step": 2188
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.9690751445086707e-05,
+      "loss": 5.9359,
+      "step": 2189
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.968208092485549e-05,
+      "loss": 5.7806,
+      "step": 2190
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9673410404624276e-05,
+      "loss": 6.0562,
+      "step": 2191
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.966473988439306e-05,
+      "loss": 5.9853,
+      "step": 2192
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.965606936416185e-05,
+      "loss": 6.0157,
+      "step": 2193
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9647398843930637e-05,
+      "loss": 5.9844,
+      "step": 2194
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9638728323699423e-05,
+      "loss": 5.9106,
+      "step": 2195
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.963005780346821e-05,
+      "loss": 5.8019,
+      "step": 2196
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.9621387283236995e-05,
+      "loss": 5.8265,
+      "step": 2197
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.961271676300578e-05,
+      "loss": 5.7261,
+      "step": 2198
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9604046242774567e-05,
+      "loss": 5.928,
+      "step": 2199
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9595375722543353e-05,
+      "loss": 5.9904,
+      "step": 2200
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.958670520231214e-05,
+      "loss": 5.909,
+      "step": 2201
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9578034682080925e-05,
+      "loss": 5.7943,
+      "step": 2202
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.956936416184971e-05,
+      "loss": 5.7112,
+      "step": 2203
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9560693641618497e-05,
+      "loss": 5.9078,
+      "step": 2204
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9552023121387283e-05,
+      "loss": 5.8801,
+      "step": 2205
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.954335260115607e-05,
+      "loss": 5.8273,
+      "step": 2206
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.9534682080924856e-05,
+      "loss": 5.7893,
+      "step": 2207
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.952601156069364e-05,
+      "loss": 5.9664,
+      "step": 2208
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.951734104046243e-05,
+      "loss": 5.8679,
+      "step": 2209
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9508670520231217e-05,
+      "loss": 5.7122,
+      "step": 2210
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.95e-05,
+      "loss": 5.9805,
+      "step": 2211
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9491329479768786e-05,
+      "loss": 5.8587,
+      "step": 2212
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9482658959537572e-05,
+      "loss": 6.0415,
+      "step": 2213
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9473988439306358e-05,
+      "loss": 5.9353,
+      "step": 2214
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9465317919075144e-05,
+      "loss": 6.0354,
+      "step": 2215
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.945664739884393e-05,
+      "loss": 5.8155,
+      "step": 2216
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.944797687861272e-05,
+      "loss": 5.7355,
+      "step": 2217
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9439306358381505e-05,
+      "loss": 5.7742,
+      "step": 2218
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.943063583815029e-05,
+      "loss": 5.6955,
+      "step": 2219
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9421965317919077e-05,
+      "loss": 5.9375,
+      "step": 2220
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.941329479768786e-05,
+      "loss": 5.8294,
+      "step": 2221
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9404624277456646e-05,
+      "loss": 5.7937,
+      "step": 2222
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9395953757225432e-05,
+      "loss": 6.1219,
+      "step": 2223
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9387283236994218e-05,
+      "loss": 5.7679,
+      "step": 2224
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.9378612716763008e-05,
+      "loss": 5.7906,
+      "step": 2225
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9369942196531794e-05,
+      "loss": 5.7577,
+      "step": 2226
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.936127167630058e-05,
+      "loss": 5.7737,
+      "step": 2227
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9352601156069366e-05,
+      "loss": 5.4421,
+      "step": 2228
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.934393063583815e-05,
+      "loss": 5.5939,
+      "step": 2229
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9335260115606938e-05,
+      "loss": 5.561,
+      "step": 2230
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9326589595375724e-05,
+      "loss": 5.583,
+      "step": 2231
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9317919075144506e-05,
+      "loss": 5.6607,
+      "step": 2232
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9309248554913296e-05,
+      "loss": 5.6732,
+      "step": 2233
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9300578034682082e-05,
+      "loss": 5.2681,
+      "step": 2234
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9291907514450868e-05,
+      "loss": 6.2231,
+      "step": 2235
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9283236994219654e-05,
+      "loss": 6.1439,
+      "step": 2236
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.927456647398844e-05,
+      "loss": 6.2924,
+      "step": 2237
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9265895953757226e-05,
+      "loss": 6.3126,
+      "step": 2238
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9257225433526012e-05,
+      "loss": 5.964,
+      "step": 2239
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9248554913294798e-05,
+      "loss": 5.978,
+      "step": 2240
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9239884393063587e-05,
+      "loss": 5.914,
+      "step": 2241
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.923121387283237e-05,
+      "loss": 5.9158,
+      "step": 2242
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.9222543352601156e-05,
+      "loss": 6.0084,
+      "step": 2243
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9213872832369942e-05,
+      "loss": 5.9381,
+      "step": 2244
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9205202312138728e-05,
+      "loss": 5.8511,
+      "step": 2245
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9196531791907514e-05,
+      "loss": 5.8697,
+      "step": 2246
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.91878612716763e-05,
+      "loss": 5.8746,
+      "step": 2247
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9179190751445086e-05,
+      "loss": 6.0981,
+      "step": 2248
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9170520231213876e-05,
+      "loss": 5.8153,
+      "step": 2249
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9161849710982662e-05,
+      "loss": 5.6926,
+      "step": 2250
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9153179190751448e-05,
+      "loss": 5.7182,
+      "step": 2251
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.9144508670520234e-05,
+      "loss": 6.0182,
+      "step": 2252
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9135838150289016e-05,
+      "loss": 5.7788,
+      "step": 2253
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9127167630057802e-05,
+      "loss": 5.9277,
+      "step": 2254
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.911849710982659e-05,
+      "loss": 5.8535,
+      "step": 2255
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9109826589595375e-05,
+      "loss": 5.8902,
+      "step": 2256
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.910115606936416e-05,
+      "loss": 5.7086,
+      "step": 2257
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.909248554913295e-05,
+      "loss": 5.854,
+      "step": 2258
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9083815028901736e-05,
+      "loss": 5.8853,
+      "step": 2259
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9075144508670522e-05,
+      "loss": 5.9381,
+      "step": 2260
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.9066473988439308e-05,
+      "loss": 5.7999,
+      "step": 2261
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9057803468208094e-05,
+      "loss": 5.8425,
+      "step": 2262
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9049132947976877e-05,
+      "loss": 5.9574,
+      "step": 2263
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9040462427745663e-05,
+      "loss": 5.9362,
+      "step": 2264
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.903179190751445e-05,
+      "loss": 5.8513,
+      "step": 2265
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9023121387283238e-05,
+      "loss": 5.7271,
+      "step": 2266
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9014450867052024e-05,
+      "loss": 5.9644,
+      "step": 2267
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.900578034682081e-05,
+      "loss": 5.898,
+      "step": 2268
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.8997109826589596e-05,
+      "loss": 5.9778,
+      "step": 2269
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.8988439306358382e-05,
+      "loss": 5.8847,
+      "step": 2270
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.897976878612717e-05,
+      "loss": 5.8235,
+      "step": 2271
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.8971098265895954e-05,
+      "loss": 5.6186,
+      "step": 2272
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.8962427745664737e-05,
+      "loss": 5.8497,
+      "step": 2273
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.8953757225433527e-05,
+      "loss": 5.8187,
+      "step": 2274
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.8945086705202313e-05,
+      "loss": 5.7876,
+      "step": 2275
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.89364161849711e-05,
+      "loss": 5.7189,
+      "step": 2276
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.8927745664739885e-05,
+      "loss": 5.7783,
+      "step": 2277
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.891907514450867e-05,
+      "loss": 5.6498,
+      "step": 2278
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.8910404624277457e-05,
+      "loss": 5.9379,
+      "step": 2279
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.8901734104046243e-05,
+      "loss": 5.637,
+      "step": 2280
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.889306358381503e-05,
+      "loss": 5.4664,
+      "step": 2281
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.8884393063583818e-05,
+      "loss": 5.6196,
+      "step": 2282
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.8875722543352604e-05,
+      "loss": 5.3326,
+      "step": 2283
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.8867052023121387e-05,
+      "loss": 5.1158,
+      "step": 2284
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.8858381502890173e-05,
+      "loss": 6.2264,
+      "step": 2285
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.884971098265896e-05,
+      "loss": 5.9113,
+      "step": 2286
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.8841040462427745e-05,
+      "loss": 5.9849,
+      "step": 2287
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.883236994219653e-05,
+      "loss": 6.066,
+      "step": 2288
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8823699421965317e-05,
+      "loss": 5.8755,
+      "step": 2289
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8815028901734106e-05,
+      "loss": 5.995,
+      "step": 2290
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8806358381502893e-05,
+      "loss": 5.9367,
+      "step": 2291
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.879768786127168e-05,
+      "loss": 5.9388,
+      "step": 2292
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8789017341040465e-05,
+      "loss": 5.8545,
+      "step": 2293
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8780346820809247e-05,
+      "loss": 5.8932,
+      "step": 2294
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.8771676300578033e-05,
+      "loss": 5.8832,
+      "step": 2295
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.876300578034682e-05,
+      "loss": 6.0522,
+      "step": 2296
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8754335260115605e-05,
+      "loss": 6.0445,
+      "step": 2297
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8745664739884395e-05,
+      "loss": 5.8901,
+      "step": 2298
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.873699421965318e-05,
+      "loss": 5.9049,
+      "step": 2299
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8728323699421967e-05,
+      "loss": 5.7732,
+      "step": 2300
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8719653179190753e-05,
+      "loss": 5.9032,
+      "step": 2301
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.871098265895954e-05,
+      "loss": 5.7948,
+      "step": 2302
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8702312138728325e-05,
+      "loss": 5.8802,
+      "step": 2303
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.869364161849711e-05,
+      "loss": 5.9257,
+      "step": 2304
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.8684971098265894e-05,
+      "loss": 5.6379,
+      "step": 2305
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8676300578034683e-05,
+      "loss": 5.7577,
+      "step": 2306
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.866763005780347e-05,
+      "loss": 5.8644,
+      "step": 2307
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8658959537572255e-05,
+      "loss": 5.9362,
+      "step": 2308
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.865028901734104e-05,
+      "loss": 5.989,
+      "step": 2309
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8641618497109827e-05,
+      "loss": 5.9142,
+      "step": 2310
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8632947976878613e-05,
+      "loss": 5.7755,
+      "step": 2311
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.86242774566474e-05,
+      "loss": 5.8638,
+      "step": 2312
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8615606936416185e-05,
+      "loss": 5.8418,
+      "step": 2313
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.8606936416184975e-05,
+      "loss": 5.9023,
+      "step": 2314
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8598265895953757e-05,
+      "loss": 5.6851,
+      "step": 2315
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8589595375722543e-05,
+      "loss": 5.6745,
+      "step": 2316
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.858092485549133e-05,
+      "loss": 5.7629,
+      "step": 2317
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8572254335260115e-05,
+      "loss": 5.8477,
+      "step": 2318
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.85635838150289e-05,
+      "loss": 5.8884,
+      "step": 2319
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8554913294797688e-05,
+      "loss": 5.994,
+      "step": 2320
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8546242774566474e-05,
+      "loss": 5.8139,
+      "step": 2321
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.8537572254335263e-05,
+      "loss": 5.9078,
+      "step": 2322
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.852890173410405e-05,
+      "loss": 5.8183,
+      "step": 2323
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8520231213872835e-05,
+      "loss": 5.7777,
+      "step": 2324
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8511560693641618e-05,
+      "loss": 5.9104,
+      "step": 2325
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8502890173410404e-05,
+      "loss": 5.7804,
+      "step": 2326
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.849421965317919e-05,
+      "loss": 5.7629,
+      "step": 2327
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8485549132947976e-05,
+      "loss": 5.5441,
+      "step": 2328
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8476878612716762e-05,
+      "loss": 5.7745,
+      "step": 2329
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.846820809248555e-05,
+      "loss": 5.3581,
+      "step": 2330
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8459537572254337e-05,
+      "loss": 5.4102,
+      "step": 2331
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.8450867052023123e-05,
+      "loss": 5.3548,
+      "step": 2332
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.844219653179191e-05,
+      "loss": 5.6484,
+      "step": 2333
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8433526011560695e-05,
+      "loss": 5.0388,
+      "step": 2334
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.842485549132948e-05,
+      "loss": 6.1362,
+      "step": 2335
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8416184971098264e-05,
+      "loss": 6.0266,
+      "step": 2336
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.840751445086705e-05,
+      "loss": 6.0138,
+      "step": 2337
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.839884393063584e-05,
+      "loss": 5.8972,
+      "step": 2338
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8390173410404626e-05,
+      "loss": 5.9116,
+      "step": 2339
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.838150289017341e-05,
+      "loss": 6.0317,
+      "step": 2340
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.8372832369942198e-05,
+      "loss": 6.0209,
+      "step": 2341
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8364161849710984e-05,
+      "loss": 5.9583,
+      "step": 2342
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.835549132947977e-05,
+      "loss": 5.883,
+      "step": 2343
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8346820809248556e-05,
+      "loss": 5.8375,
+      "step": 2344
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8338150289017342e-05,
+      "loss": 5.7954,
+      "step": 2345
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8329479768786124e-05,
+      "loss": 5.7585,
+      "step": 2346
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8320809248554914e-05,
+      "loss": 5.8205,
+      "step": 2347
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.83121387283237e-05,
+      "loss": 5.8951,
+      "step": 2348
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8303468208092486e-05,
+      "loss": 5.6683,
+      "step": 2349
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.8294797687861272e-05,
+      "loss": 5.9732,
+      "step": 2350
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8286127167630058e-05,
+      "loss": 5.7928,
+      "step": 2351
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8277456647398844e-05,
+      "loss": 5.7368,
+      "step": 2352
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.826878612716763e-05,
+      "loss": 5.8595,
+      "step": 2353
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8260115606936416e-05,
+      "loss": 5.7854,
+      "step": 2354
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8251445086705205e-05,
+      "loss": 5.7616,
+      "step": 2355
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8242774566473988e-05,
+      "loss": 5.9021,
+      "step": 2356
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8234104046242774e-05,
+      "loss": 5.7978,
+      "step": 2357
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.822543352601156e-05,
+      "loss": 5.717,
+      "step": 2358
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.8216763005780346e-05,
+      "loss": 5.9142,
+      "step": 2359
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8208092485549132e-05,
+      "loss": 5.7107,
+      "step": 2360
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8199421965317918e-05,
+      "loss": 5.8734,
+      "step": 2361
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8190751445086704e-05,
+      "loss": 5.9213,
+      "step": 2362
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8182080924855494e-05,
+      "loss": 5.8821,
+      "step": 2363
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.817341040462428e-05,
+      "loss": 5.8741,
+      "step": 2364
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8164739884393066e-05,
+      "loss": 5.7771,
+      "step": 2365
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8156069364161852e-05,
+      "loss": 5.8247,
+      "step": 2366
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.8147398843930634e-05,
+      "loss": 5.8684,
+      "step": 2367
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.813872832369942e-05,
+      "loss": 5.6977,
+      "step": 2368
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8130057803468207e-05,
+      "loss": 5.8325,
+      "step": 2369
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8121387283236993e-05,
+      "loss": 5.7856,
+      "step": 2370
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8112716763005782e-05,
+      "loss": 5.8098,
+      "step": 2371
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8104046242774568e-05,
+      "loss": 5.9409,
+      "step": 2372
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8095375722543354e-05,
+      "loss": 5.747,
+      "step": 2373
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.808670520231214e-05,
+      "loss": 5.6705,
+      "step": 2374
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8078034682080926e-05,
+      "loss": 5.8976,
+      "step": 2375
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8078034682080926e-05,
+      "loss": 5.7613,
+      "step": 2376
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.8069364161849712e-05,
+      "loss": 5.6958,
+      "step": 2377
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8060693641618495e-05,
+      "loss": 5.5693,
+      "step": 2378
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.805202312138728e-05,
+      "loss": 5.4969,
+      "step": 2379
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.804335260115607e-05,
+      "loss": 5.7208,
+      "step": 2380
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8034682080924856e-05,
+      "loss": 5.6195,
+      "step": 2381
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8026011560693642e-05,
+      "loss": 5.5073,
+      "step": 2382
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.801734104046243e-05,
+      "loss": 5.6617,
+      "step": 2383
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8008670520231214e-05,
+      "loss": 5.0815,
+      "step": 2384
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.8e-05,
+      "loss": 6.1643,
+      "step": 2385
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.7991329479768786e-05,
+      "loss": 6.1349,
+      "step": 2386
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.7982658959537573e-05,
+      "loss": 5.9339,
+      "step": 2387
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.7973988439306362e-05,
+      "loss": 5.9392,
+      "step": 2388
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.7965317919075145e-05,
+      "loss": 5.8989,
+      "step": 2389
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.795664739884393e-05,
+      "loss": 5.9807,
+      "step": 2390
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.7947976878612717e-05,
+      "loss": 5.8674,
+      "step": 2391
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.7939306358381503e-05,
+      "loss": 5.9927,
+      "step": 2392
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.793063583815029e-05,
+      "loss": 6.0194,
+      "step": 2393
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.7921965317919075e-05,
+      "loss": 5.6762,
+      "step": 2394
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.791329479768786e-05,
+      "loss": 5.6687,
+      "step": 2395
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.790462427745665e-05,
+      "loss": 5.9903,
+      "step": 2396
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.7895953757225436e-05,
+      "loss": 5.6912,
+      "step": 2397
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.7887283236994222e-05,
+      "loss": 5.7228,
+      "step": 2398
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.7878612716763005e-05,
+      "loss": 5.7489,
+      "step": 2399
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.786994219653179e-05,
+      "loss": 5.6966,
+      "step": 2400
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.7861271676300577e-05,
+      "loss": 5.845,
+      "step": 2401
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.7852601156069363e-05,
+      "loss": 5.6786,
+      "step": 2402
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.784393063583815e-05,
+      "loss": 5.7793,
+      "step": 2403
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.783526011560694e-05,
+      "loss": 5.8104,
+      "step": 2404
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.7826589595375725e-05,
+      "loss": 5.7722,
+      "step": 2405
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.781791907514451e-05,
+      "loss": 5.9954,
+      "step": 2406
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.7809248554913297e-05,
+      "loss": 5.886,
+      "step": 2407
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.7800578034682083e-05,
+      "loss": 5.8043,
+      "step": 2408
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.7791907514450865e-05,
+      "loss": 5.6609,
+      "step": 2409
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.778323699421965e-05,
+      "loss": 5.7315,
+      "step": 2410
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.7774566473988437e-05,
+      "loss": 5.9009,
+      "step": 2411
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.7765895953757227e-05,
+      "loss": 5.7775,
+      "step": 2412
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7757225433526013e-05,
+      "loss": 5.8441,
+      "step": 2413
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.77485549132948e-05,
+      "loss": 5.7423,
+      "step": 2414
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7739884393063585e-05,
+      "loss": 5.7726,
+      "step": 2415
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.773121387283237e-05,
+      "loss": 5.9488,
+      "step": 2416
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7722543352601157e-05,
+      "loss": 5.6598,
+      "step": 2417
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7713872832369943e-05,
+      "loss": 5.886,
+      "step": 2418
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.770520231213873e-05,
+      "loss": 5.6192,
+      "step": 2419
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7696531791907515e-05,
+      "loss": 5.8178,
+      "step": 2420
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.76878612716763e-05,
+      "loss": 5.8053,
+      "step": 2421
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7679190751445087e-05,
+      "loss": 5.8303,
+      "step": 2422
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7670520231213873e-05,
+      "loss": 5.6097,
+      "step": 2423
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.766184971098266e-05,
+      "loss": 5.5961,
+      "step": 2424
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7653179190751445e-05,
+      "loss": 5.8571,
+      "step": 2425
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.764450867052023e-05,
+      "loss": 5.6164,
+      "step": 2426
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7635838150289017e-05,
+      "loss": 5.7092,
+      "step": 2427
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7627167630057807e-05,
+      "loss": 5.6414,
+      "step": 2428
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7618497109826593e-05,
+      "loss": 5.7609,
+      "step": 2429
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7609826589595375e-05,
+      "loss": 5.4113,
+      "step": 2430
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.760115606936416e-05,
+      "loss": 5.6451,
+      "step": 2431
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7592485549132947e-05,
+      "loss": 5.4009,
+      "step": 2432
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7583815028901733e-05,
+      "loss": 5.4458,
+      "step": 2433
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.757514450867052e-05,
+      "loss": 5.0465,
+      "step": 2434
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7566473988439306e-05,
+      "loss": 6.1399,
+      "step": 2435
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7557803468208095e-05,
+      "loss": 5.8701,
+      "step": 2436
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.754913294797688e-05,
+      "loss": 6.062,
+      "step": 2437
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7540462427745667e-05,
+      "loss": 5.9964,
+      "step": 2438
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7531791907514453e-05,
+      "loss": 5.9304,
+      "step": 2439
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.752312138728324e-05,
+      "loss": 6.0802,
+      "step": 2440
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7514450867052022e-05,
+      "loss": 5.9618,
+      "step": 2441
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7505780346820808e-05,
+      "loss": 5.885,
+      "step": 2442
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7497109826589594e-05,
+      "loss": 5.8447,
+      "step": 2443
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.748843930635838e-05,
+      "loss": 5.8697,
+      "step": 2444
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.747976878612717e-05,
+      "loss": 5.697,
+      "step": 2445
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7471098265895955e-05,
+      "loss": 5.7746,
+      "step": 2446
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.746242774566474e-05,
+      "loss": 5.794,
+      "step": 2447
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7453757225433527e-05,
+      "loss": 6.0465,
+      "step": 2448
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7445086705202313e-05,
+      "loss": 5.8485,
+      "step": 2449
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.74364161849711e-05,
+      "loss": 5.9628,
+      "step": 2450
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7427745664739882e-05,
+      "loss": 5.9109,
+      "step": 2451
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7419075144508668e-05,
+      "loss": 5.8678,
+      "step": 2452
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7410404624277458e-05,
+      "loss": 5.7104,
+      "step": 2453
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7401734104046244e-05,
+      "loss": 5.9611,
+      "step": 2454
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.739306358381503e-05,
+      "loss": 5.7044,
+      "step": 2455
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7384393063583816e-05,
+      "loss": 5.9397,
+      "step": 2456
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.73757225433526e-05,
+      "loss": 5.8025,
+      "step": 2457
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7367052023121388e-05,
+      "loss": 5.796,
+      "step": 2458
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7358381502890174e-05,
+      "loss": 5.7652,
+      "step": 2459
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.734971098265896e-05,
+      "loss": 5.7638,
+      "step": 2460
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7341040462427746e-05,
+      "loss": 5.8298,
+      "step": 2461
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7332369942196532e-05,
+      "loss": 5.7334,
+      "step": 2462
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7323699421965318e-05,
+      "loss": 5.8306,
+      "step": 2463
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7315028901734104e-05,
+      "loss": 5.8407,
+      "step": 2464
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.730635838150289e-05,
+      "loss": 5.8458,
+      "step": 2465
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.7297687861271676e-05,
+      "loss": 5.734,
+      "step": 2466
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7289017341040462e-05,
+      "loss": 5.7351,
+      "step": 2467
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7280346820809248e-05,
+      "loss": 5.8421,
+      "step": 2468
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7271676300578037e-05,
+      "loss": 5.7891,
+      "step": 2469
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7263005780346824e-05,
+      "loss": 5.7793,
+      "step": 2470
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.725433526011561e-05,
+      "loss": 5.7239,
+      "step": 2471
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7245664739884392e-05,
+      "loss": 5.7405,
+      "step": 2472
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7236994219653178e-05,
+      "loss": 5.7682,
+      "step": 2473
