diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,28633 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9998427260812582,
+  "global_step": 4768,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2e-08,
+      "loss": 2.5328,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4e-08,
+      "loss": 2.5251,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 2.5486,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8e-08,
+      "loss": 2.5435,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 2.523,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 2.4938,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4e-07,
+      "loss": 2.6218,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6e-07,
+      "loss": 2.5497,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-07,
+      "loss": 2.5893,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 2.6216,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2e-07,
+      "loss": 2.5735,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 2.53,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6e-07,
+      "loss": 2.5208,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8e-07,
+      "loss": 2.6361,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 2.5842,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2e-07,
+      "loss": 2.5468,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4000000000000003e-07,
+      "loss": 2.562,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-07,
+      "loss": 2.4546,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8e-07,
+      "loss": 2.5826,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 2.5109,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 2.5289,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4e-07,
+      "loss": 2.4865,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6000000000000004e-07,
+      "loss": 2.5702,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 2.5875,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 2.5467,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2e-07,
+      "loss": 2.5894,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.4e-07,
+      "loss": 2.5949,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6e-07,
+      "loss": 2.5374,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.800000000000001e-07,
+      "loss": 2.6216,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 2.6127,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.200000000000001e-07,
+      "loss": 2.616,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.4e-07,
+      "loss": 2.6476,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.6e-07,
+      "loss": 2.5542,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 2.3941,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 2.5965,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.2e-07,
+      "loss": 2.5442,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.4e-07,
+      "loss": 2.5416,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.6e-07,
+      "loss": 2.5582,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.8e-07,
+      "loss": 2.5553,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 2.6,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.200000000000001e-07,
+      "loss": 2.5582,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.400000000000001e-07,
+      "loss": 2.5786,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.6e-07,
+      "loss": 2.6029,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.8e-07,
+      "loss": 2.501,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 2.4712,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 2.4941,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.400000000000001e-07,
+      "loss": 2.5221,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 2.597,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.800000000000001e-07,
+      "loss": 2.5149,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 2.561,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.02e-06,
+      "loss": 2.5512,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.04e-06,
+      "loss": 2.5369,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.06e-06,
+      "loss": 2.5369,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.08e-06,
+      "loss": 2.5062,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1e-06,
+      "loss": 2.4963,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.12e-06,
+      "loss": 2.5464,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.14e-06,
+      "loss": 2.5689,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1600000000000001e-06,
+      "loss": 2.5137,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1800000000000001e-06,
+      "loss": 2.4484,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 2.538,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2200000000000002e-06,
+      "loss": 2.4938,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2400000000000002e-06,
+      "loss": 2.5767,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.26e-06,
+      "loss": 2.4947,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.28e-06,
+      "loss": 2.5412,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3e-06,
+      "loss": 2.5663,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.32e-06,
+      "loss": 2.4796,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.34e-06,
+      "loss": 2.4463,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3600000000000001e-06,
+      "loss": 2.4821,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 2.5057,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 2.5066,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.42e-06,
+      "loss": 2.4425,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.44e-06,
+      "loss": 2.5584,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.46e-06,
+      "loss": 2.428,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.48e-06,
+      "loss": 2.4955,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5e-06,
+      "loss": 2.4006,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.52e-06,
+      "loss": 2.5879,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.54e-06,
+      "loss": 2.516,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.56e-06,
+      "loss": 2.4413,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5800000000000001e-06,
+      "loss": 2.4099,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 2.5,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6200000000000002e-06,
+      "loss": 2.4217,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6400000000000002e-06,
+      "loss": 2.3733,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6600000000000002e-06,
+      "loss": 2.5123,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 2.4603,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 2.4808,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.72e-06,
+      "loss": 2.4482,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.74e-06,
+      "loss": 2.5016,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.76e-06,
+      "loss": 2.5101,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7800000000000001e-06,