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.7228323699421964e-05,
+      "loss": 5.6477,
+      "step": 2474
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.721965317919075e-05,
+      "loss": 5.7496,
+      "step": 2475
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7210982658959536e-05,
+      "loss": 5.8322,
+      "step": 2476
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7202312138728326e-05,
+      "loss": 5.4921,
+      "step": 2477
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7193641618497112e-05,
+      "loss": 5.7352,
+      "step": 2478
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7184971098265898e-05,
+      "loss": 5.5792,
+      "step": 2479
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7176300578034684e-05,
+      "loss": 5.5514,
+      "step": 2480
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.716763005780347e-05,
+      "loss": 5.3088,
+      "step": 2481
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7158959537572253e-05,
+      "loss": 5.4825,
+      "step": 2482
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.715028901734104e-05,
+      "loss": 5.2472,
+      "step": 2483
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.7141618497109825e-05,
+      "loss": 5.2535,
+      "step": 2484
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.7132947976878614e-05,
+      "loss": 5.9436,
+      "step": 2485
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.71242774566474e-05,
+      "loss": 6.0199,
+      "step": 2486
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.7115606936416186e-05,
+      "loss": 5.9255,
+      "step": 2487
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.7106936416184972e-05,
+      "loss": 6.0809,
+      "step": 2488
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.7098265895953758e-05,
+      "loss": 6.0269,
+      "step": 2489
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.7089595375722544e-05,
+      "loss": 5.9089,
+      "step": 2490
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.708092485549133e-05,
+      "loss": 5.8801,
+      "step": 2491
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.7072254335260116e-05,
+      "loss": 6.0189,
+      "step": 2492
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.7063583815028902e-05,
+      "loss": 5.7706,
+      "step": 2493
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.705491329479769e-05,
+      "loss": 5.9608,
+      "step": 2494
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.7046242774566474e-05,
+      "loss": 5.708,
+      "step": 2495
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.703757225433526e-05,
+      "loss": 5.8503,
+      "step": 2496
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.7028901734104046e-05,
+      "loss": 5.7902,
+      "step": 2497
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.7020231213872832e-05,
+      "loss": 5.5513,
+      "step": 2498
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.701156069364162e-05,
+      "loss": 5.6838,
+      "step": 2499
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.7002890173410405e-05,
+      "loss": 5.6359,
+      "step": 2500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.6994219653179194e-05,
+      "loss": 5.8483,
+      "step": 2501
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.698554913294798e-05,
+      "loss": 5.5859,
+      "step": 2502
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6976878612716763e-05,
+      "loss": 5.57,
+      "step": 2503
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.696820809248555e-05,
+      "loss": 5.8073,
+      "step": 2504
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6959537572254335e-05,
+      "loss": 5.8256,
+      "step": 2505
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.695086705202312e-05,
+      "loss": 5.8205,
+      "step": 2506
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6942196531791907e-05,
+      "loss": 5.7547,
+      "step": 2507
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6933526011560693e-05,
+      "loss": 5.7452,
+      "step": 2508
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6924855491329482e-05,
+      "loss": 5.7918,
+      "step": 2509
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.6916184971098268e-05,
+      "loss": 5.7579,
+      "step": 2510
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6907514450867054e-05,
+      "loss": 5.8938,
+      "step": 2511
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.689884393063584e-05,
+      "loss": 5.665,
+      "step": 2512
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6890173410404623e-05,
+      "loss": 5.8527,
+      "step": 2513
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.688150289017341e-05,
+      "loss": 5.8047,
+      "step": 2514
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6872832369942195e-05,
+      "loss": 5.8263,
+      "step": 2515
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.686416184971098e-05,
+      "loss": 5.6899,
+      "step": 2516
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.685549132947977e-05,
+      "loss": 5.7482,
+      "step": 2517
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6846820809248557e-05,
+      "loss": 5.772,
+      "step": 2518
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.6838150289017343e-05,
+      "loss": 5.7592,
+      "step": 2519
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.682947976878613e-05,
+      "loss": 5.7406,
+      "step": 2520
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6820809248554915e-05,
+      "loss": 5.7932,
+      "step": 2521
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.68121387283237e-05,
+      "loss": 5.8434,
+      "step": 2522
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6803468208092487e-05,
+      "loss": 5.7906,
+      "step": 2523
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.679479768786127e-05,
+      "loss": 5.5825,
+      "step": 2524
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.678612716763006e-05,
+      "loss": 5.6761,
+      "step": 2525
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6777456647398845e-05,
+      "loss": 5.8026,
+      "step": 2526
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.676878612716763e-05,
+      "loss": 5.8687,
+      "step": 2527
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6760115606936417e-05,
+      "loss": 5.6937,
+      "step": 2528
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6751445086705203e-05,
+      "loss": 5.5874,
+      "step": 2529
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.674277456647399e-05,
+      "loss": 5.5517,
+      "step": 2530
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6734104046242775e-05,
+      "loss": 5.4865,
+      "step": 2531
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.672543352601156e-05,
+      "loss": 5.4401,
+      "step": 2532
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.671676300578035e-05,
+      "loss": 5.2314,
+      "step": 2533
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6708092485549133e-05,
+      "loss": 5.0264,
+      "step": 2534
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.669942196531792e-05,
+      "loss": 6.0651,
+      "step": 2535
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.6690751445086705e-05,
+      "loss": 6.0035,
+      "step": 2536
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.668208092485549e-05,
+      "loss": 5.8465,
+      "step": 2537
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6673410404624277e-05,
+      "loss": 6.0662,
+      "step": 2538
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6664739884393063e-05,
+      "loss": 5.8727,
+      "step": 2539
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.665606936416185e-05,
+      "loss": 5.889,
+      "step": 2540
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6647398843930635e-05,
+      "loss": 6.0304,
+      "step": 2541
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6638728323699425e-05,
+      "loss": 5.9526,
+      "step": 2542
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.663005780346821e-05,
+      "loss": 5.8444,
+      "step": 2543
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6621387283236993e-05,
+      "loss": 5.8988,
+      "step": 2544
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.661271676300578e-05,
+      "loss": 5.7957,
+      "step": 2545
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6604046242774565e-05,
+      "loss": 5.8324,
+      "step": 2546
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.659537572254335e-05,
+      "loss": 5.8167,
+      "step": 2547
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6586705202312138e-05,
+      "loss": 5.6456,
+      "step": 2548
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6578034682080924e-05,
+      "loss": 5.8471,
+      "step": 2549
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6569364161849713e-05,
+      "loss": 5.7674,
+      "step": 2550
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.65606936416185e-05,
+      "loss": 5.7604,
+      "step": 2551
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6552023121387285e-05,
+      "loss": 5.8115,
+      "step": 2552
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.654335260115607e-05,
+      "loss": 5.6845,
+      "step": 2553
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6534682080924857e-05,
+      "loss": 5.7811,
+      "step": 2554
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.652601156069364e-05,
+      "loss": 5.5056,
+      "step": 2555
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.6517341040462426e-05,
+      "loss": 5.7144,
+      "step": 2556
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.6508670520231212e-05,
+      "loss": 5.8278,
+      "step": 2557
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.65e-05,
+      "loss": 5.6451,
+      "step": 2558
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.6491329479768787e-05,
+      "loss": 5.9493,
+      "step": 2559
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.6482658959537573e-05,
+      "loss": 5.6491,
+      "step": 2560
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.647398843930636e-05,
+      "loss": 5.5918,
+      "step": 2561
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.6465317919075145e-05,
+      "loss": 5.617,
+      "step": 2562
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.645664739884393e-05,
+      "loss": 5.6557,
+      "step": 2563
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.6447976878612717e-05,
+      "loss": 5.7325,
+      "step": 2564
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.64393063583815e-05,
+      "loss": 5.4946,
+      "step": 2565
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.643063583815029e-05,
+      "loss": 5.6293,
+      "step": 2566
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6421965317919076e-05,
+      "loss": 5.5907,
+      "step": 2567
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.641329479768786e-05,
+      "loss": 5.7002,
+      "step": 2568
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6404624277456648e-05,
+      "loss": 5.9036,
+      "step": 2569
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6395953757225434e-05,
+      "loss": 5.648,
+      "step": 2570
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.638728323699422e-05,
+      "loss": 5.8209,
+      "step": 2571
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6378612716763006e-05,
+      "loss": 5.7224,
+      "step": 2572
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6369942196531792e-05,
+      "loss": 5.6328,
+      "step": 2573
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.636127167630058e-05,
+      "loss": 5.7095,
+      "step": 2574
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.6352601156069367e-05,
+      "loss": 5.7692,
+      "step": 2575
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.634393063583815e-05,
+      "loss": 5.6285,
+      "step": 2576
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.6335260115606936e-05,
+      "loss": 5.6246,
+      "step": 2577
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.6326589595375722e-05,
+      "loss": 5.633,
+      "step": 2578
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.6317919075144508e-05,
+      "loss": 5.4647,
+      "step": 2579
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.6309248554913294e-05,
+      "loss": 5.3744,
+      "step": 2580
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.630057803468208e-05,
+      "loss": 5.4028,
+      "step": 2581
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.629190751445087e-05,
+      "loss": 5.5155,
+      "step": 2582
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6283236994219656e-05,
+      "loss": 5.3093,
+      "step": 2583
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.627456647398844e-05,
+      "loss": 4.9617,
+      "step": 2584
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6265895953757228e-05,
+      "loss": 6.1572,
+      "step": 2585
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.625722543352601e-05,
+      "loss": 5.8823,
+      "step": 2586
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6248554913294796e-05,
+      "loss": 5.9516,
+      "step": 2587
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6239884393063582e-05,
+      "loss": 5.9849,
+      "step": 2588
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.623121387283237e-05,
+      "loss": 5.9636,
+      "step": 2589
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6222543352601158e-05,
+      "loss": 6.1109,
+      "step": 2590
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6213872832369944e-05,
+      "loss": 5.8612,
+      "step": 2591
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.620520231213873e-05,
+      "loss": 5.9639,
+      "step": 2592
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6196531791907516e-05,
+      "loss": 5.8634,
+      "step": 2593
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6187861271676302e-05,
+      "loss": 5.7623,
+      "step": 2594
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6179190751445088e-05,
+      "loss": 5.823,
+      "step": 2595
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.617052023121387e-05,
+      "loss": 5.7119,
+      "step": 2596
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6161849710982657e-05,
+      "loss": 5.8206,
+      "step": 2597
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6153179190751446e-05,
+      "loss": 5.7334,
+      "step": 2598
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6144508670520232e-05,
+      "loss": 5.7251,
+      "step": 2599
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6135838150289018e-05,
+      "loss": 5.8418,
+      "step": 2600
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.6127167630057804e-05,
+      "loss": 5.8566,
+      "step": 2601
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.611849710982659e-05,
+      "loss": 5.8191,
+      "step": 2602
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.6109826589595376e-05,
+      "loss": 5.7651,
+      "step": 2603
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.6101156069364162e-05,
+      "loss": 5.884,
+      "step": 2604
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.6092485549132948e-05,
+      "loss": 5.5994,
+      "step": 2605
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.6083815028901738e-05,
+      "loss": 5.6655,
+      "step": 2606
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.607514450867052e-05,
+      "loss": 5.7522,
+      "step": 2607
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.6066473988439306e-05,
+      "loss": 5.4767,
+      "step": 2608
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.6057803468208092e-05,
+      "loss": 5.6421,
+      "step": 2609
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.604913294797688e-05,
+      "loss": 5.7198,
+      "step": 2610
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.6040462427745664e-05,
+      "loss": 5.7154,
+      "step": 2611
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.603179190751445e-05,
+      "loss": 5.5807,
+      "step": 2612
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.6023121387283237e-05,
+      "loss": 5.8537,
+      "step": 2613
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.6014450867052026e-05,
+      "loss": 5.6865,
+      "step": 2614
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.6005780346820812e-05,
+      "loss": 5.8455,
+      "step": 2615
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.5997109826589598e-05,
+      "loss": 5.8058,
+      "step": 2616
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.598843930635838e-05,
+      "loss": 5.6953,
+      "step": 2617
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.5979768786127167e-05,
+      "loss": 5.5456,
+      "step": 2618
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.5971098265895953e-05,
+      "loss": 5.7384,
+      "step": 2619
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.596242774566474e-05,
+      "loss": 5.6043,
+      "step": 2620
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.5953757225433525e-05,
+      "loss": 5.4919,
+      "step": 2621
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.5945086705202314e-05,
+      "loss": 5.6323,
+      "step": 2622
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.59364161849711e-05,
+      "loss": 5.6517,
+      "step": 2623
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.5927745664739886e-05,
+      "loss": 5.8249,
+      "step": 2624
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.5919075144508672e-05,
+      "loss": 5.6079,
+      "step": 2625
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.591040462427746e-05,
+      "loss": 5.5786,
+      "step": 2626
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5901734104046244e-05,
+      "loss": 5.293,
+      "step": 2627
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5893063583815027e-05,
+      "loss": 5.3446,
+      "step": 2628
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5884393063583813e-05,
+      "loss": 5.7171,
+      "step": 2629
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5875722543352602e-05,
+      "loss": 5.7661,
+      "step": 2630
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.586705202312139e-05,
+      "loss": 5.2179,
+      "step": 2631
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5858381502890175e-05,
+      "loss": 5.2124,
+      "step": 2632
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.584971098265896e-05,
+      "loss": 5.2354,
+      "step": 2633
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5841040462427747e-05,
+      "loss": 4.9928,
+      "step": 2634
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5832369942196533e-05,
+      "loss": 6.08,
+      "step": 2635
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.582369942196532e-05,
+      "loss": 5.9955,
+      "step": 2636
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5815028901734105e-05,
+      "loss": 6.165,
+      "step": 2637
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5806358381502887e-05,
+      "loss": 5.768,
+      "step": 2638
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5797687861271677e-05,
+      "loss": 5.8993,
+      "step": 2639
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5789017341040463e-05,
+      "loss": 5.7331,
+      "step": 2640
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.578034682080925e-05,
+      "loss": 5.7988,
+      "step": 2641
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5771676300578035e-05,
+      "loss": 5.8557,
+      "step": 2642
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.576300578034682e-05,
+      "loss": 5.8678,
+      "step": 2643
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.5754335260115607e-05,
+      "loss": 5.6033,
+      "step": 2644
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5745664739884393e-05,
+      "loss": 5.691,
+      "step": 2645
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.573699421965318e-05,
+      "loss": 5.8058,
+      "step": 2646
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.572832369942197e-05,
+      "loss": 5.8767,
+      "step": 2647
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.571965317919075e-05,
+      "loss": 5.8109,
+      "step": 2648
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5710982658959537e-05,
+      "loss": 5.833,
+      "step": 2649
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5702312138728323e-05,
+      "loss": 5.6479,
+      "step": 2650
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.569364161849711e-05,
+      "loss": 5.8673,
+      "step": 2651
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.5684971098265895e-05,
+      "loss": 5.7182,
+      "step": 2652
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.567630057803468e-05,
+      "loss": 5.815,
+      "step": 2653
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5667630057803467e-05,
+      "loss": 5.5822,
+      "step": 2654
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5658959537572257e-05,
+      "loss": 5.7133,
+      "step": 2655
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5650289017341043e-05,
+      "loss": 5.7653,
+      "step": 2656
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.564161849710983e-05,
+      "loss": 5.7088,
+      "step": 2657
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5632947976878615e-05,
+      "loss": 5.8168,
+      "step": 2658
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5624277456647397e-05,
+      "loss": 5.6596,
+      "step": 2659
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5615606936416184e-05,
+      "loss": 5.6555,
+      "step": 2660
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.560693641618497e-05,
+      "loss": 5.8318,
+      "step": 2661
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5598265895953756e-05,
+      "loss": 5.9238,
+      "step": 2662
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5589595375722545e-05,
+      "loss": 5.5852,
+      "step": 2663
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.558092485549133e-05,
+      "loss": 5.8859,
+      "step": 2664
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5572254335260117e-05,
+      "loss": 5.3367,
+      "step": 2665
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5563583815028903e-05,
+      "loss": 5.5798,
+      "step": 2666
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.555491329479769e-05,
+      "loss": 5.76,
+      "step": 2667
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5546242774566475e-05,
+      "loss": 5.7235,
+      "step": 2668
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5537572254335258e-05,
+      "loss": 5.6132,
+      "step": 2669
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5528901734104044e-05,
+      "loss": 5.5993,
+      "step": 2670
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5520231213872833e-05,
+      "loss": 5.593,
+      "step": 2671
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.551156069364162e-05,
+      "loss": 5.4318,
+      "step": 2672
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5502890173410405e-05,
+      "loss": 5.456,
+      "step": 2673
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.549421965317919e-05,
+      "loss": 5.2267,
+      "step": 2674
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5485549132947977e-05,
+      "loss": 5.3482,
+      "step": 2675
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5476878612716763e-05,
+      "loss": 4.9006,
+      "step": 2676
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.546820809248555e-05,
+      "loss": 5.8986,
+      "step": 2677
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5459537572254335e-05,
+      "loss": 5.9064,
+      "step": 2678
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.545086705202312e-05,
+      "loss": 5.9584,
+      "step": 2679
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5442196531791908e-05,
+      "loss": 5.7993,
+      "step": 2680
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.5433526011560694e-05,
+      "loss": 5.9554,
+      "step": 2681
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.542485549132948e-05,
+      "loss": 5.6799,
+      "step": 2682
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.5416184971098266e-05,
+      "loss": 5.656,
+      "step": 2683
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.5407514450867052e-05,
+      "loss": 5.8713,
+      "step": 2684
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.5398843930635838e-05,
+      "loss": 5.8011,
+      "step": 2685
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.5390173410404624e-05,
+      "loss": 5.7338,
+      "step": 2686
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.5381502890173413e-05,
+      "loss": 5.7878,
+      "step": 2687
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.53728323699422e-05,
+      "loss": 5.6378,
+      "step": 2688
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.5364161849710985e-05,
+      "loss": 5.7442,
+      "step": 2689
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.5355491329479768e-05,
+      "loss": 5.6688,
+      "step": 2690
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.5346820809248554e-05,
+      "loss": 5.7041,
+      "step": 2691
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.533815028901734e-05,
+      "loss": 5.8323,
+      "step": 2692
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.5329479768786126e-05,
+      "loss": 5.7745,
+      "step": 2693
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.5320809248554912e-05,
+      "loss": 5.6874,
+      "step": 2694
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.53121387283237e-05,
+      "loss": 5.7412,
+      "step": 2695
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.5303468208092487e-05,
+      "loss": 5.5903,
+      "step": 2696
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.5294797687861274e-05,
+      "loss": 5.6311,
+      "step": 2697
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.528612716763006e-05,
+      "loss": 5.5192,
+      "step": 2698
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.5277456647398846e-05,
+      "loss": 5.6743,
+      "step": 2699
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.5268786127167628e-05,
+      "loss": 5.7855,
+      "step": 2700
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.5260115606936414e-05,
+      "loss": 5.7361,
+      "step": 2701
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.5251445086705202e-05,
+      "loss": 5.7432,
+      "step": 2702
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.524277456647399e-05,
+      "loss": 5.9032,
+      "step": 2703
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.5234104046242776e-05,
+      "loss": 5.5624,
+      "step": 2704
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.5225433526011562e-05,
+      "loss": 5.6814,
+      "step": 2705
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.5216763005780348e-05,
+      "loss": 5.7267,
+      "step": 2706
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.5208092485549134e-05,
+      "loss": 5.4694,
+      "step": 2707
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.5199421965317918e-05,
+      "loss": 5.3978,
+      "step": 2708
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.5190751445086704e-05,
+      "loss": 5.459,
+      "step": 2709
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.518208092485549e-05,
+      "loss": 5.6644,
+      "step": 2710
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.517341040462428e-05,
+      "loss": 5.5751,
+      "step": 2711
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.5164739884393066e-05,
+      "loss": 5.693,
+      "step": 2712
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.515606936416185e-05,
+      "loss": 5.6663,
+      "step": 2713
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.5147398843930636e-05,
+      "loss": 5.6767,
+      "step": 2714
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.5138728323699422e-05,
+      "loss": 5.5692,
+      "step": 2715
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.5130057803468208e-05,
+      "loss": 5.5518,
+      "step": 2716
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.5121387283236994e-05,
+      "loss": 5.5293,
+      "step": 2717
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.511271676300578e-05,
+      "loss": 5.5791,
+      "step": 2718
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.5104046242774568e-05,
+      "loss": 5.4694,
+      "step": 2719
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.5095375722543354e-05,
+      "loss": 5.6932,
+      "step": 2720
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.508670520231214e-05,
+      "loss": 5.2995,
+      "step": 2721
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.5078034682080926e-05,
+      "loss": 5.4273,
+      "step": 2722
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.5069364161849712e-05,
+      "loss": 5.311,
+      "step": 2723
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.5060693641618496e-05,
+      "loss": 5.3742,
+      "step": 2724
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.5052023121387282e-05,
+      "loss": 5.4605,
+      "step": 2725
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.5043352601156069e-05,
+      "loss": 4.8729,
+      "step": 2726
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.5034682080924858e-05,
+      "loss": 5.9304,
+      "step": 2727
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.5026011560693644e-05,
+      "loss": 5.9629,
+      "step": 2728
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.5017341040462428e-05,
+      "loss": 5.7124,
+      "step": 2729
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.5008670520231214e-05,
+      "loss": 5.674,
+      "step": 2730
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.5e-05,
+      "loss": 5.8389,
+      "step": 2731
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.4991329479768786e-05,
+      "loss": 5.8728,
+      "step": 2732
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.4982658959537572e-05,
+      "loss": 6.028,
+      "step": 2733
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.4973988439306358e-05,
+      "loss": 5.6434,
+      "step": 2734
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.4965317919075145e-05,
+      "loss": 5.6105,
+      "step": 2735
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.495664739884393e-05,
+      "loss": 5.7201,
+      "step": 2736
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.4947976878612717e-05,
+      "loss": 5.6228,
+      "step": 2737
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.4939306358381504e-05,
+      "loss": 5.6791,
+      "step": 2738
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.4930635838150289e-05,
+      "loss": 5.821,
+      "step": 2739
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.4921965317919075e-05,
+      "loss": 5.5984,
+      "step": 2740
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.491329479768786e-05,
+      "loss": 5.8845,
+      "step": 2741
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.4904624277456648e-05,
+      "loss": 5.739,
+      "step": 2742
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.4895953757225434e-05,
+      "loss": 5.6074,
+      "step": 2743
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.488728323699422e-05,
+      "loss": 5.6447,
+      "step": 2744
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.4878612716763005e-05,
+      "loss": 5.6274,
+      "step": 2745
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.4869942196531793e-05,
+      "loss": 5.6934,
+      "step": 2746
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.4861271676300579e-05,
+      "loss": 5.5832,
+      "step": 2747
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.4852601156069365e-05,
+      "loss": 5.783,
+      "step": 2748
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.484393063583815e-05,
+      "loss": 5.72,
+      "step": 2749
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.4835260115606937e-05,
+      "loss": 5.7491,
+      "step": 2750
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.4826589595375723e-05,
+      "loss": 5.6584,
+      "step": 2751
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.4817919075144509e-05,
+      "loss": 5.6998,
+      "step": 2752
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.4809248554913295e-05,
+      "loss": 5.6885,
+      "step": 2753
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.4800578034682083e-05,
+      "loss": 5.6141,
+      "step": 2754
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.4791907514450867e-05,
+      "loss": 5.8275,
+      "step": 2755
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.4783236994219653e-05,
+      "loss": 5.7048,
+      "step": 2756
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.4774566473988439e-05,
+      "loss": 5.6465,
+      "step": 2757
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.4765895953757227e-05,
+      "loss": 5.5732,
+      "step": 2758
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.4757225433526013e-05,
+      "loss": 5.7426,
+      "step": 2759
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.4748554913294797e-05,
+      "loss": 5.5264,
+      "step": 2760
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.4739884393063583e-05,
+      "loss": 5.5125,
+      "step": 2761
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.473121387283237e-05,
+      "loss": 5.4877,
+      "step": 2762
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.4722543352601157e-05,
+      "loss": 5.6246,
+      "step": 2763
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.4713872832369943e-05,
+      "loss": 5.6908,
+      "step": 2764
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.4705202312138727e-05,
+      "loss": 5.5966,
+      "step": 2765
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.4696531791907515e-05,
+      "loss": 5.6561,
+      "step": 2766
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.4687861271676301e-05,
+      "loss": 5.6936,
+      "step": 2767
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.4679190751445087e-05,
+      "loss": 5.6424,
+      "step": 2768
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.4670520231213873e-05,
+      "loss": 5.3524,
+      "step": 2769
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.4661849710982659e-05,
+      "loss": 5.5335,
+      "step": 2770
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.4653179190751445e-05,
+      "loss": 5.2202,
+      "step": 2771
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.4644508670520231e-05,
+      "loss": 5.4553,
+      "step": 2772
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.4635838150289017e-05,
+      "loss": 5.2891,
+      "step": 2773
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.4627167630057805e-05,
+      "loss": 5.1782,
+      "step": 2774
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.4618497109826591e-05,
+      "loss": 5.3018,
+      "step": 2775
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.4609826589595375e-05,
+      "loss": 5.093,
+      "step": 2776
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.4601156069364161e-05,
+      "loss": 5.9164,
+      "step": 2777
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.4592485549132949e-05,
+      "loss": 5.7576,
+      "step": 2778
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.4583815028901735e-05,
+      "loss": 5.6063,
+      "step": 2779
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.4575144508670521e-05,
+      "loss": 5.8208,
+      "step": 2780
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.4566473988439305e-05,
+      "loss": 5.7642,
+      "step": 2781
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.4557803468208093e-05,
+      "loss": 5.8087,
+      "step": 2782
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.454913294797688e-05,
+      "loss": 5.8315,
+      "step": 2783
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.4540462427745665e-05,
+      "loss": 5.5638,
+      "step": 2784
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.4531791907514451e-05,
+      "loss": 5.7431,
+      "step": 2785
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.4523121387283237e-05,
+      "loss": 5.6606,
+      "step": 2786
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.4514450867052023e-05,
+      "loss": 5.6295,
+      "step": 2787
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.450578034682081e-05,
+      "loss": 5.8203,
+      "step": 2788
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.4497109826589595e-05,
+      "loss": 5.6978,
+      "step": 2789
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.4488439306358383e-05,
+      "loss": 5.7141,
+      "step": 2790
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.4479768786127167e-05,
+      "loss": 5.7373,
+      "step": 2791
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.4471098265895954e-05,
+      "loss": 5.627,
+      "step": 2792
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.446242774566474e-05,
+      "loss": 5.6982,
+      "step": 2793
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.4453757225433527e-05,
+      "loss": 5.5,
+      "step": 2794
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.4445086705202313e-05,
+      "loss": 5.5686,
+      "step": 2795
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.4436416184971098e-05,
+      "loss": 5.5784,
+      "step": 2796
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4427745664739884e-05,
+      "loss": 5.6769,
+      "step": 2797
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4419075144508671e-05,
+      "loss": 5.6965,
+      "step": 2798
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4410404624277457e-05,
+      "loss": 5.6073,
+      "step": 2799
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4401734104046243e-05,
+      "loss": 5.677,
+      "step": 2800
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.439306358381503e-05,
+      "loss": 5.5439,
+      "step": 2801
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4384393063583816e-05,
+      "loss": 5.6112,
+      "step": 2802
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4375722543352602e-05,
+      "loss": 5.6052,
+      "step": 2803
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4367052023121388e-05,
+      "loss": 5.6854,
+      "step": 2804
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.4358381502890174e-05,
+      "loss": 5.4811,
+      "step": 2805
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.4349710982658961e-05,
+      "loss": 5.6022,
+      "step": 2806
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.4341040462427746e-05,
+      "loss": 5.4854,
+      "step": 2807
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.4332369942196532e-05,
+      "loss": 5.4632,
+      "step": 2808
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.4323699421965318e-05,
+      "loss": 5.5912,
+      "step": 2809
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.4315028901734106e-05,
+      "loss": 5.5738,
+      "step": 2810
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.4306358381502892e-05,
+      "loss": 5.5753,
+      "step": 2811
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.4297687861271676e-05,
+      "loss": 5.509,
+      "step": 2812
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.4289017341040462e-05,
+      "loss": 5.3802,
+      "step": 2813
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.428034682080925e-05,
+      "loss": 5.6099,
+      "step": 2814
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.4271676300578036e-05,
+      "loss": 5.5734,
+      "step": 2815
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.4263005780346822e-05,
+      "loss": 5.6986,
+      "step": 2816
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.4254335260115606e-05,
+      "loss": 5.5833,
+      "step": 2817
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.4245664739884394e-05,
+      "loss": 5.4466,
+      "step": 2818
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.423699421965318e-05,
+      "loss": 5.4855,
+      "step": 2819
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.4228323699421966e-05,
+      "loss": 5.374,
+      "step": 2820
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.4219653179190752e-05,
+      "loss": 5.5274,
+      "step": 2821
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.4210982658959538e-05,
+      "loss": 5.3772,
+      "step": 2822
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.4202312138728324e-05,
+      "loss": 5.3739,
+      "step": 2823
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.419364161849711e-05,
+      "loss": 5.2144,
+      "step": 2824
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.4184971098265896e-05,
+      "loss": 5.266,
+      "step": 2825
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.4176300578034684e-05,
+      "loss": 4.9162,
+      "step": 2826
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.416763005780347e-05,
+      "loss": 6.1522,
+      "step": 2827
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.4158959537572254e-05,
+      "loss": 5.9026,
+      "step": 2828
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.415028901734104e-05,
+      "loss": 5.9744,
+      "step": 2829
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.4141618497109826e-05,
+      "loss": 5.9505,
+      "step": 2830
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.4132947976878614e-05,
+      "loss": 5.8591,
+      "step": 2831
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.41242774566474e-05,
+      "loss": 6.0634,
+      "step": 2832
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.4115606936416184e-05,
+      "loss": 5.7648,
+      "step": 2833
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.410693641618497e-05,
+      "loss": 6.0653,
+      "step": 2834
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.4098265895953758e-05,
+      "loss": 5.7087,
+      "step": 2835
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.4089595375722544e-05,
+      "loss": 5.5984,
+      "step": 2836
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.408092485549133e-05,
+      "loss": 5.8286,
+      "step": 2837
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.4072254335260114e-05,
+      "loss": 5.7271,
+      "step": 2838
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.4063583815028902e-05,
+      "loss": 5.9021,
+      "step": 2839
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.4054913294797688e-05,
+      "loss": 5.8112,
+      "step": 2840
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.4046242774566474e-05,
+      "loss": 5.6854,
+      "step": 2841
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.403757225433526e-05,
+      "loss": 5.6105,
+      "step": 2842
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.4028901734104046e-05,
+      "loss": 5.64,
+      "step": 2843
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.4020231213872832e-05,
+      "loss": 5.6432,
+      "step": 2844
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.4011560693641618e-05,
+      "loss": 5.7826,
+      "step": 2845
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.4002890173410404e-05,
+      "loss": 5.6829,
+      "step": 2846
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3994219653179192e-05,
+      "loss": 5.7455,
+      "step": 2847
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3985549132947977e-05,
+      "loss": 5.7564,
+      "step": 2848
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.3976878612716763e-05,
+      "loss": 5.6405,
+      "step": 2849
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3968208092485549e-05,
+      "loss": 5.5576,
+      "step": 2850
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3959537572254336e-05,
+      "loss": 5.6231,
+      "step": 2851
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3950867052023122e-05,
+      "loss": 5.6484,
+      "step": 2852
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3942196531791908e-05,
+      "loss": 5.5752,
+      "step": 2853
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3933526011560693e-05,
+      "loss": 5.6808,
+      "step": 2854
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.392485549132948e-05,
+      "loss": 5.664,
+      "step": 2855
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3916184971098266e-05,
+      "loss": 5.6131,
+      "step": 2856
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3907514450867053e-05,
+      "loss": 5.8567,
+      "step": 2857
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.3898843930635839e-05,
+      "loss": 5.444,
+      "step": 2858
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3890173410404625e-05,
+      "loss": 5.726,
+      "step": 2859
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.388150289017341e-05,
+      "loss": 5.5616,
+      "step": 2860
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3872832369942197e-05,
+      "loss": 5.4651,
+      "step": 2861
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3864161849710983e-05,
+      "loss": 5.7654,
+      "step": 2862
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.385549132947977e-05,
+      "loss": 5.4449,
+      "step": 2863
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3846820809248555e-05,
+      "loss": 5.5052,
+      "step": 2864
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.383815028901734e-05,
+      "loss": 5.3912,
+      "step": 2865
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3829479768786127e-05,
+      "loss": 5.4972,
+      "step": 2866
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.3820809248554915e-05,
+      "loss": 5.5674,
+      "step": 2867
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.38121387283237e-05,
+      "loss": 5.6931,
+      "step": 2868
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3803468208092485e-05,
+      "loss": 5.4033,
+      "step": 2869
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3803468208092485e-05,
+      "loss": 5.3575,
+      "step": 2870
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3794797687861271e-05,
+      "loss": 5.3,
+      "step": 2871
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3786127167630059e-05,
+      "loss": 5.292,
+      "step": 2872
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3777456647398845e-05,
+      "loss": 5.2626,
+      "step": 2873
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.376878612716763e-05,
+      "loss": 5.2629,
+      "step": 2874
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3760115606936415e-05,
+      "loss": 5.0757,
+      "step": 2875
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.3751445086705203e-05,
+      "loss": 4.8716,
+      "step": 2876
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3742774566473989e-05,
+      "loss": 5.93,
+      "step": 2877
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3734104046242775e-05,
+      "loss": 5.9746,
+      "step": 2878
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3725433526011561e-05,
+      "loss": 6.0137,
+      "step": 2879
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3716763005780349e-05,
+      "loss": 5.8386,
+      "step": 2880
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3708092485549133e-05,
+      "loss": 5.681,
+      "step": 2881
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3699421965317919e-05,
+      "loss": 5.747,
+      "step": 2882
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3690751445086705e-05,
+      "loss": 5.7465,
+      "step": 2883
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3682080924855493e-05,
+      "loss": 5.8221,
+      "step": 2884
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.3673410404624279e-05,
+      "loss": 5.7454,
+      "step": 2885
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3664739884393063e-05,
+      "loss": 5.5505,
+      "step": 2886
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.365606936416185e-05,
+      "loss": 5.5473,
+      "step": 2887
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3647398843930637e-05,
+      "loss": 5.8301,
+      "step": 2888
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3638728323699423e-05,
+      "loss": 5.8494,
+      "step": 2889
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3630057803468209e-05,
+      "loss": 5.6024,
+      "step": 2890
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3621387283236993e-05,
+      "loss": 5.7373,
+      "step": 2891
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3612716763005781e-05,
+      "loss": 5.8057,
+      "step": 2892
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3604046242774567e-05,
+      "loss": 5.6828,
+      "step": 2893
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.3595375722543353e-05,
+      "loss": 5.6656,
+      "step": 2894
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.358670520231214e-05,
+      "loss": 5.6848,
+      "step": 2895
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3578034682080925e-05,
+      "loss": 5.6269,
+      "step": 2896
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3569364161849711e-05,
+      "loss": 5.728,
+      "step": 2897
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3560693641618497e-05,
+      "loss": 5.671,
+      "step": 2898
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3552023121387283e-05,
+      "loss": 5.5758,
+      "step": 2899
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3543352601156071e-05,
+      "loss": 5.6257,
+      "step": 2900
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3534682080924855e-05,
+      "loss": 5.4657,
+      "step": 2901
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3526011560693641e-05,
+      "loss": 5.5953,
+      "step": 2902
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.3517341040462427e-05,
+      "loss": 5.6747,
+      "step": 2903
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3508670520231215e-05,
+      "loss": 5.5604,
+      "step": 2904
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 5.6901,
+      "step": 2905
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3491329479768787e-05,
+      "loss": 5.812,
+      "step": 2906
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3482658959537572e-05,
+      "loss": 5.589,
+      "step": 2907
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.347398843930636e-05,
+      "loss": 5.6632,
+      "step": 2908
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3465317919075145e-05,
+      "loss": 5.7988,
+      "step": 2909
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3456647398843931e-05,
+      "loss": 5.6632,
+      "step": 2910
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3447976878612717e-05,
+      "loss": 5.6039,
+      "step": 2911
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.3439306358381503e-05,
+      "loss": 5.5833,
+      "step": 2912
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.343063583815029e-05,
+      "loss": 5.6497,
+      "step": 2913
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3421965317919075e-05,
+      "loss": 5.5337,
+      "step": 2914
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3413294797687862e-05,
+      "loss": 5.5154,
+      "step": 2915
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.340462427745665e-05,
+      "loss": 5.5108,
+      "step": 2916
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3395953757225434e-05,
+      "loss": 5.6465,
+      "step": 2917
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.338728323699422e-05,
+      "loss": 5.6277,
+      "step": 2918
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3378612716763006e-05,
+      "loss": 5.6413,
+      "step": 2919
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.3369942196531793e-05,
+      "loss": 5.5713,
+      "step": 2920
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.336127167630058e-05,
+      "loss": 5.3309,
+      "step": 2921
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3352601156069364e-05,
+      "loss": 5.3572,
+      "step": 2922
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.334393063583815e-05,
+      "loss": 5.3908,
+      "step": 2923
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3335260115606938e-05,
+      "loss": 5.2371,
+      "step": 2924
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3326589595375724e-05,
+      "loss": 5.3896,
+      "step": 2925
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.331791907514451e-05,
+      "loss": 4.9373,
+      "step": 2926
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3309248554913294e-05,
+      "loss": 6.1502,
+      "step": 2927
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.330057803468208e-05,
+      "loss": 5.9275,
+      "step": 2928
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3291907514450868e-05,
+      "loss": 5.9494,
+      "step": 2929
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.3283236994219654e-05,
+      "loss": 5.8384,
+      "step": 2930
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.327456647398844e-05,
+      "loss": 5.7688,
+      "step": 2931
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3265895953757226e-05,
+      "loss": 5.7801,
+      "step": 2932
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3257225433526012e-05,
+      "loss": 5.8038,
+      "step": 2933
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3248554913294798e-05,
+      "loss": 5.8212,
+      "step": 2934
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3239884393063584e-05,
+      "loss": 5.778,
+      "step": 2935
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.323121387283237e-05,
+      "loss": 5.5002,
+      "step": 2936
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3222543352601158e-05,
+      "loss": 5.6765,
+      "step": 2937
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3213872832369942e-05,
+      "loss": 5.6491,
+      "step": 2938
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.3205202312138728e-05,
+      "loss": 5.6756,
+      "step": 2939
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3196531791907514e-05,
+      "loss": 5.6965,
+      "step": 2940
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3187861271676302e-05,
+      "loss": 5.8501,
+      "step": 2941
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3179190751445088e-05,
+      "loss": 5.6718,
+      "step": 2942
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3170520231213872e-05,
+      "loss": 5.6694,
+      "step": 2943
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3161849710982658e-05,
+      "loss": 5.609,
+      "step": 2944
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3153179190751446e-05,
+      "loss": 5.7405,
+      "step": 2945
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3144508670520232e-05,
+      "loss": 5.7105,
+      "step": 2946
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3135838150289018e-05,
+      "loss": 5.5683,
+      "step": 2947
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.3127167630057802e-05,
+      "loss": 5.7268,
+      "step": 2948
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.311849710982659e-05,
+      "loss": 5.5074,
+      "step": 2949
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3109826589595376e-05,
+      "loss": 5.5898,
+      "step": 2950
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3101156069364162e-05,
+      "loss": 5.634,
+      "step": 2951
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3092485549132948e-05,
+      "loss": 5.568,
+      "step": 2952
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3083815028901734e-05,
+      "loss": 5.444,
+      "step": 2953
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.307514450867052e-05,
+      "loss": 5.6497,
+      "step": 2954
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3066473988439306e-05,
+      "loss": 5.6537,
+      "step": 2955
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.3057803468208092e-05,
+      "loss": 5.7532,
+      "step": 2956
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.304913294797688e-05,
+      "loss": 5.6731,
+      "step": 2957
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3040462427745664e-05,
+      "loss": 5.6814,
+      "step": 2958
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.303179190751445e-05,
+      "loss": 5.663,
+      "step": 2959
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3023121387283236e-05,
+      "loss": 5.6581,
+      "step": 2960
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.3014450867052024e-05,
+      "loss": 5.7157,
+      "step": 2961
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.300578034682081e-05,
+      "loss": 5.8843,
+      "step": 2962
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.2997109826589596e-05,
+      "loss": 5.6964,
+      "step": 2963
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.298843930635838e-05,
+      "loss": 5.6063,
+      "step": 2964
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.2979768786127168e-05,
+      "loss": 5.6032,
+      "step": 2965
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2971098265895954e-05,
+      "loss": 5.643,
+      "step": 2966
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.296242774566474e-05,
+      "loss": 5.5077,
+      "step": 2967
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2953757225433526e-05,
+      "loss": 5.4264,
+      "step": 2968
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2945086705202312e-05,
+      "loss": 5.5827,
+      "step": 2969
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2936416184971098e-05,
+      "loss": 5.5361,
+      "step": 2970
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2927745664739885e-05,
+      "loss": 5.4286,
+      "step": 2971
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.291907514450867e-05,
+      "loss": 5.3946,
+      "step": 2972
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2910404624277458e-05,
+      "loss": 5.2598,
+      "step": 2973
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.2901734104046243e-05,
+      "loss": 5.3033,
+      "step": 2974
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.2893063583815029e-05,
+      "loss": 5.2774,
+      "step": 2975
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.2884393063583815e-05,
+      "loss": 4.9094,
+      "step": 2976
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.2875722543352602e-05,
+      "loss": 6.0885,
+      "step": 2977
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.2867052023121388e-05,
+      "loss": 5.7902,
+      "step": 2978
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.2858381502890173e-05,
+      "loss": 5.6974,
+      "step": 2979
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.2849710982658959e-05,
+      "loss": 5.7826,
+      "step": 2980
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.2841040462427747e-05,
+      "loss": 5.8182,
+      "step": 2981
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.2832369942196533e-05,
+      "loss": 5.8481,
+      "step": 2982
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.2823699421965319e-05,
+      "loss": 5.6565,
+      "step": 2983
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2815028901734103e-05,
+      "loss": 5.6938,
+      "step": 2984
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.280635838150289e-05,
+      "loss": 5.7552,
+      "step": 2985
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2797687861271677e-05,
+      "loss": 5.5694,
+      "step": 2986
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2789017341040463e-05,
+      "loss": 5.659,
+      "step": 2987
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2780346820809249e-05,
+      "loss": 5.4247,
+      "step": 2988
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2771676300578037e-05,
+      "loss": 5.7123,
+      "step": 2989
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2763005780346821e-05,
+      "loss": 5.6619,
+      "step": 2990
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2754335260115607e-05,
+      "loss": 5.5788,
+      "step": 2991
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.2745664739884393e-05,
+      "loss": 5.6175,
+      "step": 2992
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.273699421965318e-05,
+      "loss": 5.6433,
+      "step": 2993
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2728323699421967e-05,
+      "loss": 5.6871,
+      "step": 2994
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2719653179190751e-05,
+      "loss": 5.5603,
+      "step": 2995
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2710982658959537e-05,
+      "loss": 5.7456,
+      "step": 2996
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2702312138728325e-05,
+      "loss": 5.6589,
+      "step": 2997
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.269364161849711e-05,
+      "loss": 5.5256,
+      "step": 2998
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2684971098265897e-05,
+      "loss": 5.5635,
+      "step": 2999
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2676300578034681e-05,
+      "loss": 5.6103,
+      "step": 3000
+    },
+    {
+      "epoch": 3.36,
+      "eval_loss": 5.628564834594727,
+      "eval_runtime": 963.4096,
+      "eval_samples_per_second": 2.742,
+      "eval_steps_per_second": 0.344,
+      "eval_wer": 1.5293172212631705,
+      "step": 3000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2667630057803469e-05,
+      "loss": 5.5514,
+      "step": 3001
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2658959537572255e-05,
+      "loss": 5.5598,
+      "step": 3002
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2650289017341041e-05,
+      "loss": 5.6766,
+      "step": 3003
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2641618497109827e-05,
+      "loss": 5.6639,
+      "step": 3004
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2632947976878613e-05,
+      "loss": 5.675,
+      "step": 3005
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2624277456647399e-05,
+      "loss": 5.6434,
+      "step": 3006
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2615606936416185e-05,
+      "loss": 5.594,
+      "step": 3007
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2606936416184971e-05,
+      "loss": 5.3594,
+      "step": 3008
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2598265895953759e-05,
+      "loss": 5.4268,
+      "step": 3009
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2589595375722543e-05,
+      "loss": 5.6629,
+      "step": 3010
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.258092485549133e-05,
+      "loss": 5.7721,
+      "step": 3011
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2572254335260115e-05,
+      "loss": 5.6332,
+      "step": 3012
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2563583815028903e-05,
+      "loss": 5.5658,
+      "step": 3013
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2554913294797689e-05,
+      "loss": 5.5719,
+      "step": 3014
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2546242774566475e-05,
+      "loss": 5.56,
+      "step": 3015
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.253757225433526e-05,
+      "loss": 5.5266,
+      "step": 3016
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2528901734104047e-05,
+      "loss": 5.6051,
+      "step": 3017
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2520231213872833e-05,
+      "loss": 5.2491,
+      "step": 3018
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.251156069364162e-05,
+      "loss": 5.7448,
+      "step": 3019
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2502890173410405e-05,
+      "loss": 5.3691,
+      "step": 3020
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.249421965317919e-05,
+      "loss": 5.5387,
+      "step": 3021
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2485549132947977e-05,
+      "loss": 5.5543,
+      "step": 3022
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2476878612716763e-05,
+      "loss": 4.9667,
+      "step": 3023
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.246820809248555e-05,
+      "loss": 5.0539,
+      "step": 3024
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2459537572254335e-05,
+      "loss": 4.9561,
+      "step": 3025
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2450867052023121e-05,
+      "loss": 4.8042,
+      "step": 3026
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2442196531791907e-05,
+      "loss": 6.0697,
+      "step": 3027
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2433526011560694e-05,
+      "loss": 5.8133,
+      "step": 3028
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.242485549132948e-05,
+      "loss": 5.9345,
+      "step": 3029
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2416184971098267e-05,
+      "loss": 5.8965,
+      "step": 3030
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2407514450867052e-05,
+      "loss": 5.6318,
+      "step": 3031
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2398843930635838e-05,
+      "loss": 5.7387,
+      "step": 3032
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2390173410404624e-05,
+      "loss": 5.8718,
+      "step": 3033
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2381502890173411e-05,
+      "loss": 5.6787,
+      "step": 3034
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2381502890173411e-05,
+      "loss": 5.7652,
+      "step": 3035
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2372832369942197e-05,
+      "loss": 5.8762,
+      "step": 3036
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.2364161849710982e-05,
+      "loss": 5.6691,
+      "step": 3037
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2355491329479768e-05,
+      "loss": 5.6802,
+      "step": 3038
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2346820809248556e-05,
+      "loss": 5.7367,
+      "step": 3039
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2338150289017342e-05,
+      "loss": 5.7739,
+      "step": 3040
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2329479768786128e-05,