+      "loss": 2.4886,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 2.4994,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8200000000000002e-06,
+      "loss": 2.5504,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8400000000000002e-06,
+      "loss": 2.4196,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8600000000000002e-06,
+      "loss": 2.399,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 2.4237,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 2.4667,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 2.4934,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.94e-06,
+      "loss": 2.3601,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9600000000000003e-06,
+      "loss": 2.4602,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.98e-06,
+      "loss": 2.4906,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 2.4347,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.02e-06,
+      "loss": 2.4296,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.04e-06,
+      "loss": 2.4267,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.06e-06,
+      "loss": 2.5489,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.08e-06,
+      "loss": 2.3433,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 2.4054,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.12e-06,
+      "loss": 2.4272,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1400000000000003e-06,
+      "loss": 2.4808,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.16e-06,
+      "loss": 2.5048,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1800000000000003e-06,
+      "loss": 2.4035,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2e-06,
+      "loss": 2.5168,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2200000000000003e-06,
+      "loss": 2.4057,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.24e-06,
+      "loss": 2.3867,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2600000000000004e-06,
+      "loss": 2.4764,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.28e-06,
+      "loss": 2.3829,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 2.5088,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3200000000000002e-06,
+      "loss": 2.4269,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3400000000000005e-06,
+      "loss": 2.4616,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3600000000000003e-06,
+      "loss": 2.4613,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.38e-06,
+      "loss": 2.4278,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 2.4212,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.42e-06,
+      "loss": 2.424,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.4400000000000004e-06,
+      "loss": 2.3639,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.46e-06,
+      "loss": 2.3569,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.4800000000000004e-06,
+      "loss": 2.4177,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5e-06,
+      "loss": 2.4697,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.52e-06,
+      "loss": 2.4481,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5400000000000002e-06,
+      "loss": 2.4008,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.56e-06,
+      "loss": 2.357,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5800000000000003e-06,
+      "loss": 2.403,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.6e-06,
+      "loss": 2.3758,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.6200000000000003e-06,
+      "loss": 2.4465,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.64e-06,
+      "loss": 2.4126,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.6600000000000004e-06,
+      "loss": 2.3781,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.68e-06,
+      "loss": 2.4524,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 2.5464,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7200000000000002e-06,
+      "loss": 2.3527,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7400000000000004e-06,
+      "loss": 2.4728,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7600000000000003e-06,
+      "loss": 2.4063,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.7800000000000005e-06,
+      "loss": 2.3898,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 2.4136,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.82e-06,
+      "loss": 2.2612,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.84e-06,
+      "loss": 2.3523,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.86e-06,
+      "loss": 2.3139,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.88e-06,
+      "loss": 2.4568,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9e-06,
+      "loss": 2.3361,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.92e-06,
+      "loss": 2.3848,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 2.4558,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.96e-06,
+      "loss": 2.3316,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9800000000000003e-06,
+      "loss": 2.4527,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3e-06,
+      "loss": 2.4386,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.0200000000000003e-06,
+      "loss": 2.3402,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.04e-06,
+      "loss": 2.4051,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.0600000000000003e-06,
+      "loss": 2.4011,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.08e-06,
+      "loss": 2.3627,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 2.3318,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.12e-06,
+      "loss": 2.4862,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1400000000000004e-06,
+      "loss": 2.3589,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1600000000000002e-06,
+      "loss": 2.4397,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1800000000000005e-06,
+      "loss": 2.3911,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 2.3437,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2200000000000005e-06,
+      "loss": 2.4328,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2400000000000003e-06,
+      "loss": 2.3783,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2600000000000006e-06,
+      "loss": 2.4134,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2800000000000004e-06,
+      "loss": 2.3977,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 2.2898,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3200000000000004e-06,
+      "loss": 2.4317,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.3400000000000006e-06,
+      "loss": 2.4104,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.3600000000000004e-06,
+      "loss": 2.3403,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.3800000000000007e-06,
+      "loss": 2.3863,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 2.4678,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.4200000000000007e-06,
+      "loss": 2.4299,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.44e-06,
+      "loss": 2.3295,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.46e-06,