+      "loss": 5.4654,
+      "step": 3041
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2320809248554914e-05,
+      "loss": 5.5281,
+      "step": 3042
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.23121387283237e-05,
+      "loss": 5.5844,
+      "step": 3043
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2303468208092486e-05,
+      "loss": 5.7248,
+      "step": 3044
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2294797687861272e-05,
+      "loss": 5.5293,
+      "step": 3045
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.2286127167630058e-05,
+      "loss": 5.5775,
+      "step": 3046
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2277456647398846e-05,
+      "loss": 5.3979,
+      "step": 3047
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.226878612716763e-05,
+      "loss": 5.6615,
+      "step": 3048
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2260115606936416e-05,
+      "loss": 5.4493,
+      "step": 3049
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2251445086705202e-05,
+      "loss": 5.6367,
+      "step": 3050
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.224277456647399e-05,
+      "loss": 5.6763,
+      "step": 3051
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2234104046242776e-05,
+      "loss": 5.628,
+      "step": 3052
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.222543352601156e-05,
+      "loss": 5.6526,
+      "step": 3053
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2216763005780346e-05,
+      "loss": 5.6357,
+      "step": 3054
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.2208092485549134e-05,
+      "loss": 5.5822,
+      "step": 3055
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.219942196531792e-05,
+      "loss": 5.3459,
+      "step": 3056
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.2190751445086706e-05,
+      "loss": 5.7369,
+      "step": 3057
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.218208092485549e-05,
+      "loss": 5.5875,
+      "step": 3058
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.2173410404624278e-05,
+      "loss": 5.4312,
+      "step": 3059
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.2164739884393064e-05,
+      "loss": 5.5474,
+      "step": 3060
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.215606936416185e-05,
+      "loss": 5.858,
+      "step": 3061
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.2147398843930636e-05,
+      "loss": 5.5917,
+      "step": 3062
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.2138728323699422e-05,
+      "loss": 5.406,
+      "step": 3063
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.2130057803468208e-05,
+      "loss": 5.7969,
+      "step": 3064
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.2121387283236994e-05,
+      "loss": 5.6181,
+      "step": 3065
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.211271676300578e-05,
+      "loss": 5.4218,
+      "step": 3066
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.2104046242774568e-05,
+      "loss": 5.5227,
+      "step": 3067
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.2095375722543354e-05,
+      "loss": 5.5388,
+      "step": 3068
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.2086705202312138e-05,
+      "loss": 5.639,
+      "step": 3069
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.2078034682080924e-05,
+      "loss": 5.3934,
+      "step": 3070
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.2069364161849712e-05,
+      "loss": 5.314,
+      "step": 3071
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.2060693641618498e-05,
+      "loss": 5.4486,
+      "step": 3072
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.2052023121387284e-05,
+      "loss": 5.1007,
+      "step": 3073
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.2043352601156068e-05,
+      "loss": 5.3029,
+      "step": 3074
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.2034682080924856e-05,
+      "loss": 5.2348,
+      "step": 3075
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.2026011560693642e-05,
+      "loss": 4.9958,
+      "step": 3076
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.2017341040462428e-05,
+      "loss": 5.8186,
+      "step": 3077
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.2008670520231214e-05,
+      "loss": 5.974,
+      "step": 3078
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.2e-05,
+      "loss": 6.0019,
+      "step": 3079
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1991329479768786e-05,
+      "loss": 5.7854,
+      "step": 3080
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.1982658959537572e-05,
+      "loss": 5.7262,
+      "step": 3081
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1973988439306358e-05,
+      "loss": 5.8513,
+      "step": 3082
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1965317919075146e-05,
+      "loss": 5.7491,
+      "step": 3083
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.195664739884393e-05,
+      "loss": 5.6564,
+      "step": 3084
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1947976878612717e-05,
+      "loss": 5.6798,
+      "step": 3085
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1939306358381503e-05,
+      "loss": 5.7074,
+      "step": 3086
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.193063583815029e-05,
+      "loss": 5.7537,
+      "step": 3087
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1921965317919076e-05,
+      "loss": 5.6411,
+      "step": 3088
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.191329479768786e-05,
+      "loss": 5.627,
+      "step": 3089
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.1904624277456647e-05,
+      "loss": 5.5377,
+      "step": 3090
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1895953757225434e-05,
+      "loss": 5.5858,
+      "step": 3091
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.188728323699422e-05,
+      "loss": 5.5075,
+      "step": 3092
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1878612716763006e-05,
+      "loss": 5.5744,
+      "step": 3093
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1869942196531793e-05,
+      "loss": 5.6306,
+      "step": 3094
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1861271676300579e-05,
+      "loss": 5.5498,
+      "step": 3095
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1852601156069365e-05,
+      "loss": 5.7186,
+      "step": 3096
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.184393063583815e-05,
+      "loss": 5.726,
+      "step": 3097
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1835260115606937e-05,
+      "loss": 5.6807,
+      "step": 3098
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.1826589595375724e-05,
+      "loss": 5.5657,
+      "step": 3099
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1817919075144509e-05,
+      "loss": 5.4097,
+      "step": 3100
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1809248554913295e-05,
+      "loss": 5.6154,
+      "step": 3101
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.180057803468208e-05,
+      "loss": 5.5753,
+      "step": 3102
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1791907514450869e-05,
+      "loss": 5.6889,
+      "step": 3103
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1783236994219655e-05,
+      "loss": 5.5099,
+      "step": 3104
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1774566473988439e-05,
+      "loss": 5.6581,
+      "step": 3105
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1765895953757225e-05,
+      "loss": 5.5523,
+      "step": 3106
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1757225433526013e-05,
+      "loss": 5.6115,
+      "step": 3107
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.1748554913294799e-05,
+      "loss": 5.4777,
+      "step": 3108
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1739884393063585e-05,
+      "loss": 5.6418,
+      "step": 3109
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1731213872832369e-05,
+      "loss": 5.6885,
+      "step": 3110
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1722543352601157e-05,
+      "loss": 5.4585,
+      "step": 3111
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1713872832369943e-05,
+      "loss": 5.4977,
+      "step": 3112
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1705202312138729e-05,
+      "loss": 5.667,
+      "step": 3113
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1696531791907515e-05,
+      "loss": 5.5255,
+      "step": 3114
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1687861271676301e-05,
+      "loss": 5.5002,
+      "step": 3115
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1679190751445087e-05,
+      "loss": 5.3443,
+      "step": 3116
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.1670520231213873e-05,
+      "loss": 5.5492,
+      "step": 3117
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.1661849710982659e-05,
+      "loss": 5.5212,
+      "step": 3118
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.1653179190751445e-05,
+      "loss": 5.3944,
+      "step": 3119
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.1644508670520233e-05,
+      "loss": 5.4058,
+      "step": 3120
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.1635838150289017e-05,
+      "loss": 5.3455,
+      "step": 3121
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.1627167630057803e-05,
+      "loss": 5.3037,
+      "step": 3122
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.161849710982659e-05,
+      "loss": 5.628,
+      "step": 3123
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.1609826589595377e-05,
+      "loss": 5.1686,
+      "step": 3124
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.1601156069364163e-05,
+      "loss": 5.0519,
+      "step": 3125
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.1592485549132947e-05,
+      "loss": 5.11,
+      "step": 3126
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.1583815028901733e-05,
+      "loss": 5.9271,
+      "step": 3127
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.1575144508670521e-05,
+      "loss": 5.9688,
+      "step": 3128
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.1566473988439307e-05,
+      "loss": 5.952,
+      "step": 3129
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.1557803468208093e-05,
+      "loss": 5.7777,
+      "step": 3130
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.1549132947976877e-05,
+      "loss": 5.6022,
+      "step": 3131
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.1540462427745665e-05,
+      "loss": 5.7213,
+      "step": 3132
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.1531791907514451e-05,
+      "loss": 5.9327,
+      "step": 3133
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.1523121387283237e-05,
+      "loss": 5.856,
+      "step": 3134
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.1514450867052023e-05,
+      "loss": 5.694,
+      "step": 3135
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.150578034682081e-05,
+      "loss": 5.7436,
+      "step": 3136
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.1497109826589595e-05,
+      "loss": 5.8559,
+      "step": 3137
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.1488439306358381e-05,
+      "loss": 5.6893,
+      "step": 3138
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.1479768786127167e-05,
+      "loss": 5.5717,
+      "step": 3139
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.1471098265895955e-05,
+      "loss": 5.6533,
+      "step": 3140
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.146242774566474e-05,
+      "loss": 5.7293,
+      "step": 3141
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.1453757225433526e-05,
+      "loss": 5.7048,
+      "step": 3142
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.1445086705202312e-05,
+      "loss": 5.7593,
+      "step": 3143
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.14364161849711e-05,
+      "loss": 5.6287,
+      "step": 3144
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.1427745664739885e-05,
+      "loss": 5.5643,
+      "step": 3145
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.1419075144508671e-05,
+      "loss": 5.7273,
+      "step": 3146
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.1410404624277456e-05,
+      "loss": 5.5071,
+      "step": 3147
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.1401734104046243e-05,
+      "loss": 5.6757,
+      "step": 3148
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.139306358381503e-05,
+      "loss": 5.6321,
+      "step": 3149
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.1384393063583815e-05,
+      "loss": 5.571,
+      "step": 3150
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.1375722543352602e-05,
+      "loss": 5.5227,
+      "step": 3151
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.1367052023121388e-05,
+      "loss": 5.5668,
+      "step": 3152
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.1358381502890174e-05,
+      "loss": 5.5278,
+      "step": 3153
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.134971098265896e-05,
+      "loss": 5.6379,
+      "step": 3154
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.1341040462427746e-05,
+      "loss": 5.563,
+      "step": 3155
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.1332369942196533e-05,
+      "loss": 5.6754,
+      "step": 3156
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.1323699421965318e-05,
+      "loss": 5.6001,
+      "step": 3157
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.1315028901734104e-05,
+      "loss": 5.4656,
+      "step": 3158
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.130635838150289e-05,
+      "loss": 5.5101,
+      "step": 3159
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.1297687861271678e-05,
+      "loss": 5.616,
+      "step": 3160
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.1289017341040464e-05,
+      "loss": 5.6909,
+      "step": 3161
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.1280346820809248e-05,
+      "loss": 5.3755,
+      "step": 3162
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.1271676300578034e-05,
+      "loss": 5.5321,
+      "step": 3163
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.1263005780346822e-05,
+      "loss": 5.3617,
+      "step": 3164
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.1254335260115608e-05,
+      "loss": 5.5978,
+      "step": 3165
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.1245664739884394e-05,
+      "loss": 5.5225,
+      "step": 3166
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.1236994219653178e-05,
+      "loss": 5.5617,
+      "step": 3167
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.1228323699421966e-05,
+      "loss": 5.3622,
+      "step": 3168
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.1219653179190752e-05,
+      "loss": 5.3402,
+      "step": 3169
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.1210982658959538e-05,
+      "loss": 5.2446,
+      "step": 3170
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.1202312138728324e-05,
+      "loss": 5.4,
+      "step": 3171
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.119364161849711e-05,
+      "loss": 5.3239,
+      "step": 3172
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.1184971098265896e-05,
+      "loss": 5.1082,
+      "step": 3173
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.1176300578034682e-05,
+      "loss": 5.3441,
+      "step": 3174
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.1167630057803468e-05,
+      "loss": 5.1152,
+      "step": 3175
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.1158959537572256e-05,
+      "loss": 4.8902,
+      "step": 3176
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.1150289017341042e-05,
+      "loss": 5.7921,
+      "step": 3177
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.1141618497109826e-05,
+      "loss": 5.742,
+      "step": 3178
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.1132947976878612e-05,
+      "loss": 5.6973,
+      "step": 3179
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.11242774566474e-05,
+      "loss": 5.9033,
+      "step": 3180
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.1115606936416186e-05,
+      "loss": 5.8872,
+      "step": 3181
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.1106936416184972e-05,
+      "loss": 5.7765,
+      "step": 3182
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.1098265895953756e-05,
+      "loss": 5.7492,
+      "step": 3183
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.1089595375722544e-05,
+      "loss": 5.5967,
+      "step": 3184
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.108092485549133e-05,
+      "loss": 5.6828,
+      "step": 3185
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.1072254335260116e-05,
+      "loss": 5.5791,
+      "step": 3186
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.1063583815028902e-05,
+      "loss": 5.6585,
+      "step": 3187
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.1054913294797688e-05,
+      "loss": 5.5961,
+      "step": 3188
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.1046242774566474e-05,
+      "loss": 5.4987,
+      "step": 3189
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.103757225433526e-05,
+      "loss": 5.5074,
+      "step": 3190
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.1028901734104046e-05,
+      "loss": 5.5923,
+      "step": 3191
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.1020231213872834e-05,
+      "loss": 5.7602,
+      "step": 3192
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.1011560693641618e-05,
+      "loss": 5.7839,
+      "step": 3193
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.1002890173410404e-05,
+      "loss": 5.4832,
+      "step": 3194
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.099421965317919e-05,
+      "loss": 5.4421,
+      "step": 3195
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.0985549132947978e-05,
+      "loss": 5.6083,
+      "step": 3196
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.0976878612716764e-05,
+      "loss": 5.6291,
+      "step": 3197
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.0968208092485549e-05,
+      "loss": 5.699,
+      "step": 3198
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.0959537572254335e-05,
+      "loss": 5.5122,
+      "step": 3199
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.0950867052023122e-05,
+      "loss": 5.633,
+      "step": 3200
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.0942196531791908e-05,
+      "loss": 5.5618,
+      "step": 3201
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.0933526011560694e-05,
+      "loss": 5.6074,
+      "step": 3202
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.092485549132948e-05,
+      "loss": 5.562,
+      "step": 3203
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.0916184971098266e-05,
+      "loss": 5.4524,
+      "step": 3204
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.0907514450867052e-05,
+      "loss": 5.423,
+      "step": 3205
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.0898843930635838e-05,
+      "loss": 5.7104,
+      "step": 3206
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.0890173410404625e-05,
+      "loss": 5.6318,
+      "step": 3207
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.0881502890173412e-05,
+      "loss": 5.5891,
+      "step": 3208
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.0872832369942197e-05,
+      "loss": 5.5275,
+      "step": 3209
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.0864161849710983e-05,
+      "loss": 5.4398,
+      "step": 3210
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.0855491329479769e-05,
+      "loss": 5.4696,
+      "step": 3211
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.0846820809248556e-05,
+      "loss": 5.6052,
+      "step": 3212
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.0838150289017342e-05,
+      "loss": 5.4387,
+      "step": 3213
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.0829479768786127e-05,
+      "loss": 5.4683,
+      "step": 3214
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.0820809248554913e-05,
+      "loss": 5.4882,
+      "step": 3215
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.0812138728323699e-05,
+      "loss": 5.7377,
+      "step": 3216
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.0803468208092487e-05,
+      "loss": 5.4334,
+      "step": 3217
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.0794797687861273e-05,
+      "loss": 5.2632,
+      "step": 3218
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.0786127167630057e-05,
+      "loss": 5.4831,
+      "step": 3219
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.0777456647398843e-05,
+      "loss": 5.2438,
+      "step": 3220
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.076878612716763e-05,
+      "loss": 5.4429,
+      "step": 3221
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.0760115606936417e-05,
+      "loss": 5.3507,
+      "step": 3222
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.0751445086705203e-05,
+      "loss": 5.1922,
+      "step": 3223
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.0742774566473987e-05,
+      "loss": 5.1152,
+      "step": 3224
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.0734104046242775e-05,
+      "loss": 5.3503,
+      "step": 3225
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.072543352601156e-05,
+      "loss": 4.7656,
+      "step": 3226
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.0716763005780347e-05,
+      "loss": 5.9204,
+      "step": 3227
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.0708092485549133e-05,
+      "loss": 6.1084,
+      "step": 3228
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.069942196531792e-05,
+      "loss": 5.8536,
+      "step": 3229
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.0690751445086705e-05,
+      "loss": 5.5421,
+      "step": 3230
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.0682080924855491e-05,
+      "loss": 5.6363,
+      "step": 3231
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.0673410404624277e-05,
+      "loss": 5.7784,
+      "step": 3232
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.0664739884393065e-05,
+      "loss": 5.7341,
+      "step": 3233
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.065606936416185e-05,
+      "loss": 5.6498,
+      "step": 3234
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.0647398843930635e-05,
+      "loss": 5.7545,
+      "step": 3235
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.0638728323699421e-05,
+      "loss": 5.6899,
+      "step": 3236
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.0630057803468209e-05,
+      "loss": 5.7626,
+      "step": 3237
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.0621387283236995e-05,
+      "loss": 5.7104,
+      "step": 3238
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.0612716763005781e-05,
+      "loss": 5.6912,
+      "step": 3239
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.0604046242774565e-05,
+      "loss": 5.4971,
+      "step": 3240
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.0595375722543353e-05,
+      "loss": 5.5994,
+      "step": 3241
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.0586705202312139e-05,
+      "loss": 5.7041,
+      "step": 3242
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.0578034682080925e-05,
+      "loss": 5.4435,
+      "step": 3243
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.0569364161849711e-05,
+      "loss": 5.5153,
+      "step": 3244
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.0560693641618497e-05,
+      "loss": 5.4923,
+      "step": 3245
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.0552023121387283e-05,
+      "loss": 5.5449,
+      "step": 3246
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.054335260115607e-05,
+      "loss": 5.5867,
+      "step": 3247
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.0534682080924855e-05,
+      "loss": 5.4546,
+      "step": 3248
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.0526011560693643e-05,
+      "loss": 5.6559,
+      "step": 3249
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.0517341040462427e-05,
+      "loss": 5.7721,
+      "step": 3250
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.0508670520231213e-05,
+      "loss": 5.6395,
+      "step": 3251
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.05e-05,
+      "loss": 5.6019,
+      "step": 3252
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.0491329479768787e-05,
+      "loss": 5.4498,
+      "step": 3253
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.0482658959537573e-05,
+      "loss": 5.5684,
+      "step": 3254
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.047398843930636e-05,
+      "loss": 5.7895,
+      "step": 3255
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.0465317919075144e-05,
+      "loss": 5.4451,
+      "step": 3256
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.0456647398843931e-05,
+      "loss": 5.7652,
+      "step": 3257
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.0447976878612717e-05,
+      "loss": 5.5397,
+      "step": 3258
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.0439306358381503e-05,
+      "loss": 5.6289,
+      "step": 3259
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.043063583815029e-05,
+      "loss": 5.5504,
+      "step": 3260
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.0421965317919075e-05,
+      "loss": 5.693,
+      "step": 3261
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.0413294797687861e-05,
+      "loss": 5.3476,
+      "step": 3262
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.0404624277456647e-05,
+      "loss": 5.4788,
+      "step": 3263
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.0395953757225434e-05,
+      "loss": 5.5461,
+      "step": 3264
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.0387283236994221e-05,
+      "loss": 5.4592,
+      "step": 3265
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.0378612716763006e-05,
+      "loss": 5.3356,
+      "step": 3266
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.0369942196531792e-05,
+      "loss": 5.6962,
+      "step": 3267
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.0361271676300578e-05,
+      "loss": 5.0594,
+      "step": 3268
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.0352601156069365e-05,
+      "loss": 5.4772,
+      "step": 3269
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.0343930635838151e-05,
+      "loss": 5.5389,
+      "step": 3270
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.0335260115606936e-05,
+      "loss": 5.5507,
+      "step": 3271
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.0326589595375722e-05,
+      "loss": 5.1602,
+      "step": 3272
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.031791907514451e-05,
+      "loss": 5.169,
+      "step": 3273
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.0309248554913296e-05,
+      "loss": 5.2393,
+      "step": 3274
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.0300578034682082e-05,
+      "loss": 5.2815,
+      "step": 3275
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.0291907514450866e-05,
+      "loss": 4.9484,
+      "step": 3276
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.0283236994219654e-05,
+      "loss": 6.1248,
+      "step": 3277
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.027456647398844e-05,
+      "loss": 5.6994,
+      "step": 3278
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.0265895953757226e-05,
+      "loss": 5.7342,
+      "step": 3279
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.0257225433526012e-05,
+      "loss": 5.8371,
+      "step": 3280
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.02485549132948e-05,
+      "loss": 5.7241,
+      "step": 3281
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.0239884393063584e-05,
+      "loss": 5.5694,
+      "step": 3282
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.023121387283237e-05,
+      "loss": 5.7798,
+      "step": 3283
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.0222543352601156e-05,
+      "loss": 5.6896,
+      "step": 3284
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.0213872832369944e-05,
+      "loss": 5.6491,
+      "step": 3285
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.020520231213873e-05,
+      "loss": 5.7205,
+      "step": 3286
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.0196531791907514e-05,
+      "loss": 5.7268,
+      "step": 3287
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.01878612716763e-05,
+      "loss": 5.4796,
+      "step": 3288
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.0179190751445088e-05,
+      "loss": 5.6932,
+      "step": 3289
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.0170520231213874e-05,
+      "loss": 5.5444,
+      "step": 3290
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.016184971098266e-05,
+      "loss": 5.3382,
+      "step": 3291
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.0153179190751444e-05,
+      "loss": 5.6928,
+      "step": 3292
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.0144508670520232e-05,
+      "loss": 5.5849,
+      "step": 3293
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.0135838150289018e-05,
+      "loss": 5.5088,
+      "step": 3294
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.0127167630057804e-05,
+      "loss": 5.58,
+      "step": 3295
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.011849710982659e-05,
+      "loss": 5.5318,
+      "step": 3296
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.0109826589595376e-05,
+      "loss": 5.4011,
+      "step": 3297
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.0101156069364162e-05,
+      "loss": 5.679,
+      "step": 3298
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.0092485549132948e-05,
+      "loss": 5.5461,
+      "step": 3299
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.0083815028901734e-05,
+      "loss": 5.4849,
+      "step": 3300
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.0075144508670522e-05,
+      "loss": 5.5411,
+      "step": 3301
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.0066473988439306e-05,
+      "loss": 5.6731,
+      "step": 3302
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.0057803468208092e-05,
+      "loss": 5.6884,
+      "step": 3303
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.0049132947976878e-05,
+      "loss": 5.5925,
+      "step": 3304
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.0040462427745666e-05,
+      "loss": 5.5572,
+      "step": 3305
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.0031791907514452e-05,
+      "loss": 5.4926,
+      "step": 3306
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.0023121387283238e-05,
+      "loss": 5.6307,
+      "step": 3307
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.0014450867052022e-05,
+      "loss": 5.5715,
+      "step": 3308
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.000578034682081e-05,
+      "loss": 5.3649,
+      "step": 3309
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.997109826589596e-06,
+      "loss": 5.3332,
+      "step": 3310
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.988439306358382e-06,
+      "loss": 5.4992,
+      "step": 3311
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.979768786127168e-06,
+      "loss": 5.5227,
+      "step": 3312
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.971098265895953e-06,
+      "loss": 5.5136,
+      "step": 3313
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.96242774566474e-06,
+      "loss": 5.5142,
+      "step": 3314
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.953757225433526e-06,
+      "loss": 5.5193,
+      "step": 3315
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.945086705202312e-06,
+      "loss": 5.4802,
+      "step": 3316
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.936416184971098e-06,
+      "loss": 5.4288,
+      "step": 3317
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.927745664739884e-06,
+      "loss": 5.5119,
+      "step": 3318
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.91907514450867e-06,
+      "loss": 5.612,
+      "step": 3319
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.910404624277457e-06,
+      "loss": 5.429,
+      "step": 3320
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.901734104046243e-06,
+      "loss": 5.3,
+      "step": 3321
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.89306358381503e-06,
+      "loss": 5.3607,
+      "step": 3322
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 9.884393063583815e-06,
+      "loss": 4.9827,
+      "step": 3323
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 9.8757225433526e-06,
+      "loss": 5.4008,
+      "step": 3324
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 9.867052023121387e-06,
+      "loss": 5.148,
+      "step": 3325
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 9.858381502890174e-06,
+      "loss": 4.9032,
+      "step": 3326
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 9.84971098265896e-06,
+      "loss": 5.7864,
+      "step": 3327
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 9.841040462427745e-06,
+      "loss": 5.6067,
+      "step": 3328
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 9.83236994219653e-06,
+      "loss": 5.9594,
+      "step": 3329
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 9.823699421965319e-06,
+      "loss": 5.8329,
+      "step": 3330
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 9.815028901734105e-06,
+      "loss": 5.7011,
+      "step": 3331
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 9.80635838150289e-06,
+      "loss": 5.7911,
+      "step": 3332
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 9.797687861271677e-06,
+      "loss": 5.8068,
+      "step": 3333
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 9.789017341040463e-06,
+      "loss": 5.6477,
+      "step": 3334
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 9.780346820809249e-06,
+      "loss": 5.7618,
+      "step": 3335
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 9.771676300578035e-06,
+      "loss": 5.6346,
+      "step": 3336
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 9.76300578034682e-06,
+      "loss": 5.6266,
+      "step": 3337
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 9.754335260115609e-06,
+      "loss": 5.5229,
+      "step": 3338
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 9.745664739884393e-06,
+      "loss": 5.7211,
+      "step": 3339
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 9.736994219653179e-06,
+      "loss": 5.558,
+      "step": 3340
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 9.728323699421965e-06,
+      "loss": 5.7719,
+      "step": 3341
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 9.719653179190753e-06,
+      "loss": 5.4744,
+      "step": 3342
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 9.710982658959539e-06,
+      "loss": 5.5841,
+      "step": 3343
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 9.702312138728323e-06,
+      "loss": 5.4243,
+      "step": 3344
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 9.693641618497109e-06,
+      "loss": 5.6836,
+      "step": 3345
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 9.684971098265897e-06,
+      "loss": 5.5375,
+      "step": 3346
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 9.676300578034683e-06,
+      "loss": 5.5239,
+      "step": 3347
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 9.667630057803469e-06,
+      "loss": 5.448,
+      "step": 3348
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 9.658959537572253e-06,
+      "loss": 5.3941,
+      "step": 3349
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 9.650289017341041e-06,
+      "loss": 5.489,
+      "step": 3350
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 9.641618497109827e-06,
+      "loss": 5.5909,
+      "step": 3351
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 9.632947976878613e-06,
+      "loss": 5.6812,
+      "step": 3352
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 9.624277456647399e-06,
+      "loss": 5.6058,
+      "step": 3353
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 9.615606936416185e-06,
+      "loss": 5.7068,
+      "step": 3354
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 9.606936416184971e-06,
+      "loss": 5.6959,
+      "step": 3355
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 9.598265895953757e-06,
+      "loss": 5.5311,
+      "step": 3356
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 9.589595375722543e-06,
+      "loss": 5.7124,
+      "step": 3357
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 9.580924855491331e-06,
+      "loss": 5.3961,
+      "step": 3358
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 9.572254335260117e-06,
+      "loss": 5.482,
+      "step": 3359
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 9.563583815028901e-06,
+      "loss": 5.4687,
+      "step": 3360
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 9.554913294797687e-06,
+      "loss": 5.5163,
+      "step": 3361
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 9.546242774566475e-06,
+      "loss": 5.4843,
+      "step": 3362
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 9.537572254335261e-06,
+      "loss": 5.4397,
+      "step": 3363
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 9.528901734104047e-06,
+      "loss": 5.4796,
+      "step": 3364
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 9.520231213872831e-06,
+      "loss": 5.6783,
+      "step": 3365
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 9.511560693641619e-06,
+      "loss": 5.4836,
+      "step": 3366
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 9.502890173410405e-06,
+      "loss": 5.6436,
+      "step": 3367
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 9.494219653179191e-06,
+      "loss": 5.5662,
+      "step": 3368
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 9.485549132947977e-06,
+      "loss": 5.4997,
+      "step": 3369
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 9.476878612716763e-06,
+      "loss": 5.3907,
+      "step": 3370
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 9.46820809248555e-06,
+      "loss": 5.2503,
+      "step": 3371
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 9.459537572254335e-06,
+      "loss": 4.9228,
+      "step": 3372
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 9.450867052023121e-06,
+      "loss": 5.181,
+      "step": 3373
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 9.442196531791909e-06,
+      "loss": 5.1699,
+      "step": 3374
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 9.433526011560693e-06,
+      "loss": 5.221,
+      "step": 3375
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 9.42485549132948e-06,
+      "loss": 4.73,
+      "step": 3376
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.416184971098266e-06,
+      "loss": 5.6821,
+      "step": 3377
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.407514450867053e-06,
+      "loss": 5.8673,
+      "step": 3378
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.39884393063584e-06,
+      "loss": 5.86,
+      "step": 3379
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.390173410404624e-06,
+      "loss": 5.8991,
+      "step": 3380
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.38150289017341e-06,
+      "loss": 5.6999,
+      "step": 3381
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.372832369942197e-06,
+      "loss": 5.6251,
+      "step": 3382
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.364161849710983e-06,
+      "loss": 5.8161,
+      "step": 3383
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.35549132947977e-06,
+      "loss": 5.666,
+      "step": 3384
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.346820809248555e-06,
+      "loss": 5.7066,
+      "step": 3385
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.338150289017342e-06,
+      "loss": 5.5966,
+      "step": 3386
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.329479768786128e-06,
+      "loss": 5.5349,
+      "step": 3387
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.320809248554914e-06,
+      "loss": 5.6323,
+      "step": 3388
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.3121387283237e-06,
+      "loss": 5.4048,
+      "step": 3389
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.303468208092487e-06,
+      "loss": 5.4749,
+      "step": 3390
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.294797687861272e-06,
+      "loss": 5.5911,
+      "step": 3391
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.286127167630058e-06,
+      "loss": 5.6177,
+      "step": 3392
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.277456647398844e-06,
+      "loss": 5.505,
+      "step": 3393
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.268786127167631e-06,
+      "loss": 5.3547,
+      "step": 3394
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 9.260115606936418e-06,
+      "loss": 5.5391,
+      "step": 3395
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 9.251445086705202e-06,
+      "loss": 5.6448,
+      "step": 3396
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 9.242774566473988e-06,
+      "loss": 5.5639,
+      "step": 3397
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 9.234104046242776e-06,
+      "loss": 5.5724,
+      "step": 3398
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 9.225433526011562e-06,
+      "loss": 5.4731,
+      "step": 3399
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 9.216763005780348e-06,
+      "loss": 5.5266,
+      "step": 3400
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 9.208092485549132e-06,
+      "loss": 5.5887,
+      "step": 3401
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 9.19942196531792e-06,
+      "loss": 5.395,
+      "step": 3402
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 9.190751445086706e-06,
+      "loss": 5.5672,
+      "step": 3403
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 9.182080924855492e-06,
+      "loss": 5.2987,
+      "step": 3404
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 9.173410404624278e-06,
+      "loss": 5.4032,
+      "step": 3405
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 9.164739884393062e-06,
+      "loss": 5.5088,
+      "step": 3406
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 9.15606936416185e-06,
+      "loss": 5.5319,
+      "step": 3407
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 9.147398843930636e-06,
+      "loss": 5.4853,
+      "step": 3408
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 9.138728323699422e-06,
+      "loss": 5.6006,
+      "step": 3409
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 9.130057803468208e-06,
+      "loss": 5.5714,
+      "step": 3410
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 9.121387283236994e-06,
+      "loss": 5.2823,
+      "step": 3411
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.11271676300578e-06,
+      "loss": 5.4126,
+      "step": 3412
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.104046242774566e-06,
+      "loss": 5.673,
+      "step": 3413
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.095375722543352e-06,
+      "loss": 5.5772,
+      "step": 3414
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.08670520231214e-06,
+      "loss": 5.299,
+      "step": 3415
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.078034682080926e-06,
+      "loss": 5.2596,
+      "step": 3416
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.06936416184971e-06,
+      "loss": 5.3817,
+      "step": 3417
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.060693641618496e-06,
+      "loss": 5.284,
+      "step": 3418
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.052023121387284e-06,
+      "loss": 5.321,
+      "step": 3419
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.04335260115607e-06,
+      "loss": 5.5135,
+      "step": 3420
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 9.034682080924856e-06,
+      "loss": 5.3163,
+      "step": 3421
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 9.02601156069364e-06,
+      "loss": 5.5079,
+      "step": 3422
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 9.017341040462428e-06,
+      "loss": 5.275,
+      "step": 3423
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 9.008670520231214e-06,
+      "loss": 5.2395,
+      "step": 3424
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 9e-06,
+      "loss": 5.0462,
+      "step": 3425
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.991329479768786e-06,
+      "loss": 4.9179,
+      "step": 3426
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.982658959537572e-06,
+      "loss": 5.8885,
+      "step": 3427
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.973988439306358e-06,
+      "loss": 5.6934,
+      "step": 3428
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.965317919075144e-06,
+      "loss": 5.6682,
+      "step": 3429
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.95664739884393e-06,
+      "loss": 5.8066,
+      "step": 3430
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.947976878612718e-06,
+      "loss": 5.6061,
+      "step": 3431
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.939306358381502e-06,
+      "loss": 5.6719,
+      "step": 3432
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.930635838150288e-06,
+      "loss": 5.6758,
+      "step": 3433
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.921965317919075e-06,
+      "loss": 5.7598,
+      "step": 3434
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.913294797687862e-06,
+      "loss": 5.634,
+      "step": 3435
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.904624277456648e-06,
+      "loss": 5.4991,
+      "step": 3436
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.895953757225433e-06,
+      "loss": 5.4845,
+      "step": 3437
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 8.887283236994219e-06,
+      "loss": 5.5317,
+      "step": 3438
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.878612716763006e-06,
+      "loss": 5.6291,
+      "step": 3439
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.869942196531792e-06,
+      "loss": 5.669,
+      "step": 3440
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.861271676300578e-06,
+      "loss": 5.4467,
+      "step": 3441
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.852601156069364e-06,
+      "loss": 5.4621,
+      "step": 3442
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.84393063583815e-06,
+      "loss": 5.6147,
+      "step": 3443
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.835260115606937e-06,
+      "loss": 5.6472,
+      "step": 3444
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.826589595375723e-06,
+      "loss": 5.4466,
+      "step": 3445
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.817919075144509e-06,
+      "loss": 5.4615,
+      "step": 3446
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 8.809248554913296e-06,
+      "loss": 5.7464,
+      "step": 3447
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.80057803468208e-06,
+      "loss": 5.464,
+      "step": 3448
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.791907514450867e-06,
+      "loss": 5.5783,
+      "step": 3449
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.783236994219653e-06,
+      "loss": 5.4568,
+      "step": 3450
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.77456647398844e-06,
+      "loss": 5.5032,
+      "step": 3451
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.765895953757227e-06,
+      "loss": 5.5118,
+      "step": 3452
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.757225433526011e-06,
+      "loss": 5.4459,
+      "step": 3453
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.748554913294797e-06,
+      "loss": 5.4365,
+      "step": 3454
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.739884393063585e-06,
+      "loss": 5.5142,
+      "step": 3455
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 8.73121387283237e-06,
+      "loss": 5.5697,
+      "step": 3456
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.722543352601157e-06,
+      "loss": 5.373,
+      "step": 3457
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.713872832369941e-06,
+      "loss": 5.4767,
+      "step": 3458
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.705202312138729e-06,
+      "loss": 5.4697,
+      "step": 3459
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.696531791907515e-06,
+      "loss": 5.4302,
+      "step": 3460
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.6878612716763e-06,
+      "loss": 5.4301,
+      "step": 3461
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.679190751445087e-06,
+      "loss": 5.3418,
+      "step": 3462
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.670520231213873e-06,
+      "loss": 5.6869,
+      "step": 3463
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.661849710982659e-06,
+      "loss": 5.5117,
+      "step": 3464
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 8.653179190751445e-06,
+      "loss": 5.5476,
+      "step": 3465
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.644508670520231e-06,
+      "loss": 5.3168,
+      "step": 3466
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.635838150289019e-06,
+      "loss": 5.5367,
+      "step": 3467
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.627167630057805e-06,
+      "loss": 5.4535,
+      "step": 3468
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.618497109826589e-06,
+      "loss": 5.0921,
+      "step": 3469
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.609826589595375e-06,
+      "loss": 5.2822,
+      "step": 3470
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.601156069364163e-06,
+      "loss": 5.1608,
+      "step": 3471
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.592485549132949e-06,
+      "loss": 5.0344,
+      "step": 3472
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.583815028901735e-06,
+      "loss": 5.2681,
+      "step": 3473
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 8.57514450867052e-06,
+      "loss": 5.095,
+      "step": 3474
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.566473988439307e-06,
+      "loss": 4.7826,
+      "step": 3475
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.557803468208093e-06,
+      "loss": 4.6893,
+      "step": 3476
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.549132947976879e-06,
+      "loss": 5.9106,
+      "step": 3477
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.540462427745665e-06,
+      "loss": 5.7778,
+      "step": 3478
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.531791907514451e-06,
+      "loss": 5.8266,
+      "step": 3479
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.523121387283237e-06,
+      "loss": 5.7697,
+      "step": 3480
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.514450867052023e-06,
+      "loss": 5.7253,
+      "step": 3481
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.50578034682081e-06,
+      "loss": 5.5624,
+      "step": 3482
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 8.497109826589597e-06,
+      "loss": 5.5537,
+      "step": 3483
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.488439306358381e-06,
+      "loss": 5.779,
+      "step": 3484
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.479768786127167e-06,
+      "loss": 5.7908,
+      "step": 3485
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.471098265895953e-06,
+      "loss": 5.6212,
+      "step": 3486
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.462427745664741e-06,
+      "loss": 5.6012,
+      "step": 3487
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.453757225433527e-06,
+      "loss": 5.5634,
+      "step": 3488
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.445086705202311e-06,
+      "loss": 5.6714,
+      "step": 3489
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.436416184971098e-06,
+      "loss": 5.696,
+      "step": 3490
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.427745664739885e-06,
+      "loss": 5.5268,
+      "step": 3491
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 8.419075144508671e-06,
+      "loss": 5.5505,
+      "step": 3492
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.410404624277457e-06,
+      "loss": 5.5611,
+      "step": 3493
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.401734104046243e-06,
+      "loss": 5.4414,
+      "step": 3494
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.39306358381503e-06,
+      "loss": 5.7505,
+      "step": 3495
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.384393063583815e-06,
+      "loss": 5.4042,
+      "step": 3496
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.375722543352601e-06,
+      "loss": 5.44,
+      "step": 3497
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.367052023121387e-06,
+      "loss": 5.5623,
+      "step": 3498
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.358381502890175e-06,
+      "loss": 5.6444,
+      "step": 3499
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.34971098265896e-06,
+      "loss": 5.614,
+      "step": 3500
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 8.341040462427746e-06,
+      "loss": 5.5986,
+      "step": 3501
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.332369942196532e-06,
+      "loss": 5.5144,
+      "step": 3502
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.323699421965318e-06,
+      "loss": 5.4865,
+      "step": 3503
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.315028901734105e-06,
+      "loss": 5.4893,
+      "step": 3504
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.30635838150289e-06,
+      "loss": 5.5918,
+      "step": 3505
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.297687861271676e-06,
+      "loss": 5.3655,
+      "step": 3506
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.289017341040462e-06,
+      "loss": 5.4239,
+      "step": 3507
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.28034682080925e-06,
+      "loss": 5.3287,
+      "step": 3508
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.271676300578036e-06,
+      "loss": 5.3553,
+      "step": 3509
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 8.26300578034682e-06,
+      "loss": 5.4762,
+      "step": 3510
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.254335260115606e-06,
+      "loss": 5.5153,
+      "step": 3511
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.245664739884394e-06,
+      "loss": 5.403,
+      "step": 3512
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.23699421965318e-06,
+      "loss": 5.4356,
+      "step": 3513
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.228323699421966e-06,
+      "loss": 5.3369,
+      "step": 3514
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.21965317919075e-06,
+      "loss": 5.5507,
+      "step": 3515
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.210982658959538e-06,
+      "loss": 5.4163,
+      "step": 3516
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.202312138728324e-06,
+      "loss": 5.3608,
+      "step": 3517
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 8.19364161849711e-06,
+      "loss": 5.4609,
+      "step": 3518
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.184971098265896e-06,
+      "loss": 5.3087,
+      "step": 3519
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.176300578034684e-06,
+      "loss": 5.1929,
+      "step": 3520
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.167630057803468e-06,
+      "loss": 5.3111,
+      "step": 3521
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.158959537572254e-06,
+      "loss": 5.3876,
+      "step": 3522
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.15028901734104e-06,
+      "loss": 5.1074,
+      "step": 3523
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.141618497109828e-06,
+      "loss": 4.9495,
+      "step": 3524
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.132947976878614e-06,
+      "loss": 4.6214,
+      "step": 3525
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.124277456647398e-06,
+      "loss": 4.8187,
+      "step": 3526
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.115606936416184e-06,
+      "loss": 5.821,
+      "step": 3527
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.106936416184972e-06,
+      "loss": 5.6868,
+      "step": 3528
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.098265895953758e-06,
+      "loss": 5.7121,
+      "step": 3529
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.089595375722544e-06,
+      "loss": 5.6924,
+      "step": 3530
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.080924855491328e-06,
+      "loss": 5.5363,
+      "step": 3531
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.072254335260116e-06,
+      "loss": 5.591,
+      "step": 3532
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.063583815028902e-06,
+      "loss": 5.5238,
+      "step": 3533
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.054913294797688e-06,
+      "loss": 5.6523,
+      "step": 3534
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.046242774566474e-06,
+      "loss": 5.5083,
+      "step": 3535
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 8.03757225433526e-06,
+      "loss": 5.5706,
+      "step": 3536
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.028901734104046e-06,
+      "loss": 5.6409,
+      "step": 3537
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.020231213872832e-06,
+      "loss": 5.5604,
+      "step": 3538
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.011560693641618e-06,
+      "loss": 5.4895,
+      "step": 3539
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 8.002890173410406e-06,
+      "loss": 5.5668,
+      "step": 3540
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.99421965317919e-06,
+      "loss": 5.5715,
+      "step": 3541
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.985549132947976e-06,
+      "loss": 5.5666,
+      "step": 3542
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.976878612716762e-06,
+      "loss": 5.7126,
+      "step": 3543
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.96820809248555e-06,
+      "loss": 5.6439,
+      "step": 3544
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 7.959537572254336e-06,
+      "loss": 5.4554,
+      "step": 3545
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.950867052023122e-06,
+      "loss": 5.5298,
+      "step": 3546
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.942196531791907e-06,
+      "loss": 5.5417,
+      "step": 3547
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.933526011560694e-06,
+      "loss": 5.4646,
+      "step": 3548
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.92485549132948e-06,
+      "loss": 5.4887,
+      "step": 3549
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.916184971098266e-06,
+      "loss": 5.4271,
+      "step": 3550
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.907514450867052e-06,
+      "loss": 5.6838,
+      "step": 3551
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.898843930635838e-06,
+      "loss": 5.6321,
+      "step": 3552
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.890173410404624e-06,
+      "loss": 5.4512,
+      "step": 3553
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.88150289017341e-06,
+      "loss": 5.4568,
+      "step": 3554
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.872832369942196e-06,
+      "loss": 5.6132,
+      "step": 3555
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.864161849710984e-06,
+      "loss": 5.4101,
+      "step": 3556
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.855491329479769e-06,
+      "loss": 5.3861,
+      "step": 3557
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.846820809248555e-06,
+      "loss": 5.4457,
+      "step": 3558
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.83815028901734e-06,
+      "loss": 5.6077,
+      "step": 3559
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.829479768786128e-06,
+      "loss": 5.698,
+      "step": 3560
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.820809248554914e-06,
+      "loss": 5.4414,
+      "step": 3561
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.812138728323699e-06,
+      "loss": 5.4865,
+      "step": 3562
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.803468208092485e-06,
+      "loss": 5.208,
+      "step": 3563
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.794797687861272e-06,
+      "loss": 5.2664,
+      "step": 3564
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.786127167630059e-06,
+      "loss": 5.1079,
+      "step": 3565
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.777456647398845e-06,
+      "loss": 5.0525,
+      "step": 3566
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.768786127167629e-06,
+      "loss": 5.0169,
+      "step": 3567
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.760115606936417e-06,
+      "loss": 4.8933,
+      "step": 3568
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.751445086705203e-06,
+      "loss": 5.9399,
+      "step": 3569
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.742774566473989e-06,
+      "loss": 5.8718,
+      "step": 3570
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.734104046242775e-06,
+      "loss": 5.7029,
+      "step": 3571
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.72543352601156e-06,
+      "loss": 5.7177,
+      "step": 3572
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.716763005780347e-06,
+      "loss": 5.596,
+      "step": 3573
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.708092485549133e-06,
+      "loss": 5.7729,
+      "step": 3574
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.699421965317919e-06,
+      "loss": 5.6205,
+      "step": 3575
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.690751445086707e-06,
+      "loss": 5.6507,
+      "step": 3576
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.682080924855493e-06,
+      "loss": 5.5795,
+      "step": 3577
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.673410404624277e-06,
+      "loss": 5.509,
+      "step": 3578
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.664739884393063e-06,
+      "loss": 5.4501,
+      "step": 3579
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.65606936416185e-06,
+      "loss": 5.3618,
+      "step": 3580
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 7.647398843930637e-06,
+      "loss": 5.4645,
+      "step": 3581
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.638728323699423e-06,
+      "loss": 5.4466,
+      "step": 3582
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.630057803468207e-06,
+      "loss": 5.5472,
+      "step": 3583
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.621387283236995e-06,
+      "loss": 5.5897,
+      "step": 3584
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.612716763005781e-06,
+      "loss": 5.5282,
+      "step": 3585
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.604046242774567e-06,
+      "loss": 5.6499,
+      "step": 3586
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.595375722543352e-06,
+      "loss": 5.4691,
+      "step": 3587
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.58670520231214e-06,
+      "loss": 5.3952,
+      "step": 3588
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.578034682080925e-06,
+      "loss": 5.401,
+      "step": 3589
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 7.569364161849711e-06,
+      "loss": 5.5355,
+      "step": 3590
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.560693641618497e-06,
+      "loss": 5.6674,
+      "step": 3591
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.552023121387284e-06,
+      "loss": 5.3742,
+      "step": 3592
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.54335260115607e-06,
+      "loss": 5.395,
+      "step": 3593
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.534682080924856e-06,
+      "loss": 5.3908,
+      "step": 3594
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.526011560693641e-06,
+      "loss": 5.5114,
+      "step": 3595
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.517341040462429e-06,
+      "loss": 5.2531,
+      "step": 3596
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.508670520231214e-06,
+      "loss": 5.6168,
+      "step": 3597
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.5e-06,
+      "loss": 5.5318,
+      "step": 3598
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 7.491329479768786e-06,
+      "loss": 5.4796,
+      "step": 3599
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.482658959537572e-06,
+      "loss": 5.4562,
+      "step": 3600
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.473988439306358e-06,
+      "loss": 5.2017,
+      "step": 3601
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.465317919075144e-06,
+      "loss": 5.2842,
+      "step": 3602
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.45664739884393e-06,
+      "loss": 5.4777,
+      "step": 3603
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.447976878612717e-06,
+      "loss": 5.5017,
+      "step": 3604
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.4393063583815024e-06,
+      "loss": 5.3677,
+      "step": 3605
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.430635838150289e-06,
+      "loss": 5.6138,
+      "step": 3606
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.421965317919075e-06,
+      "loss": 5.3491,
+      "step": 3607
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 7.413294797687861e-06,
+      "loss": 5.3186,
+      "step": 3608
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.404624277456647e-06,
+      "loss": 5.3083,
+      "step": 3609
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.3959537572254334e-06,