+      "loss": 2.3113,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.48e-06,
+      "loss": 2.3897,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.5e-06,
+      "loss": 2.4306,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.52e-06,
+      "loss": 2.3533,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.54e-06,
+      "loss": 2.4473,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.5600000000000002e-06,
+      "loss": 2.3461,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.58e-06,
+      "loss": 2.3761,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 2.4173,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.62e-06,
+      "loss": 2.3709,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.6400000000000003e-06,
+      "loss": 2.4205,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.66e-06,
+      "loss": 2.3207,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.6800000000000003e-06,
+      "loss": 2.4172,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.7e-06,
+      "loss": 2.3705,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.7200000000000004e-06,
+      "loss": 2.3972,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.74e-06,
+      "loss": 2.3359,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.7600000000000004e-06,
+      "loss": 2.4597,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.7800000000000002e-06,
+      "loss": 2.3975,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 2.3311,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.820000000000001e-06,
+      "loss": 2.4286,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.8400000000000005e-06,
+      "loss": 2.2739,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.86e-06,
+      "loss": 2.3067,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.88e-06,
+      "loss": 2.433,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 2.3872,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 2.3819,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.94e-06,
+      "loss": 2.3915,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.96e-06,
+      "loss": 2.4285,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.980000000000001e-06,
+      "loss": 2.302,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 2.4101,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.0200000000000005e-06,
+      "loss": 2.2795,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.04e-06,
+      "loss": 2.3601,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.060000000000001e-06,
+      "loss": 2.3147,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.08e-06,
+      "loss": 2.2904,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1e-06,
+      "loss": 2.3637,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.12e-06,
+      "loss": 2.3818,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.14e-06,
+      "loss": 2.3136,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.16e-06,
+      "loss": 2.4371,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.18e-06,
+      "loss": 2.372,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 2.4436,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.22e-06,
+      "loss": 2.2448,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.24e-06,
+      "loss": 2.3976,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.26e-06,
+      "loss": 2.4157,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.2800000000000005e-06,
+      "loss": 2.3369,
+      "step": 214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.3e-06,
+      "loss": 2.3693,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.32e-06,
+      "loss": 2.4596,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.34e-06,
+      "loss": 2.2976,
+      "step": 217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.360000000000001e-06,
+      "loss": 2.4137,
+      "step": 218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.38e-06,
+      "loss": 2.2767,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.4e-06,
+      "loss": 2.2396,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.42e-06,
+      "loss": 2.3585,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.440000000000001e-06,
+      "loss": 2.4332,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.4600000000000005e-06,
+      "loss": 2.3455,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.48e-06,
+      "loss": 2.3663,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.5e-06,
+      "loss": 2.4619,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.520000000000001e-06,
+      "loss": 2.3736,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.540000000000001e-06,
+      "loss": 2.3941,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.56e-06,
+      "loss": 2.3656,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.58e-06,
+      "loss": 2.3217,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 2.3652,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.620000000000001e-06,
+      "loss": 2.3815,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.6400000000000005e-06,
+      "loss": 2.4291,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.66e-06,
+      "loss": 2.3979,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.680000000000001e-06,
+      "loss": 2.2746,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7e-06,
+      "loss": 2.4069,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.7200000000000005e-06,
+      "loss": 2.3824,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.74e-06,
+      "loss": 2.4456,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.76e-06,
+      "loss": 2.3342,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.78e-06,
+      "loss": 2.3679,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 2.3962,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.8200000000000004e-06,
+      "loss": 2.4315,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.84e-06,
+      "loss": 2.3256,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.86e-06,
+      "loss": 2.3288,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.880000000000001e-06,
+      "loss": 2.3903,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 2.3203,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.92e-06,
+      "loss": 2.311,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.94e-06,
+      "loss": 2.307,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.960000000000001e-06,
+      "loss": 2.2922,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.980000000000001e-06,
+      "loss": 2.3276,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5e-06,
+      "loss": 2.3496,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.02e-06,
+      "loss": 2.2829,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.04e-06,
+      "loss": 2.2565,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.060000000000001e-06,
+      "loss": 2.3674,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.0800000000000005e-06,
+      "loss": 2.3762,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1e-06,
+      "loss": 2.4177,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.12e-06,