+      "loss": 5.485,
+      "step": 3610
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.3872832369942195e-06,
+      "loss": 5.3027,
+      "step": 3611
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.378612716763006e-06,
+      "loss": 5.2588,
+      "step": 3612
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.3699421965317915e-06,
+      "loss": 5.1553,
+      "step": 3613
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.3612716763005784e-06,
+      "loss": 5.3258,
+      "step": 3614
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.352601156069364e-06,
+      "loss": 5.0672,
+      "step": 3615
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.3439306358381505e-06,
+      "loss": 5.1219,
+      "step": 3616
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 7.3352601156069365e-06,
+      "loss": 5.2877,
+      "step": 3617
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.3265895953757226e-06,
+      "loss": 4.386,
+      "step": 3618
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.317919075144509e-06,
+      "loss": 5.7595,
+      "step": 3619
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.3092485549132955e-06,
+      "loss": 5.6148,
+      "step": 3620
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.300578034682081e-06,
+      "loss": 5.6086,
+      "step": 3621
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.2919075144508675e-06,
+      "loss": 5.5809,
+      "step": 3622
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.283236994219653e-06,
+      "loss": 5.5729,
+      "step": 3623
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.27456647398844e-06,
+      "loss": 5.7469,
+      "step": 3624
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 7.265895953757226e-06,
+      "loss": 5.7577,
+      "step": 3625
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.257225433526012e-06,
+      "loss": 5.6336,
+      "step": 3626
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.248554913294798e-06,
+      "loss": 5.5427,
+      "step": 3627
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.239884393063584e-06,
+      "loss": 5.5956,
+      "step": 3628
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.23121387283237e-06,
+      "loss": 5.5067,
+      "step": 3629
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.222543352601157e-06,
+      "loss": 5.5008,
+      "step": 3630
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.213872832369942e-06,
+      "loss": 5.4654,
+      "step": 3631
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.205202312138729e-06,
+      "loss": 5.6511,
+      "step": 3632
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.196531791907515e-06,
+      "loss": 5.5786,
+      "step": 3633
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 7.187861271676301e-06,
+      "loss": 5.5961,
+      "step": 3634
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.179190751445087e-06,
+      "loss": 5.4493,
+      "step": 3635
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.170520231213873e-06,
+      "loss": 5.2953,
+      "step": 3636
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.161849710982659e-06,
+      "loss": 5.2933,
+      "step": 3637
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.153179190751446e-06,
+      "loss": 5.4574,
+      "step": 3638
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.144508670520231e-06,
+      "loss": 5.545,
+      "step": 3639
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.135838150289018e-06,
+      "loss": 5.4104,
+      "step": 3640
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.127167630057803e-06,
+      "loss": 5.2942,
+      "step": 3641
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.11849710982659e-06,
+      "loss": 5.4481,
+      "step": 3642
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 7.109826589595376e-06,
+      "loss": 5.5304,
+      "step": 3643
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.101156069364162e-06,
+      "loss": 5.489,
+      "step": 3644
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.092485549132948e-06,
+      "loss": 5.3456,
+      "step": 3645
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.083815028901735e-06,
+      "loss": 5.4695,
+      "step": 3646
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.07514450867052e-06,
+      "loss": 5.5354,
+      "step": 3647
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.066473988439307e-06,
+      "loss": 5.3256,
+      "step": 3648
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.057803468208092e-06,
+      "loss": 5.3375,
+      "step": 3649
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.049132947976879e-06,
+      "loss": 5.4668,
+      "step": 3650
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.040462427745665e-06,
+      "loss": 5.4604,
+      "step": 3651
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 7.031791907514451e-06,
+      "loss": 5.3502,
+      "step": 3652
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 7.023121387283237e-06,
+      "loss": 5.274,
+      "step": 3653
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 7.014450867052023e-06,
+      "loss": 5.3775,
+      "step": 3654
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 7.005780346820809e-06,
+      "loss": 5.4104,
+      "step": 3655
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.997109826589596e-06,
+      "loss": 5.4348,
+      "step": 3656
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.988439306358381e-06,
+      "loss": 5.4416,
+      "step": 3657
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.979768786127168e-06,
+      "loss": 5.5153,
+      "step": 3658
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.971098265895954e-06,
+      "loss": 5.2215,
+      "step": 3659
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.96242774566474e-06,
+      "loss": 5.2146,
+      "step": 3660
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 6.953757225433526e-06,
+      "loss": 5.1412,
+      "step": 3661
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.945086705202312e-06,
+      "loss": 5.2433,
+      "step": 3662
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.936416184971098e-06,
+      "loss": 5.2952,
+      "step": 3663
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.927745664739885e-06,
+      "loss": 4.9746,
+      "step": 3664
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.91907514450867e-06,
+      "loss": 4.9116,
+      "step": 3665
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.910404624277457e-06,
+      "loss": 4.9154,
+      "step": 3666
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.9017341040462425e-06,
+      "loss": 5.0213,
+      "step": 3667
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.893063583815029e-06,
+      "loss": 4.7458,
+      "step": 3668
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.884393063583815e-06,
+      "loss": 5.7256,
+      "step": 3669
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.875722543352601e-06,
+      "loss": 5.7897,
+      "step": 3670
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.8670520231213874e-06,
+      "loss": 5.7648,
+      "step": 3671
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.858381502890174e-06,
+      "loss": 5.6714,
+      "step": 3672
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.8497109826589595e-06,
+      "loss": 5.5076,
+      "step": 3673
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.841040462427746e-06,
+      "loss": 5.4878,
+      "step": 3674
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.832369942196532e-06,
+      "loss": 5.7035,
+      "step": 3675
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.8236994219653185e-06,
+      "loss": 5.5621,
+      "step": 3676
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.8150289017341045e-06,
+      "loss": 5.6086,
+      "step": 3677
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.8063583815028905e-06,
+      "loss": 5.3863,
+      "step": 3678
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.7976878612716766e-06,
+      "loss": 5.5429,
+      "step": 3679
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.789017341040463e-06,
+      "loss": 5.3949,
+      "step": 3680
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.780346820809249e-06,
+      "loss": 5.4903,
+      "step": 3681
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.7716763005780355e-06,
+      "loss": 5.4246,
+      "step": 3682
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.763005780346821e-06,
+      "loss": 5.5207,
+      "step": 3683
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.754335260115608e-06,
+      "loss": 5.4503,
+      "step": 3684
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.745664739884394e-06,
+      "loss": 5.4425,
+      "step": 3685
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.73699421965318e-06,
+      "loss": 5.5081,
+      "step": 3686
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.728323699421966e-06,
+      "loss": 5.4848,
+      "step": 3687
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 6.719653179190752e-06,
+      "loss": 5.3988,
+      "step": 3688
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.710982658959538e-06,
+      "loss": 5.2964,
+      "step": 3689
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.702312138728325e-06,
+      "loss": 5.4345,
+      "step": 3690
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.69364161849711e-06,
+      "loss": 5.5969,
+      "step": 3691
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.684971098265897e-06,
+      "loss": 5.2485,
+      "step": 3692
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.676300578034682e-06,
+      "loss": 5.3298,
+      "step": 3693
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.667630057803469e-06,
+      "loss": 5.3918,
+      "step": 3694
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.658959537572255e-06,
+      "loss": 5.3045,
+      "step": 3695
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.65028901734104e-06,
+      "loss": 5.2889,
+      "step": 3696
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.641618497109827e-06,
+      "loss": 5.4483,
+      "step": 3697
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.632947976878613e-06,
+      "loss": 5.7342,
+      "step": 3698
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.624277456647399e-06,
+      "loss": 5.5169,
+      "step": 3699
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.615606936416185e-06,
+      "loss": 5.39,
+      "step": 3700
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.606936416184971e-06,
+      "loss": 5.4244,
+      "step": 3701
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.598265895953757e-06,
+      "loss": 5.3642,
+      "step": 3702
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.589595375722544e-06,
+      "loss": 5.4081,
+      "step": 3703
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.580924855491329e-06,
+      "loss": 5.3685,
+      "step": 3704
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.572254335260116e-06,
+      "loss": 5.4131,
+      "step": 3705
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.563583815028901e-06,
+      "loss": 5.4601,
+      "step": 3706
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.554913294797688e-06,
+      "loss": 5.5335,
+      "step": 3707
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.546242774566474e-06,
+      "loss": 5.3859,
+      "step": 3708
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.53757225433526e-06,
+      "loss": 5.1838,
+      "step": 3709
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.528901734104046e-06,
+      "loss": 5.4508,
+      "step": 3710
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.520231213872832e-06,
+      "loss": 5.2734,
+      "step": 3711
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.511560693641618e-06,
+      "loss": 5.4388,
+      "step": 3712
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.502890173410405e-06,
+      "loss": 5.3669,
+      "step": 3713
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.49421965317919e-06,
+      "loss": 5.1228,
+      "step": 3714
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 6.485549132947977e-06,
+      "loss": 5.0465,
+      "step": 3715
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.476878612716763e-06,
+      "loss": 4.7836,
+      "step": 3716
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.468208092485549e-06,
+      "loss": 4.7763,
+      "step": 3717
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.459537572254335e-06,
+      "loss": 4.628,
+      "step": 3718
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.450867052023121e-06,
+      "loss": 5.669,
+      "step": 3719
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.442196531791907e-06,
+      "loss": 5.7938,
+      "step": 3720
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.433526011560694e-06,
+      "loss": 5.688,
+      "step": 3721
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.424855491329479e-06,
+      "loss": 5.8128,
+      "step": 3722
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.416184971098266e-06,
+      "loss": 5.6888,
+      "step": 3723
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.4075144508670515e-06,
+      "loss": 5.5521,
+      "step": 3724
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.398843930635838e-06,
+      "loss": 5.6251,
+      "step": 3725
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.390173410404624e-06,
+      "loss": 5.6391,
+      "step": 3726
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.3815028901734104e-06,
+      "loss": 5.4291,
+      "step": 3727
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.3728323699421965e-06,
+      "loss": 5.641,
+      "step": 3728
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.364161849710983e-06,
+      "loss": 5.5213,
+      "step": 3729
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.3554913294797685e-06,
+      "loss": 5.4575,
+      "step": 3730
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.346820809248555e-06,
+      "loss": 5.4698,
+      "step": 3731
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.338150289017341e-06,
+      "loss": 5.5217,
+      "step": 3732
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.3294797687861275e-06,
+      "loss": 5.4711,
+      "step": 3733
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.3208092485549135e-06,
+      "loss": 5.4258,
+      "step": 3734
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.3121387283236995e-06,
+      "loss": 5.3761,
+      "step": 3735
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.303468208092486e-06,
+      "loss": 5.4354,
+      "step": 3736
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.294797687861272e-06,
+      "loss": 5.5238,
+      "step": 3737
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.286127167630058e-06,
+      "loss": 5.3126,
+      "step": 3738
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.2774566473988445e-06,
+      "loss": 5.5175,
+      "step": 3739
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.26878612716763e-06,
+      "loss": 5.3509,
+      "step": 3740
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.260115606936417e-06,
+      "loss": 5.4884,
+      "step": 3741
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.251445086705203e-06,
+      "loss": 5.4431,
+      "step": 3742
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.242774566473989e-06,
+      "loss": 5.6268,
+      "step": 3743
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.234104046242775e-06,
+      "loss": 5.6274,
+      "step": 3744
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.225433526011561e-06,
+      "loss": 5.3049,
+      "step": 3745
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.216763005780347e-06,
+      "loss": 5.2942,
+      "step": 3746
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.208092485549134e-06,
+      "loss": 5.4219,
+      "step": 3747
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.199421965317919e-06,
+      "loss": 5.5211,
+      "step": 3748
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.190751445086706e-06,
+      "loss": 5.2444,
+      "step": 3749
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.182080924855491e-06,
+      "loss": 5.4703,
+      "step": 3750
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.173410404624278e-06,
+      "loss": 5.4221,
+      "step": 3751
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.164739884393064e-06,
+      "loss": 5.5308,
+      "step": 3752
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.15606936416185e-06,
+      "loss": 5.3833,
+      "step": 3753
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.147398843930636e-06,
+      "loss": 5.4048,
+      "step": 3754
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.138728323699423e-06,
+      "loss": 5.395,
+      "step": 3755
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.130057803468208e-06,
+      "loss": 5.4663,
+      "step": 3756
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.121387283236995e-06,
+      "loss": 5.407,
+      "step": 3757
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.11271676300578e-06,
+      "loss": 5.4322,
+      "step": 3758
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.104046242774567e-06,
+      "loss": 5.1961,
+      "step": 3759
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.095375722543353e-06,
+      "loss": 5.4284,
+      "step": 3760
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.086705202312139e-06,
+      "loss": 5.2878,
+      "step": 3761
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.078034682080925e-06,
+      "loss": 5.3931,
+      "step": 3762
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.069364161849711e-06,
+      "loss": 5.1104,
+      "step": 3763
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.060693641618497e-06,
+      "loss": 5.2472,
+      "step": 3764
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.052023121387284e-06,
+      "loss": 5.0218,
+      "step": 3765
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.043352601156069e-06,
+      "loss": 4.965,
+      "step": 3766
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.034682080924856e-06,
+      "loss": 5.1257,
+      "step": 3767
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.026011560693642e-06,
+      "loss": 4.5959,
+      "step": 3768
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.017341040462428e-06,
+      "loss": 5.8679,
+      "step": 3769
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.008670520231214e-06,
+      "loss": 5.6611,
+      "step": 3770
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6e-06,
+      "loss": 5.7034,
+      "step": 3771
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.991329479768786e-06,
+      "loss": 5.6439,
+      "step": 3772
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.982658959537573e-06,
+      "loss": 5.6403,
+      "step": 3773
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.973988439306358e-06,
+      "loss": 5.4484,
+      "step": 3774
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.965317919075145e-06,
+      "loss": 5.7061,
+      "step": 3775
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.95664739884393e-06,
+      "loss": 5.7145,
+      "step": 3776
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.947976878612717e-06,
+      "loss": 5.4328,
+      "step": 3777
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.939306358381503e-06,
+      "loss": 5.705,
+      "step": 3778
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.930635838150289e-06,
+      "loss": 5.4016,
+      "step": 3779
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.921965317919075e-06,
+      "loss": 5.5058,
+      "step": 3780
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.913294797687862e-06,
+      "loss": 5.6396,
+      "step": 3781
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.904624277456647e-06,
+      "loss": 5.3798,
+      "step": 3782
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.895953757225434e-06,
+      "loss": 5.3374,
+      "step": 3783
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.8872832369942194e-06,
+      "loss": 5.3593,
+      "step": 3784
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.878612716763006e-06,
+      "loss": 5.5086,
+      "step": 3785
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.869942196531792e-06,
+      "loss": 5.4655,
+      "step": 3786
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.861271676300578e-06,
+      "loss": 5.5905,
+      "step": 3787
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.852601156069364e-06,
+      "loss": 5.575,
+      "step": 3788
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.8439306358381505e-06,
+      "loss": 5.6033,
+      "step": 3789
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.8352601156069365e-06,
+      "loss": 5.2759,
+      "step": 3790
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.8265895953757225e-06,
+      "loss": 5.595,
+      "step": 3791
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.8179190751445086e-06,
+      "loss": 5.5235,
+      "step": 3792
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.809248554913295e-06,
+      "loss": 5.3038,
+      "step": 3793
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.8005780346820815e-06,
+      "loss": 5.2872,
+      "step": 3794
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.791907514450867e-06,
+      "loss": 5.3684,
+      "step": 3795
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.7832369942196535e-06,
+      "loss": 5.5654,
+      "step": 3796
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.774566473988439e-06,
+      "loss": 5.4322,
+      "step": 3797
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.765895953757226e-06,
+      "loss": 5.3413,
+      "step": 3798
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.757225433526012e-06,
+      "loss": 5.4163,
+      "step": 3799
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.748554913294798e-06,
+      "loss": 5.3127,
+      "step": 3800
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.739884393063584e-06,
+      "loss": 5.3653,
+      "step": 3801
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.73121387283237e-06,
+      "loss": 5.5681,
+      "step": 3802
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.722543352601156e-06,
+      "loss": 5.4465,
+      "step": 3803
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.713872832369943e-06,
+      "loss": 5.4645,
+      "step": 3804
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.705202312138728e-06,
+      "loss": 5.5164,
+      "step": 3805
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.696531791907515e-06,
+      "loss": 5.2849,
+      "step": 3806
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.687861271676301e-06,
+      "loss": 5.2435,
+      "step": 3807
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.679190751445087e-06,
+      "loss": 5.4607,
+      "step": 3808
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.670520231213873e-06,
+      "loss": 5.2293,
+      "step": 3809
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.661849710982659e-06,
+      "loss": 5.5732,
+      "step": 3810
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.653179190751445e-06,
+      "loss": 5.1412,
+      "step": 3811
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.644508670520232e-06,
+      "loss": 5.3773,
+      "step": 3812
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.635838150289017e-06,
+      "loss": 5.3716,
+      "step": 3813
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.627167630057804e-06,
+      "loss": 5.0547,
+      "step": 3814
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.618497109826589e-06,
+      "loss": 5.0307,
+      "step": 3815
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.609826589595376e-06,
+      "loss": 5.2498,
+      "step": 3816
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.601156069364162e-06,
+      "loss": 4.8935,
+      "step": 3817
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.592485549132948e-06,
+      "loss": 4.5812,
+      "step": 3818
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.583815028901734e-06,
+      "loss": 5.7658,
+      "step": 3819
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.575144508670521e-06,
+      "loss": 5.638,
+      "step": 3820
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.566473988439306e-06,
+      "loss": 5.5145,
+      "step": 3821
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.557803468208093e-06,
+      "loss": 5.5379,
+      "step": 3822
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.549132947976878e-06,
+      "loss": 5.617,
+      "step": 3823
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.540462427745665e-06,
+      "loss": 5.4033,
+      "step": 3824
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.531791907514451e-06,
+      "loss": 5.41,
+      "step": 3825
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.523121387283237e-06,
+      "loss": 5.632,
+      "step": 3826
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.514450867052023e-06,
+      "loss": 5.3289,
+      "step": 3827
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.505780346820809e-06,
+      "loss": 5.413,
+      "step": 3828
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.497109826589595e-06,
+      "loss": 5.483,
+      "step": 3829
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.488439306358382e-06,
+      "loss": 5.5825,
+      "step": 3830
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.479768786127167e-06,
+      "loss": 5.618,
+      "step": 3831
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.471098265895954e-06,
+      "loss": 5.6455,
+      "step": 3832
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.46242774566474e-06,
+      "loss": 5.6336,
+      "step": 3833
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.453757225433526e-06,
+      "loss": 5.4939,
+      "step": 3834
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.445086705202312e-06,
+      "loss": 5.3601,
+      "step": 3835
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.436416184971098e-06,
+      "loss": 5.4885,
+      "step": 3836
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.427745664739884e-06,
+      "loss": 5.4374,
+      "step": 3837
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.419075144508671e-06,
+      "loss": 5.4324,
+      "step": 3838
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.410404624277456e-06,
+      "loss": 5.403,
+      "step": 3839
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.401734104046243e-06,
+      "loss": 5.3269,
+      "step": 3840
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.3930635838150285e-06,
+      "loss": 5.418,
+      "step": 3841
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.384393063583815e-06,
+      "loss": 5.4038,
+      "step": 3842
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.375722543352601e-06,
+      "loss": 5.3506,
+      "step": 3843
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.367052023121387e-06,
+      "loss": 5.5356,
+      "step": 3844
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.3583815028901734e-06,
+      "loss": 5.3205,
+      "step": 3845
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.34971098265896e-06,
+      "loss": 5.421,
+      "step": 3846
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.3410404624277455e-06,
+      "loss": 5.3469,
+      "step": 3847
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.332369942196532e-06,
+      "loss": 5.3372,
+      "step": 3848
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.3236994219653176e-06,
+      "loss": 5.4504,
+      "step": 3849
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.3150289017341045e-06,
+      "loss": 5.6193,
+      "step": 3850
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.3063583815028905e-06,
+      "loss": 5.3544,
+      "step": 3851
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.2976878612716765e-06,
+      "loss": 5.4228,
+      "step": 3852
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.2890173410404626e-06,
+      "loss": 5.4858,
+      "step": 3853
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.280346820809249e-06,
+      "loss": 5.3297,
+      "step": 3854
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.271676300578035e-06,
+      "loss": 5.2389,
+      "step": 3855
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.2630057803468215e-06,
+      "loss": 5.3074,
+      "step": 3856
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 5.254335260115607e-06,
+      "loss": 5.3212,
+      "step": 3857
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.2456647398843936e-06,
+      "loss": 5.2992,
+      "step": 3858
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.23699421965318e-06,
+      "loss": 5.4799,
+      "step": 3859
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.228323699421966e-06,
+      "loss": 5.2996,
+      "step": 3860
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.219653179190752e-06,
+      "loss": 5.1863,
+      "step": 3861
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.210982658959538e-06,
+      "loss": 5.1669,
+      "step": 3862
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.202312138728324e-06,
+      "loss": 5.0911,
+      "step": 3863
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.193641618497111e-06,
+      "loss": 5.0824,
+      "step": 3864
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.184971098265896e-06,
+      "loss": 4.9201,
+      "step": 3865
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.176300578034683e-06,
+      "loss": 5.4174,
+      "step": 3866
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.167630057803468e-06,
+      "loss": 4.8243,
+      "step": 3867
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.158959537572255e-06,
+      "loss": 4.482,
+      "step": 3868
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.150289017341041e-06,
+      "loss": 5.8076,
+      "step": 3869
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.141618497109827e-06,
+      "loss": 5.6973,
+      "step": 3870
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.132947976878613e-06,
+      "loss": 5.6669,
+      "step": 3871
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.1242774566474e-06,
+      "loss": 5.4994,
+      "step": 3872
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.115606936416185e-06,
+      "loss": 5.5899,
+      "step": 3873
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.106936416184972e-06,
+      "loss": 5.5274,
+      "step": 3874
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 5.098265895953757e-06,
+      "loss": 5.5697,
+      "step": 3875
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.089595375722544e-06,
+      "loss": 5.7847,
+      "step": 3876
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.08092485549133e-06,
+      "loss": 5.4643,
+      "step": 3877
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.072254335260116e-06,
+      "loss": 5.5365,
+      "step": 3878
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.063583815028902e-06,
+      "loss": 5.4133,
+      "step": 3879
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.054913294797688e-06,
+      "loss": 5.665,
+      "step": 3880
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.046242774566474e-06,
+      "loss": 5.464,
+      "step": 3881
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.037572254335261e-06,
+      "loss": 5.3159,
+      "step": 3882
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.028901734104046e-06,
+      "loss": 5.4059,
+      "step": 3883
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.020231213872833e-06,
+      "loss": 5.3504,
+      "step": 3884
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.011560693641619e-06,
+      "loss": 5.3121,
+      "step": 3885
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.002890173410405e-06,
+      "loss": 5.5284,
+      "step": 3886
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.994219653179191e-06,
+      "loss": 5.339,
+      "step": 3887
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.985549132947976e-06,
+      "loss": 5.4164,
+      "step": 3888
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.976878612716763e-06,
+      "loss": 5.4493,
+      "step": 3889
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.968208092485549e-06,
+      "loss": 5.2918,
+      "step": 3890
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.959537572254335e-06,
+      "loss": 5.5224,
+      "step": 3891
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.950867052023121e-06,
+      "loss": 5.3873,
+      "step": 3892
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.942196531791907e-06,
+      "loss": 5.4333,
+      "step": 3893
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.933526011560693e-06,
+      "loss": 5.5756,
+      "step": 3894
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.92485549132948e-06,
+      "loss": 5.3649,
+      "step": 3895
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.916184971098265e-06,
+      "loss": 5.2633,
+      "step": 3896
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.907514450867052e-06,
+      "loss": 5.4086,
+      "step": 3897
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.898843930635838e-06,
+      "loss": 5.4535,
+      "step": 3898
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.890173410404624e-06,
+      "loss": 5.3928,
+      "step": 3899
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.88150289017341e-06,
+      "loss": 5.2588,
+      "step": 3900
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.872832369942196e-06,
+      "loss": 5.4203,
+      "step": 3901
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.8641618497109825e-06,
+      "loss": 5.4053,
+      "step": 3902
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.855491329479769e-06,
+      "loss": 5.5645,
+      "step": 3903
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.8468208092485545e-06,
+      "loss": 5.3945,
+      "step": 3904
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.838150289017341e-06,
+      "loss": 5.4697,
+      "step": 3905
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.829479768786127e-06,
+      "loss": 5.4298,
+      "step": 3906
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.8208092485549135e-06,
+      "loss": 5.2034,
+      "step": 3907
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.8121387283236995e-06,
+      "loss": 5.4299,
+      "step": 3908
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.8034682080924855e-06,
+      "loss": 5.4117,
+      "step": 3909
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.7947976878612716e-06,
+      "loss": 5.432,
+      "step": 3910
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.7861271676300585e-06,
+      "loss": 5.2433,
+      "step": 3911
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.777456647398844e-06,
+      "loss": 5.216,
+      "step": 3912
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.7687861271676305e-06,
+      "loss": 4.8947,
+      "step": 3913
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.760115606936416e-06,
+      "loss": 5.2307,
+      "step": 3914
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.751445086705203e-06,
+      "loss": 5.201,
+      "step": 3915
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.742774566473989e-06,
+      "loss": 5.1331,
+      "step": 3916
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.734104046242775e-06,
+      "loss": 4.8133,
+      "step": 3917
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.725433526011561e-06,
+      "loss": 4.5742,
+      "step": 3918
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.716763005780347e-06,
+      "loss": 5.8037,
+      "step": 3919
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 4.708092485549133e-06,
+      "loss": 5.6222,
+      "step": 3920
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.69942196531792e-06,
+      "loss": 5.44,
+      "step": 3921
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.690751445086705e-06,
+      "loss": 5.6518,
+      "step": 3922
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.682080924855492e-06,
+      "loss": 5.6065,
+      "step": 3923
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.673410404624278e-06,
+      "loss": 5.4729,
+      "step": 3924
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.664739884393064e-06,
+      "loss": 5.5551,
+      "step": 3925
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.65606936416185e-06,
+      "loss": 5.6122,
+      "step": 3926
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.647398843930636e-06,
+      "loss": 5.3888,
+      "step": 3927
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.638728323699422e-06,
+      "loss": 5.7275,
+      "step": 3928
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 4.630057803468209e-06,
+      "loss": 5.3197,
+      "step": 3929
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.621387283236994e-06,
+      "loss": 5.3994,
+      "step": 3930
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.612716763005781e-06,
+      "loss": 5.5699,
+      "step": 3931
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.604046242774566e-06,
+      "loss": 5.4051,
+      "step": 3932
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.595375722543353e-06,
+      "loss": 5.4391,
+      "step": 3933
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.586705202312139e-06,
+      "loss": 5.4282,
+      "step": 3934
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.578034682080925e-06,
+      "loss": 5.3663,
+      "step": 3935
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.569364161849711e-06,
+      "loss": 5.4054,
+      "step": 3936
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.560693641618497e-06,
+      "loss": 5.1528,
+      "step": 3937
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 4.552023121387283e-06,
+      "loss": 5.6016,
+      "step": 3938
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.54335260115607e-06,
+      "loss": 5.4919,
+      "step": 3939
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.534682080924855e-06,
+      "loss": 5.4579,
+      "step": 3940
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.526011560693642e-06,
+      "loss": 5.5744,
+      "step": 3941
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.517341040462428e-06,
+      "loss": 5.4696,
+      "step": 3942
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.508670520231214e-06,
+      "loss": 5.5462,
+      "step": 3943
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.5e-06,
+      "loss": 5.4916,
+      "step": 3944
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.491329479768786e-06,
+      "loss": 5.4523,
+      "step": 3945
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.482658959537572e-06,
+      "loss": 5.5371,
+      "step": 3946
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 4.473988439306359e-06,
+      "loss": 5.2894,
+      "step": 3947
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.465317919075144e-06,
+      "loss": 5.2826,
+      "step": 3948
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.456647398843931e-06,
+      "loss": 5.4907,
+      "step": 3949
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.447976878612716e-06,
+      "loss": 5.2409,
+      "step": 3950
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.439306358381503e-06,
+      "loss": 5.395,
+      "step": 3951
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.430635838150289e-06,
+      "loss": 5.3305,
+      "step": 3952
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.421965317919075e-06,
+      "loss": 5.5753,
+      "step": 3953
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.413294797687861e-06,
+      "loss": 5.507,
+      "step": 3954
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.404624277456648e-06,
+      "loss": 5.5741,
+      "step": 3955
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 4.395953757225433e-06,
+      "loss": 5.4847,
+      "step": 3956
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.38728323699422e-06,
+      "loss": 5.2421,
+      "step": 3957
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.3786127167630054e-06,
+      "loss": 5.298,
+      "step": 3958
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.369942196531792e-06,
+      "loss": 5.2138,
+      "step": 3959
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.361271676300578e-06,
+      "loss": 5.3204,
+      "step": 3960
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.352601156069364e-06,
+      "loss": 5.1114,
+      "step": 3961
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.34393063583815e-06,
+      "loss": 5.2199,
+      "step": 3962
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.3352601156069365e-06,
+      "loss": 5.2396,
+      "step": 3963
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 4.3265895953757225e-06,
+      "loss": 4.9574,
+      "step": 3964
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.317919075144509e-06,
+      "loss": 5.1438,
+      "step": 3965
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.3092485549132946e-06,
+      "loss": 4.9272,
+      "step": 3966
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.3005780346820814e-06,
+      "loss": 4.9744,
+      "step": 3967
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.2919075144508675e-06,
+      "loss": 4.7184,
+      "step": 3968
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.2832369942196535e-06,
+      "loss": 5.6813,
+      "step": 3969
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.2745664739884395e-06,
+      "loss": 5.7092,
+      "step": 3970
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.2658959537572256e-06,
+      "loss": 5.6161,
+      "step": 3971
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.257225433526012e-06,
+      "loss": 5.3616,
+      "step": 3972
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 4.2485549132947985e-06,
+      "loss": 5.6361,
+      "step": 3973
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.239884393063584e-06,
+      "loss": 5.4913,
+      "step": 3974
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.2312138728323706e-06,
+      "loss": 5.4694,
+      "step": 3975
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.222543352601156e-06,
+      "loss": 5.5341,
+      "step": 3976
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.213872832369943e-06,
+      "loss": 5.5488,
+      "step": 3977
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.205202312138729e-06,
+      "loss": 5.4348,
+      "step": 3978
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.196531791907515e-06,
+      "loss": 5.2196,
+      "step": 3979
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.187861271676301e-06,
+      "loss": 5.464,
+      "step": 3980
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.179190751445088e-06,
+      "loss": 5.3703,
+      "step": 3981
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 4.170520231213873e-06,
+      "loss": 5.4051,
+      "step": 3982
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.161849710982659e-06,
+      "loss": 5.5208,
+      "step": 3983
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.153179190751445e-06,
+      "loss": 5.2761,
+      "step": 3984
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.144508670520231e-06,
+      "loss": 5.338,
+      "step": 3985
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.135838150289018e-06,
+      "loss": 5.3354,
+      "step": 3986
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.127167630057803e-06,
+      "loss": 5.4836,
+      "step": 3987
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.11849710982659e-06,
+      "loss": 5.5273,
+      "step": 3988
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.109826589595375e-06,
+      "loss": 5.4258,
+      "step": 3989
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.101156069364162e-06,
+      "loss": 5.5121,
+      "step": 3990
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 4.092485549132948e-06,
+      "loss": 5.3503,
+      "step": 3991
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.083815028901734e-06,
+      "loss": 5.4161,
+      "step": 3992
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.07514450867052e-06,
+      "loss": 5.3631,
+      "step": 3993
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.066473988439307e-06,
+      "loss": 5.3653,
+      "step": 3994
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.057803468208092e-06,
+      "loss": 5.4953,
+      "step": 3995
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.049132947976879e-06,
+      "loss": 5.4194,
+      "step": 3996
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.040462427745664e-06,
+      "loss": 5.2943,
+      "step": 3997
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.031791907514451e-06,
+      "loss": 5.3445,
+      "step": 3998
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.023121387283237e-06,
+      "loss": 5.1126,
+      "step": 3999
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 4.014450867052023e-06,
+      "loss": 5.3928,
+      "step": 4000
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 4.005780346820809e-06,
+      "loss": 5.3157,
+      "step": 4001
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.997109826589595e-06,
+      "loss": 5.2959,
+      "step": 4002
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.988439306358381e-06,
+      "loss": 5.5213,
+      "step": 4003
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.979768786127168e-06,
+      "loss": 5.127,
+      "step": 4004
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.971098265895953e-06,
+      "loss": 5.2902,
+      "step": 4005
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.96242774566474e-06,
+      "loss": 5.2821,
+      "step": 4006
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.953757225433526e-06,
+      "loss": 5.399,
+      "step": 4007
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.945086705202312e-06,
+      "loss": 5.4217,
+      "step": 4008
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 3.936416184971098e-06,
+      "loss": 5.3659,
+      "step": 4009
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.927745664739884e-06,
+      "loss": 5.1973,
+      "step": 4010
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.91907514450867e-06,
+      "loss": 5.3746,
+      "step": 4011
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.910404624277457e-06,
+      "loss": 5.3858,
+      "step": 4012
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.901734104046242e-06,
+      "loss": 5.0759,
+      "step": 4013
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.893063583815029e-06,
+      "loss": 4.8736,
+      "step": 4014
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.8843930635838145e-06,
+      "loss": 4.8653,
+      "step": 4015
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.875722543352601e-06,
+      "loss": 4.8575,
+      "step": 4016
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.867052023121387e-06,
+      "loss": 4.7863,
+      "step": 4017
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.858381502890173e-06,
+      "loss": 4.7399,
+      "step": 4018
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.8497109826589594e-06,
+      "loss": 5.5412,
+      "step": 4019
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.841040462427746e-06,
+      "loss": 5.6773,
+      "step": 4020
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.8323699421965315e-06,
+      "loss": 5.7314,
+      "step": 4021
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.823699421965318e-06,
+      "loss": 5.5207,
+      "step": 4022
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.8150289017341036e-06,
+      "loss": 5.4961,
+      "step": 4023
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.8063583815028905e-06,
+      "loss": 5.477,
+      "step": 4024
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.797687861271676e-06,
+      "loss": 5.5658,
+      "step": 4025
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.7890173410404625e-06,
+      "loss": 5.5225,
+      "step": 4026
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 3.7803468208092486e-06,
+      "loss": 5.4342,
+      "step": 4027
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.771676300578035e-06,
+      "loss": 5.4066,
+      "step": 4028
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.7630057803468206e-06,
+      "loss": 5.33,
+      "step": 4029
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.754335260115607e-06,
+      "loss": 5.2893,
+      "step": 4030
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.745664739884393e-06,
+      "loss": 5.6026,
+      "step": 4031
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.736994219653179e-06,
+      "loss": 5.6506,
+      "step": 4032
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.728323699421965e-06,
+      "loss": 5.4771,
+      "step": 4033
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.7196531791907512e-06,
+      "loss": 5.5133,
+      "step": 4034
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.7109826589595377e-06,
+      "loss": 5.2709,
+      "step": 4035
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 3.7023121387283237e-06,
+      "loss": 5.3767,
+      "step": 4036
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.6936416184971097e-06,
+      "loss": 5.467,
+      "step": 4037
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.6849710982658958e-06,
+      "loss": 5.4217,
+      "step": 4038
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.676300578034682e-06,
+      "loss": 5.5318,
+      "step": 4039
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.6676300578034683e-06,
+      "loss": 5.354,
+      "step": 4040
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.6589595375722543e-06,
+      "loss": 5.4003,
+      "step": 4041
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.6502890173410403e-06,
+      "loss": 5.5124,
+      "step": 4042
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.6416184971098264e-06,
+      "loss": 5.5485,
+      "step": 4043
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.632947976878613e-06,
+      "loss": 5.3448,
+      "step": 4044
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.624277456647399e-06,
+      "loss": 5.3561,
+      "step": 4045
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.615606936416185e-06,
+      "loss": 5.406,
+      "step": 4046
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.606936416184971e-06,
+      "loss": 5.3786,
+      "step": 4047
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.5982658959537574e-06,
+      "loss": 5.4319,
+      "step": 4048
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.5895953757225434e-06,
+      "loss": 5.5697,
+      "step": 4049
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.5809248554913295e-06,
+      "loss": 5.3695,
+      "step": 4050
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.5722543352601155e-06,
+      "loss": 5.3672,
+      "step": 4051
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.5635838150289015e-06,
+      "loss": 5.5382,
+      "step": 4052
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.554913294797688e-06,
+      "loss": 5.601,
+      "step": 4053
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 3.546242774566474e-06,
+      "loss": 5.3465,
+      "step": 4054
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.53757225433526e-06,
+      "loss": 5.5614,
+      "step": 4055
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.528901734104046e-06,
+      "loss": 5.4285,
+      "step": 4056
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.5202312138728325e-06,
+      "loss": 5.2436,
+      "step": 4057
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.5115606936416186e-06,
+      "loss": 5.4546,
+      "step": 4058
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.5028901734104046e-06,
+      "loss": 5.3877,
+      "step": 4059
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.4942196531791906e-06,
+      "loss": 5.2136,
+      "step": 4060
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.485549132947977e-06,
+      "loss": 5.3784,
+      "step": 4061
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.476878612716763e-06,
+      "loss": 5.3677,
+      "step": 4062
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.468208092485549e-06,
+      "loss": 5.3273,
+      "step": 4063
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.459537572254335e-06,
+      "loss": 5.1243,
+      "step": 4064
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.4508670520231212e-06,
+      "loss": 5.115,
+      "step": 4065
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.4421965317919077e-06,
+      "loss": 5.0084,
+      "step": 4066
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.4335260115606937e-06,
+      "loss": 5.0635,
+      "step": 4067
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.4248554913294798e-06,
+      "loss": 4.5676,
+      "step": 4068
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.416184971098266e-06,
+      "loss": 5.6834,
+      "step": 4069
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.4075144508670522e-06,
+      "loss": 5.6016,
+      "step": 4070
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.3988439306358383e-06,
+      "loss": 5.6715,
+      "step": 4071
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.3901734104046243e-06,
+      "loss": 5.5076,
+      "step": 4072
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.3815028901734103e-06,
+      "loss": 5.6202,
+      "step": 4073
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.372832369942197e-06,
+      "loss": 5.4785,
+      "step": 4074
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.364161849710983e-06,
+      "loss": 5.5953,
+      "step": 4075
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.355491329479769e-06,
+      "loss": 5.4966,
+      "step": 4076
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.346820809248555e-06,
+      "loss": 5.6625,
+      "step": 4077
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.338150289017341e-06,
+      "loss": 5.4155,
+      "step": 4078
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.3294797687861274e-06,
+      "loss": 5.462,
+      "step": 4079
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.3208092485549134e-06,
+      "loss": 5.3905,
+      "step": 4080
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.3121387283236995e-06,
+      "loss": 5.5187,
+      "step": 4081
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.3034682080924855e-06,
+      "loss": 5.3018,
+      "step": 4082
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.294797687861272e-06,
+      "loss": 5.3422,
+      "step": 4083
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.286127167630058e-06,
+      "loss": 5.3577,
+      "step": 4084
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.277456647398844e-06,
+      "loss": 5.4846,
+      "step": 4085
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.26878612716763e-06,
+      "loss": 5.3726,
+      "step": 4086
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.260115606936416e-06,
+      "loss": 5.3353,
+      "step": 4087
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.2514450867052026e-06,
+      "loss": 5.2292,
+      "step": 4088
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.2427745664739886e-06,
+      "loss": 5.5624,
+      "step": 4089
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.2341040462427746e-06,
+      "loss": 5.2985,
+      "step": 4090
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.2254335260115607e-06,
+      "loss": 5.4439,
+      "step": 4091
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.216763005780347e-06,
+      "loss": 5.5023,
+      "step": 4092
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.208092485549133e-06,
+      "loss": 5.2132,
+      "step": 4093
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.199421965317919e-06,
+      "loss": 5.2973,
+      "step": 4094
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.1907514450867052e-06,
+      "loss": 5.3905,
+      "step": 4095
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.1820809248554917e-06,
+      "loss": 5.191,
+      "step": 4096
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.1734104046242777e-06,
+      "loss": 5.4613,
+      "step": 4097
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.1647398843930637e-06,
+      "loss": 5.4033,
+      "step": 4098
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.1560693641618498e-06,
+      "loss": 5.4116,
+      "step": 4099
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.147398843930636e-06,
+      "loss": 5.1493,
+      "step": 4100
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.1387283236994223e-06,
+      "loss": 5.1919,
+      "step": 4101
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.1300578034682083e-06,
+      "loss": 5.3376,
+      "step": 4102
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.1213872832369943e-06,
+      "loss": 5.4106,
+      "step": 4103
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.1127167630057804e-06,
+      "loss": 5.26,
+      "step": 4104
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.104046242774567e-06,
+      "loss": 5.1842,
+      "step": 4105
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.095375722543353e-06,
+      "loss": 5.2596,
+      "step": 4106
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.086705202312139e-06,
+      "loss": 5.4097,
+      "step": 4107
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.078034682080925e-06,
+      "loss": 5.4114,
+      "step": 4108
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.0693641618497114e-06,
+      "loss": 5.4094,
+      "step": 4109
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.0606936416184974e-06,
+      "loss": 5.2388,
+      "step": 4110
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.0520231213872834e-06,
+      "loss": 5.1615,
+      "step": 4111
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.0433526011560695e-06,
+      "loss": 5.2112,
+      "step": 4112
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.0346820809248555e-06,
+      "loss": 4.949,
+      "step": 4113
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.026011560693642e-06,
+      "loss": 5.0824,
+      "step": 4114
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.017341040462428e-06,
+      "loss": 5.095,
+      "step": 4115
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.008670520231214e-06,
+      "loss": 5.1041,
+      "step": 4116
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3e-06,
+      "loss": 5.0525,
+      "step": 4117
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9913294797687865e-06,
+      "loss": 4.7439,
+      "step": 4118
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9826589595375726e-06,
+      "loss": 5.5089,
+      "step": 4119
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9739884393063586e-06,
+      "loss": 5.5699,
+      "step": 4120
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9653179190751446e-06,
+      "loss": 5.5837,
+      "step": 4121
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.956647398843931e-06,
+      "loss": 5.5633,
+      "step": 4122
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.947976878612717e-06,
+      "loss": 5.4232,
+      "step": 4123
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.939306358381503e-06,
+      "loss": 5.4054,
+      "step": 4124
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.930635838150289e-06,
+      "loss": 5.5055,
+      "step": 4125
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9219653179190752e-06,
+      "loss": 5.3313,
+      "step": 4126
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9132947976878613e-06,
+      "loss": 5.6956,
+      "step": 4127
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.9046242774566473e-06,
+      "loss": 5.3884,
+      "step": 4128
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8959537572254333e-06,
+      "loss": 5.6085,
+      "step": 4129
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8872832369942194e-06,
+      "loss": 5.6036,
+      "step": 4130
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.878612716763006e-06,
+      "loss": 5.5673,
+      "step": 4131
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.869942196531792e-06,
+      "loss": 5.4521,
+      "step": 4132
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.861271676300578e-06,
+      "loss": 5.3907,
+      "step": 4133
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.852601156069364e-06,
+      "loss": 5.6804,
+      "step": 4134
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.8439306358381504e-06,
+      "loss": 5.4903,
+      "step": 4135
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.8352601156069364e-06,
+      "loss": 5.4281,
+      "step": 4136
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.8265895953757224e-06,
+      "loss": 5.3613,
+      "step": 4137
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.8179190751445085e-06,
+      "loss": 5.4742,
+      "step": 4138
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.8092485549132945e-06,
+      "loss": 5.4028,
+      "step": 4139
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.800578034682081e-06,
+      "loss": 5.2981,
+      "step": 4140
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.791907514450867e-06,
+      "loss": 5.4054,
+      "step": 4141
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.783236994219653e-06,
+      "loss": 5.2857,
+      "step": 4142
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.774566473988439e-06,
+      "loss": 5.4931,
+      "step": 4143
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.7658959537572255e-06,
+      "loss": 5.3351,
+      "step": 4144
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.7572254335260116e-06,
+      "loss": 5.5304,
+      "step": 4145
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.7485549132947976e-06,
+      "loss": 5.4338,
+      "step": 4146
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.7398843930635836e-06,
+      "loss": 5.237,
+      "step": 4147
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.73121387283237e-06,
+      "loss": 5.3772,
+      "step": 4148
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.722543352601156e-06,
+      "loss": 5.3434,
+      "step": 4149
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.713872832369942e-06,
+      "loss": 5.3247,
+      "step": 4150
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.705202312138728e-06,
+      "loss": 5.2472,
+      "step": 4151
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6965317919075142e-06,
+      "loss": 5.181,
+      "step": 4152
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6878612716763007e-06,
+      "loss": 5.346,
+      "step": 4153
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6791907514450867e-06,
+      "loss": 5.3281,
+      "step": 4154
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6705202312138728e-06,
+      "loss": 5.1518,
+      "step": 4155
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6618497109826588e-06,
+      "loss": 5.2301,
+      "step": 4156
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6531791907514452e-06,
+      "loss": 5.1791,
+      "step": 4157
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6445086705202313e-06,
+      "loss": 5.2235,
+      "step": 4158
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6358381502890173e-06,
+      "loss": 5.193,
+      "step": 4159
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.6271676300578033e-06,
+      "loss": 5.2104,
+      "step": 4160
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.61849710982659e-06,
+      "loss": 5.3723,
+      "step": 4161
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.609826589595376e-06,
+      "loss": 5.2503,
+      "step": 4162
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.601156069364162e-06,
+      "loss": 5.439,
+      "step": 4163
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.592485549132948e-06,
+      "loss": 5.2728,
+      "step": 4164
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.583815028901734e-06,
+      "loss": 5.3316,
+      "step": 4165
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.5751445086705204e-06,
+      "loss": 4.8977,
+      "step": 4166
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.5664739884393064e-06,
+      "loss": 4.784,
+      "step": 4167
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.5578034682080925e-06,
+      "loss": 4.9362,
+      "step": 4168
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.5491329479768785e-06,
+      "loss": 5.5528,
+      "step": 4169
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.540462427745665e-06,
+      "loss": 5.5437,
+      "step": 4170
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.531791907514451e-06,
+      "loss": 5.6793,
+      "step": 4171
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.523121387283237e-06,
+      "loss": 5.5009,
+      "step": 4172
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.514450867052023e-06,
+      "loss": 5.6481,
+      "step": 4173
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.5057803468208095e-06,
+      "loss": 5.5286,
+      "step": 4174
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.4971098265895955e-06,
+      "loss": 5.5441,
+      "step": 4175
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.4884393063583816e-06,
+      "loss": 5.587,
+      "step": 4176
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.4797687861271676e-06,
+      "loss": 5.4872,
+      "step": 4177
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.4710982658959537e-06,
+      "loss": 5.4151,
+      "step": 4178
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.46242774566474e-06,
+      "loss": 5.4446,
+      "step": 4179