+      "loss": 2.3196,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.140000000000001e-06,
+      "loss": 2.3842,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.1600000000000006e-06,
+      "loss": 2.426,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.18e-06,
+      "loss": 2.3261,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2e-06,
+      "loss": 2.3683,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.220000000000001e-06,
+      "loss": 2.3414,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.240000000000001e-06,
+      "loss": 2.3236,
+      "step": 262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.2600000000000005e-06,
+      "loss": 2.2147,
+      "step": 263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.28e-06,
+      "loss": 2.3784,
+      "step": 264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.300000000000001e-06,
+      "loss": 2.3315,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.320000000000001e-06,
+      "loss": 2.3022,
+      "step": 266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.3400000000000005e-06,
+      "loss": 2.3728,
+      "step": 267
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.36e-06,
+      "loss": 2.303,
+      "step": 268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.380000000000001e-06,
+      "loss": 2.4627,
+      "step": 269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 2.367,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.420000000000001e-06,
+      "loss": 2.3701,
+      "step": 271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.4400000000000004e-06,
+      "loss": 2.4093,
+      "step": 272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.460000000000001e-06,
+      "loss": 2.3489,
+      "step": 273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.480000000000001e-06,
+      "loss": 2.3311,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 2.3895,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.5200000000000005e-06,
+      "loss": 2.3391,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.540000000000001e-06,
+      "loss": 2.267,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.560000000000001e-06,
+      "loss": 2.3282,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.580000000000001e-06,
+      "loss": 2.346,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 2.3606,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.620000000000001e-06,
+      "loss": 2.3275,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.64e-06,
+      "loss": 2.3064,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.66e-06,
+      "loss": 2.3908,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.68e-06,
+      "loss": 2.2859,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.7e-06,
+      "loss": 2.4631,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.72e-06,
+      "loss": 2.4007,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.74e-06,
+      "loss": 2.4275,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.76e-06,
+      "loss": 2.3024,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.78e-06,
+      "loss": 2.3426,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.8e-06,
+      "loss": 2.3389,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.82e-06,
+      "loss": 2.3816,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.84e-06,
+      "loss": 2.292,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.86e-06,
+      "loss": 2.4027,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 2.3467,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9e-06,
+      "loss": 2.3099,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.92e-06,
+      "loss": 2.3614,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.94e-06,
+      "loss": 2.4625,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9600000000000005e-06,
+      "loss": 2.3437,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.98e-06,
+      "loss": 2.3674,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6e-06,
+      "loss": 2.3352,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.02e-06,
+      "loss": 2.394,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.040000000000001e-06,
+      "loss": 2.3795,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.0600000000000004e-06,
+      "loss": 2.4109,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.08e-06,
+      "loss": 2.3835,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.1e-06,
+      "loss": 2.346,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.120000000000001e-06,
+      "loss": 2.4127,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.1400000000000005e-06,
+      "loss": 2.4423,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.16e-06,
+      "loss": 2.4154,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.18e-06,
+      "loss": 2.3368,
+      "step": 309
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 2.44,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.220000000000001e-06,
+      "loss": 2.3857,
+      "step": 311
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.24e-06,
+      "loss": 2.2287,
+      "step": 312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.26e-06,
+      "loss": 2.3987,
+      "step": 313
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.280000000000001e-06,
+      "loss": 2.256,
+      "step": 314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 2.2168,
+      "step": 315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.3200000000000005e-06,
+      "loss": 2.3558,
+      "step": 316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.34e-06,
+      "loss": 2.3889,
+      "step": 317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.360000000000001e-06,
+      "loss": 2.3836,
+      "step": 318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.380000000000001e-06,
+      "loss": 2.3449,
+      "step": 319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 2.3675,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.42e-06,
+      "loss": 2.398,
+      "step": 321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.440000000000001e-06,
+      "loss": 2.4071,
+      "step": 322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.460000000000001e-06,
+      "loss": 2.2562,
+      "step": 323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.480000000000001e-06,
+      "loss": 2.3139,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 2.2259,
+      "step": 325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.520000000000001e-06,
+      "loss": 2.3461,
+      "step": 326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.540000000000001e-06,
+      "loss": 2.388,
+      "step": 327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.560000000000001e-06,
+      "loss": 2.3885,
+      "step": 328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.5800000000000005e-06,
+      "loss": 2.4175,
+      "step": 329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 2.2422,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.620000000000001e-06,
+      "loss": 2.3245,
+      "step": 331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.640000000000001e-06,
+      "loss": 2.402,