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.453757225433526e-06,
+      "loss": 5.503,
+      "step": 4180
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.445086705202312e-06,
+      "loss": 5.4094,
+      "step": 4181
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.436416184971098e-06,
+      "loss": 5.439,
+      "step": 4182
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.4277456647398847e-06,
+      "loss": 5.4037,
+      "step": 4183
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.4190751445086707e-06,
+      "loss": 5.4787,
+      "step": 4184
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.4104046242774567e-06,
+      "loss": 5.3329,
+      "step": 4185
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.4017341040462428e-06,
+      "loss": 5.3859,
+      "step": 4186
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.3930635838150292e-06,
+      "loss": 5.4593,
+      "step": 4187
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.3843930635838153e-06,
+      "loss": 5.724,
+      "step": 4188
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.3757225433526013e-06,
+      "loss": 5.3159,
+      "step": 4189
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.3670520231213873e-06,
+      "loss": 5.465,
+      "step": 4190
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.3583815028901734e-06,
+      "loss": 5.3628,
+      "step": 4191
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.34971098265896e-06,
+      "loss": 5.3739,
+      "step": 4192
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.341040462427746e-06,
+      "loss": 5.5357,
+      "step": 4193
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.332369942196532e-06,
+      "loss": 5.3536,
+      "step": 4194
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.323699421965318e-06,
+      "loss": 5.6237,
+      "step": 4195
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.3150289017341044e-06,
+      "loss": 5.3012,
+      "step": 4196
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.3063583815028904e-06,
+      "loss": 5.4361,
+      "step": 4197
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.2976878612716764e-06,
+      "loss": 5.5445,
+      "step": 4198
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.2890173410404625e-06,
+      "loss": 5.4449,
+      "step": 4199
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.2803468208092485e-06,
+      "loss": 5.5569,
+      "step": 4200
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.271676300578035e-06,
+      "loss": 5.1406,
+      "step": 4201
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.263005780346821e-06,
+      "loss": 5.3514,
+      "step": 4202
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.254335260115607e-06,
+      "loss": 5.3237,
+      "step": 4203
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.245664739884393e-06,
+      "loss": 5.3567,
+      "step": 4204
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.2369942196531795e-06,
+      "loss": 5.2916,
+      "step": 4205
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.2283236994219656e-06,
+      "loss": 5.3789,
+      "step": 4206
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.2196531791907516e-06,
+      "loss": 5.3235,
+      "step": 4207
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.2109826589595376e-06,
+      "loss": 5.1856,
+      "step": 4208
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.202312138728324e-06,
+      "loss": 5.4195,
+      "step": 4209
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.19364161849711e-06,
+      "loss": 5.4399,
+      "step": 4210
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.184971098265896e-06,
+      "loss": 5.1733,
+      "step": 4211
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.176300578034682e-06,
+      "loss": 5.232,
+      "step": 4212
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.1676300578034682e-06,
+      "loss": 5.0585,
+      "step": 4213
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.1589595375722547e-06,
+      "loss": 5.2122,
+      "step": 4214
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1502890173410407e-06,
+      "loss": 5.0638,
+      "step": 4215
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1416184971098268e-06,
+      "loss": 5.0041,
+      "step": 4216
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1329479768786128e-06,
+      "loss": 5.0001,
+      "step": 4217
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1242774566473992e-06,
+      "loss": 4.7327,
+      "step": 4218
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1156069364161853e-06,
+      "loss": 5.7463,
+      "step": 4219
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.1069364161849713e-06,
+      "loss": 5.6994,
+      "step": 4220
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.0982658959537573e-06,
+      "loss": 5.6665,
+      "step": 4221
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.089595375722544e-06,
+      "loss": 5.7771,
+      "step": 4222
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.0809248554913294e-06,
+      "loss": 5.431,
+      "step": 4223
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0722543352601154e-06,
+      "loss": 5.3765,
+      "step": 4224
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0635838150289015e-06,
+      "loss": 5.614,
+      "step": 4225
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0549132947976875e-06,
+      "loss": 5.4828,
+      "step": 4226
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.046242774566474e-06,
+      "loss": 5.4925,
+      "step": 4227
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.03757225433526e-06,
+      "loss": 5.4956,
+      "step": 4228
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.028901734104046e-06,
+      "loss": 5.4918,
+      "step": 4229
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.020231213872832e-06,
+      "loss": 5.3792,
+      "step": 4230
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0115606936416185e-06,
+      "loss": 5.3477,
+      "step": 4231
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.0028901734104046e-06,
+      "loss": 5.2099,
+      "step": 4232
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.9942196531791906e-06,
+      "loss": 5.2821,
+      "step": 4233
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.9855491329479766e-06,
+      "loss": 5.4284,
+      "step": 4234
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.976878612716763e-06,
+      "loss": 5.4036,
+      "step": 4235
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.968208092485549e-06,
+      "loss": 5.3815,
+      "step": 4236
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.959537572254335e-06,
+      "loss": 5.4222,
+      "step": 4237
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.950867052023121e-06,
+      "loss": 5.4202,
+      "step": 4238
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.9421965317919072e-06,
+      "loss": 5.3342,
+      "step": 4239
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.9335260115606937e-06,
+      "loss": 5.4599,
+      "step": 4240
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.9248554913294797e-06,
+      "loss": 5.043,
+      "step": 4241
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.9161849710982658e-06,
+      "loss": 5.3202,
+      "step": 4242
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.9075144508670518e-06,
+      "loss": 5.4235,
+      "step": 4243
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.898843930635838e-06,
+      "loss": 5.4933,
+      "step": 4244
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.8901734104046243e-06,
+      "loss": 5.4596,
+      "step": 4245
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.8815028901734103e-06,
+      "loss": 5.3833,
+      "step": 4246
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.8728323699421966e-06,
+      "loss": 5.5049,
+      "step": 4247
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.8641618497109826e-06,
+      "loss": 5.293,
+      "step": 4248
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.8554913294797688e-06,
+      "loss": 5.2026,
+      "step": 4249
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.8468208092485549e-06,
+      "loss": 5.2576,
+      "step": 4250
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.838150289017341e-06,
+      "loss": 5.3964,
+      "step": 4251
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.8294797687861271e-06,
+      "loss": 5.3494,
+      "step": 4252
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.8208092485549132e-06,
+      "loss": 5.4439,
+      "step": 4253
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.8121387283236994e-06,
+      "loss": 5.1649,
+      "step": 4254
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.8034682080924855e-06,
+      "loss": 5.275,
+      "step": 4255
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.7947976878612717e-06,
+      "loss": 5.2984,
+      "step": 4256
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.7861271676300577e-06,
+      "loss": 5.1559,
+      "step": 4257
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.777456647398844e-06,
+      "loss": 5.2479,
+      "step": 4258
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.76878612716763e-06,
+      "loss": 5.3342,
+      "step": 4259
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.7601156069364163e-06,
+      "loss": 5.2025,
+      "step": 4260
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.7514450867052023e-06,
+      "loss": 5.4314,
+      "step": 4261
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.7427745664739885e-06,
+      "loss": 4.9916,
+      "step": 4262
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.7341040462427746e-06,
+      "loss": 5.0605,
+      "step": 4263
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.7254335260115606e-06,
+      "loss": 5.0654,
+      "step": 4264
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.7167630057803469e-06,
+      "loss": 4.9763,
+      "step": 4265
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.708092485549133e-06,
+      "loss": 5.0154,
+      "step": 4266
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.6994219653179191e-06,
+      "loss": 5.0163,
+      "step": 4267
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.6907514450867052e-06,
+      "loss": 4.47,
+      "step": 4268
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.6820809248554914e-06,
+      "loss": 5.5818,
+      "step": 4269
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.6734104046242775e-06,
+      "loss": 5.6156,
+      "step": 4270
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.6647398843930637e-06,
+      "loss": 5.5,
+      "step": 4271
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.6560693641618497e-06,
+      "loss": 5.3279,
+      "step": 4272
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.647398843930636e-06,
+      "loss": 5.5051,
+      "step": 4273
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.638728323699422e-06,
+      "loss": 5.5376,
+      "step": 4274
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.630057803468208e-06,
+      "loss": 5.5308,
+      "step": 4275
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.6213872832369943e-06,
+      "loss": 5.5887,
+      "step": 4276
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.6127167630057803e-06,
+      "loss": 5.3431,
+      "step": 4277
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.6040462427745666e-06,
+      "loss": 5.3809,
+      "step": 4278
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.5953757225433526e-06,
+      "loss": 5.3123,
+      "step": 4279
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.5867052023121389e-06,
+      "loss": 5.5032,
+      "step": 4280
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.5780346820809249e-06,
+      "loss": 5.3653,
+      "step": 4281
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.5693641618497111e-06,
+      "loss": 5.3123,
+      "step": 4282
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.5606936416184972e-06,
+      "loss": 5.559,
+      "step": 4283
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.5520231213872834e-06,
+      "loss": 5.412,
+      "step": 4284
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.5433526011560694e-06,
+      "loss": 5.5555,
+      "step": 4285
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.5346820809248557e-06,
+      "loss": 5.3101,
+      "step": 4286
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.5260115606936417e-06,
+      "loss": 5.4294,
+      "step": 4287
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.5173410404624278e-06,
+      "loss": 5.4257,
+      "step": 4288
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.508670520231214e-06,
+      "loss": 5.4821,
+      "step": 4289
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.5e-06,
+      "loss": 5.3454,
+      "step": 4290
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.4913294797687863e-06,
+      "loss": 5.2102,
+      "step": 4291
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.4826589595375723e-06,
+      "loss": 5.3367,
+      "step": 4292
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.4739884393063586e-06,
+      "loss": 5.3683,
+      "step": 4293
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.4653179190751446e-06,
+      "loss": 5.3063,
+      "step": 4294
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.4566473988439306e-06,
+      "loss": 5.2424,
+      "step": 4295
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.4479768786127167e-06,
+      "loss": 5.4882,
+      "step": 4296
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.439306358381503e-06,
+      "loss": 5.3725,
+      "step": 4297
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.430635838150289e-06,
+      "loss": 5.4681,
+      "step": 4298
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.4219653179190752e-06,
+      "loss": 5.3403,
+      "step": 4299
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.4132947976878612e-06,
+      "loss": 5.1856,
+      "step": 4300
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.4046242774566473e-06,
+      "loss": 5.2428,
+      "step": 4301
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.3959537572254335e-06,
+      "loss": 5.2518,
+      "step": 4302
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.3872832369942195e-06,
+      "loss": 5.4664,
+      "step": 4303
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.3786127167630058e-06,
+      "loss": 5.1299,
+      "step": 4304
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.3699421965317918e-06,
+      "loss": 5.3089,
+      "step": 4305
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.361271676300578e-06,
+      "loss": 5.4222,
+      "step": 4306
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.352601156069364e-06,
+      "loss": 5.1595,
+      "step": 4307
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.3439306358381503e-06,
+      "loss": 5.2342,
+      "step": 4308
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.3352601156069364e-06,
+      "loss": 5.2065,
+      "step": 4309
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.3265895953757226e-06,
+      "loss": 5.2441,
+      "step": 4310
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.3179190751445087e-06,
+      "loss": 5.1878,
+      "step": 4311
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.309248554913295e-06,
+      "loss": 5.162,
+      "step": 4312
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.300578034682081e-06,
+      "loss": 5.0681,
+      "step": 4313
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.291907514450867e-06,
+      "loss": 5.3732,
+      "step": 4314
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.2832369942196532e-06,
+      "loss": 4.9756,
+      "step": 4315
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.2745664739884392e-06,
+      "loss": 4.8992,
+      "step": 4316
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.2658959537572255e-06,
+      "loss": 4.8588,
+      "step": 4317
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.2572254335260115e-06,
+      "loss": 4.8248,
+      "step": 4318
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.2485549132947978e-06,
+      "loss": 5.6645,
+      "step": 4319
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.2398843930635838e-06,
+      "loss": 5.4146,
+      "step": 4320
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.23121387283237e-06,
+      "loss": 5.5123,
+      "step": 4321
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.222543352601156e-06,
+      "loss": 5.6543,
+      "step": 4322
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2138728323699423e-06,
+      "loss": 5.5127,
+      "step": 4323
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.2052023121387284e-06,
+      "loss": 5.6304,
+      "step": 4324
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.1965317919075146e-06,
+      "loss": 5.5336,
+      "step": 4325
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.1878612716763006e-06,
+      "loss": 5.4198,
+      "step": 4326
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.1878612716763006e-06,
+      "loss": 5.4849,
+      "step": 4327
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.1791907514450867e-06,
+      "loss": 5.3657,
+      "step": 4328
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.170520231213873e-06,
+      "loss": 5.5515,
+      "step": 4329
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.161849710982659e-06,
+      "loss": 5.4946,
+      "step": 4330
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.1531791907514452e-06,
+      "loss": 5.304,
+      "step": 4331
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.1445086705202312e-06,
+      "loss": 5.5548,
+      "step": 4332
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.1358381502890175e-06,
+      "loss": 5.574,
+      "step": 4333
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.1271676300578035e-06,
+      "loss": 5.3662,
+      "step": 4334
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.1184971098265898e-06,
+      "loss": 5.4107,
+      "step": 4335
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.1098265895953758e-06,
+      "loss": 5.2635,
+      "step": 4336
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.101156069364162e-06,
+      "loss": 5.4652,
+      "step": 4337
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.092485549132948e-06,
+      "loss": 5.4761,
+      "step": 4338
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0838150289017341e-06,
+      "loss": 5.2196,
+      "step": 4339
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0751445086705204e-06,
+      "loss": 5.5139,
+      "step": 4340
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0664739884393064e-06,
+      "loss": 5.3197,
+      "step": 4341
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0578034682080926e-06,
+      "loss": 5.3803,
+      "step": 4342
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0491329479768787e-06,
+      "loss": 5.4311,
+      "step": 4343
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0404624277456647e-06,
+      "loss": 5.217,
+      "step": 4344
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0317919075144507e-06,
+      "loss": 5.4947,
+      "step": 4345
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.023121387283237e-06,
+      "loss": 5.5503,
+      "step": 4346
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.014450867052023e-06,
+      "loss": 5.3921,
+      "step": 4347
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.0057803468208093e-06,
+      "loss": 5.3858,
+      "step": 4348
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 9.971098265895953e-07,
+      "loss": 5.5908,
+      "step": 4349
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 9.884393063583815e-07,
+      "loss": 5.3987,
+      "step": 4350
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 9.797687861271676e-07,
+      "loss": 5.3403,
+      "step": 4351
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 9.710982658959536e-07,
+      "loss": 5.377,
+      "step": 4352
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 9.624277456647399e-07,
+      "loss": 5.3385,
+      "step": 4353
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 9.537572254335259e-07,
+      "loss": 5.2926,
+      "step": 4354
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 9.450867052023121e-07,
+      "loss": 5.4416,
+      "step": 4355
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 9.364161849710983e-07,
+      "loss": 5.2407,
+      "step": 4356
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 9.277456647398844e-07,
+      "loss": 5.3296,
+      "step": 4357
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 9.190751445086705e-07,
+      "loss": 5.1662,
+      "step": 4358
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 9.104046242774566e-07,
+      "loss": 5.3271,
+      "step": 4359
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 9.017341040462427e-07,
+      "loss": 5.2996,
+      "step": 4360
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 8.930635838150289e-07,
+      "loss": 5.1281,
+      "step": 4361
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 8.84393063583815e-07,
+      "loss": 4.9652,
+      "step": 4362
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 8.757225433526012e-07,
+      "loss": 4.9897,
+      "step": 4363
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 8.670520231213873e-07,
+      "loss": 4.9321,
+      "step": 4364
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 8.583815028901734e-07,
+      "loss": 5.3021,
+      "step": 4365
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 8.497109826589596e-07,
+      "loss": 5.0022,
+      "step": 4366
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 8.410404624277457e-07,
+      "loss": 5.0641,
+      "step": 4367
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 8.323699421965318e-07,
+      "loss": 4.6637,
+      "step": 4368
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 8.23699421965318e-07,
+      "loss": 5.6292,
+      "step": 4369
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 8.15028901734104e-07,
+      "loss": 5.5159,
+      "step": 4370
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 8.063583815028902e-07,
+      "loss": 5.4665,
+      "step": 4371
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.976878612716763e-07,
+      "loss": 5.4702,
+      "step": 4372
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.890173410404624e-07,
+      "loss": 5.4785,
+      "step": 4373
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.803468208092486e-07,
+      "loss": 5.4958,
+      "step": 4374
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 7.716763005780347e-07,
+      "loss": 5.5724,
+      "step": 4375
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.630057803468209e-07,
+      "loss": 5.3995,
+      "step": 4376
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.54335260115607e-07,
+      "loss": 5.4278,
+      "step": 4377
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.456647398843931e-07,
+      "loss": 5.4355,
+      "step": 4378
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.369942196531793e-07,
+      "loss": 5.4488,
+      "step": 4379
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.283236994219653e-07,
+      "loss": 5.3558,
+      "step": 4380
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.196531791907515e-07,
+      "loss": 5.265,
+      "step": 4381
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.109826589595376e-07,
+      "loss": 5.4331,
+      "step": 4382
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.023121387283236e-07,
+      "loss": 5.4852,
+      "step": 4383
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 6.936416184971098e-07,
+      "loss": 5.6307,
+      "step": 4384
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.849710982658959e-07,
+      "loss": 5.4433,
+      "step": 4385
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.76300578034682e-07,
+      "loss": 5.386,
+      "step": 4386
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.676300578034682e-07,
+      "loss": 5.3767,
+      "step": 4387
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.589595375722543e-07,
+      "loss": 5.2551,
+      "step": 4388
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.502890173410405e-07,
+      "loss": 5.2543,
+      "step": 4389
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.416184971098266e-07,
+      "loss": 5.3695,
+      "step": 4390
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.329479768786127e-07,
+      "loss": 5.3616,
+      "step": 4391
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.242774566473989e-07,
+      "loss": 5.5352,
+      "step": 4392
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.15606936416185e-07,
+      "loss": 5.3109,
+      "step": 4393
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 6.069364161849712e-07,
+      "loss": 5.254,
+      "step": 4394
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.982658959537573e-07,
+      "loss": 5.5199,
+      "step": 4395
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.895953757225433e-07,
+      "loss": 5.4216,
+      "step": 4396
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.809248554913295e-07,
+      "loss": 5.1985,
+      "step": 4397
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.722543352601156e-07,
+      "loss": 5.327,
+      "step": 4398
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.635838150289018e-07,
+      "loss": 5.3288,
+      "step": 4399
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.549132947976879e-07,
+      "loss": 5.381,
+      "step": 4400
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.46242774566474e-07,
+      "loss": 5.3781,
+      "step": 4401
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.375722543352602e-07,
+      "loss": 5.1983,
+      "step": 4402
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.289017341040463e-07,
+      "loss": 5.1761,
+      "step": 4403
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.202312138728324e-07,
+      "loss": 5.1861,
+      "step": 4404
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.115606936416185e-07,
+      "loss": 5.3782,
+      "step": 4405
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 5.028901734104046e-07,
+      "loss": 5.305,
+      "step": 4406
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.942196531791908e-07,
+      "loss": 5.4233,
+      "step": 4407
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.855491329479768e-07,
+      "loss": 5.2848,
+      "step": 4408
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.768786127167629e-07,
+      "loss": 5.3075,
+      "step": 4409
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.6820809248554914e-07,
+      "loss": 5.1179,
+      "step": 4410
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.595375722543352e-07,
+      "loss": 5.1724,
+      "step": 4411
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.5086705202312137e-07,
+      "loss": 5.1686,
+      "step": 4412
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.421965317919075e-07,
+      "loss": 5.0993,
+      "step": 4413
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.3352601156069365e-07,
+      "loss": 5.2529,
+      "step": 4414
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.248554913294798e-07,
+      "loss": 4.9323,
+      "step": 4415
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.161849710982659e-07,
+      "loss": 5.1278,
+      "step": 4416
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.07514450867052e-07,
+      "loss": 4.9928,
+      "step": 4417
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.9884393063583815e-07,
+      "loss": 4.8098,
+      "step": 4418
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.901734104046243e-07,
+      "loss": 5.4066,
+      "step": 4419
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.8150289017341043e-07,
+      "loss": 5.3478,
+      "step": 4420
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.7283236994219657e-07,
+      "loss": 5.5408,
+      "step": 4421
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.6416184971098266e-07,
+      "loss": 5.363,
+      "step": 4422
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.554913294797688e-07,
+      "loss": 5.3668,
+      "step": 4423
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.468208092485549e-07,
+      "loss": 5.4698,
+      "step": 4424
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.38150289017341e-07,
+      "loss": 5.3154,
+      "step": 4425
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.2947976878612716e-07,
+      "loss": 5.3662,
+      "step": 4426
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.208092485549133e-07,
+      "loss": 5.4779,
+      "step": 4427
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.1213872832369944e-07,
+      "loss": 5.3426,
+      "step": 4428
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.034682080924856e-07,
+      "loss": 5.4471,
+      "step": 4429
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.9479768786127167e-07,
+      "loss": 5.3677,
+      "step": 4430
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.861271676300578e-07,
+      "loss": 5.4609,
+      "step": 4431
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.7745664739884395e-07,
+      "loss": 5.3772,
+      "step": 4432
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.687861271676301e-07,
+      "loss": 5.3796,
+      "step": 4433
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.601156069364162e-07,
+      "loss": 5.3896,
+      "step": 4434
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.514450867052023e-07,
+      "loss": 5.444,
+      "step": 4435
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.427745664739884e-07,
+      "loss": 5.4077,
+      "step": 4436
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.3410404624277457e-07,
+      "loss": 5.2058,
+      "step": 4437
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.2543352601156068e-07,
+      "loss": 5.4515,
+      "step": 4438
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.1676300578034682e-07,
+      "loss": 5.3286,
+      "step": 4439
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.0809248554913296e-07,
+      "loss": 5.4395,
+      "step": 4440
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.9942196531791908e-07,
+      "loss": 5.458,
+      "step": 4441
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.9075144508670522e-07,
+      "loss": 5.4846,
+      "step": 4442
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.8208092485549133e-07,
+      "loss": 5.4466,
+      "step": 4443
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.7341040462427744e-07,
+      "loss": 5.334,
+      "step": 4444
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.6473988439306358e-07,
+      "loss": 5.2473,
+      "step": 4445
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.5606936416184972e-07,
+      "loss": 5.424,
+      "step": 4446
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.4739884393063584e-07,
+      "loss": 5.4711,
+      "step": 4447
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.3872832369942197e-07,
+      "loss": 5.5162,
+      "step": 4448
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.300578034682081e-07,
+      "loss": 5.3185,
+      "step": 4449
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.213872832369942e-07,
+      "loss": 5.3675,
+      "step": 4450
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.1271676300578034e-07,
+      "loss": 5.2836,
+      "step": 4451
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0404624277456648e-07,
+      "loss": 5.2427,
+      "step": 4452
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 9.537572254335261e-08,
+      "loss": 5.5049,
+      "step": 4453
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 8.670520231213872e-08,
+      "loss": 5.2266,
+      "step": 4454
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 7.803468208092486e-08,
+      "loss": 5.0959,
+      "step": 4455
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.936416184971099e-08,
+      "loss": 5.3138,
+      "step": 4456
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.06936416184971e-08,
+      "loss": 5.1818,
+      "step": 4457
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 5.202312138728324e-08,
+      "loss": 4.8935,
+      "step": 4458
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 4.335260115606936e-08,
+      "loss": 5.278,
+      "step": 4459
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.4682080924855494e-08,
+      "loss": 4.4894,
+      "step": 4460
+    },
+    {
+      "epoch": 5.0,
+      "step": 4460,
+      "total_flos": 0.0,
+      "train_loss": 6.089906092074955,
+      "train_runtime": 28486.3453,
+      "train_samples_per_second": 5.009,
+      "train_steps_per_second": 0.157
+    }
+  ],
+  "max_steps": 4460,
+  "num_train_epochs": 5,
+  "total_flos": 0.0,
+  "trial_name": null,
+  "trial_params": null
+}