+      "step": 332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.660000000000001e-06,
+      "loss": 2.4081,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.680000000000001e-06,
+      "loss": 2.3671,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 2.3439,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.720000000000001e-06,
+      "loss": 2.3359,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.740000000000001e-06,
+      "loss": 2.3358,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.760000000000001e-06,
+      "loss": 2.43,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.780000000000001e-06,
+      "loss": 2.2929,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 2.2838,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.820000000000001e-06,
+      "loss": 2.3024,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.8400000000000014e-06,
+      "loss": 2.2875,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.860000000000001e-06,
+      "loss": 2.2141,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.88e-06,
+      "loss": 2.3402,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.9e-06,
+      "loss": 2.3455,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.92e-06,
+      "loss": 2.4424,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.9400000000000005e-06,
+      "loss": 2.3687,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.96e-06,
+      "loss": 2.3128,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 6.98e-06,
+      "loss": 2.3986,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7e-06,
+      "loss": 2.3142,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.0200000000000006e-06,
+      "loss": 2.3391,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.04e-06,
+      "loss": 2.3736,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.06e-06,
+      "loss": 2.3578,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.08e-06,
+      "loss": 2.3718,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.100000000000001e-06,
+      "loss": 2.3966,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.1200000000000004e-06,
+      "loss": 2.3172,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.14e-06,
+      "loss": 2.3746,
+      "step": 357
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.16e-06,
+      "loss": 2.3507,
+      "step": 358
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.180000000000001e-06,
+      "loss": 2.2613,
+      "step": 359
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 2.2931,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.22e-06,
+      "loss": 2.3,
+      "step": 361
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.24e-06,
+      "loss": 2.3273,
+      "step": 362
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.260000000000001e-06,
+      "loss": 2.3516,
+      "step": 363
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.280000000000001e-06,
+      "loss": 2.3544,
+      "step": 364
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.3e-06,
+      "loss": 2.4057,
+      "step": 365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.32e-06,
+      "loss": 2.3231,
+      "step": 366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.340000000000001e-06,
+      "loss": 2.3381,
+      "step": 367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.360000000000001e-06,
+      "loss": 2.3037,
+      "step": 368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.3800000000000005e-06,
+      "loss": 2.3938,
+      "step": 369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.4e-06,
+      "loss": 2.3127,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.420000000000001e-06,
+      "loss": 2.2572,
+      "step": 371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.440000000000001e-06,
+      "loss": 2.3125,
+      "step": 372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.4600000000000006e-06,
+      "loss": 2.3361,
+      "step": 373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.48e-06,
+      "loss": 2.2685,
+      "step": 374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 2.3934,
+      "step": 375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.520000000000001e-06,
+      "loss": 2.3322,
+      "step": 376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.540000000000001e-06,
+      "loss": 2.3319,
+      "step": 377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.5600000000000005e-06,
+      "loss": 2.2812,
+      "step": 378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.58e-06,
+      "loss": 2.4279,
+      "step": 379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 2.3516,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.620000000000001e-06,
+      "loss": 2.3605,
+      "step": 381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.640000000000001e-06,
+      "loss": 2.3459,
+      "step": 382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.660000000000001e-06,
+      "loss": 2.2894,
+      "step": 383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.680000000000001e-06,
+      "loss": 2.3024,
+      "step": 384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.7e-06,
+      "loss": 2.2971,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.72e-06,
+      "loss": 2.3208,
+      "step": 386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.74e-06,
+      "loss": 2.346,
+      "step": 387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.76e-06,
+      "loss": 2.3455,
+      "step": 388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.78e-06,
+      "loss": 2.4215,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 2.4596,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.820000000000001e-06,
+      "loss": 2.3747,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.840000000000001e-06,
+      "loss": 2.2961,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.860000000000001e-06,
+      "loss": 2.3224,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.88e-06,
+      "loss": 2.3472,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.9e-06,
+      "loss": 2.3214,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.92e-06,
+      "loss": 2.306,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.94e-06,
+      "loss": 2.3234,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.960000000000002e-06,
+      "loss": 2.3617,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 7.980000000000002e-06,
+      "loss": 2.3866,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 2.3019,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.020000000000001e-06,
+      "loss": 2.3604,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.040000000000001e-06,
+      "loss": 2.3669,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.06e-06,
+      "loss": 2.4265,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.08e-06,
+      "loss": 2.3686,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.1e-06,
+      "loss": 2.3911,
+      "step": 405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.120000000000002e-06,
+      "loss": 2.3336,
+      "step": 406
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.14e-06,
+      "loss": 2.2708,
+      "step": 407
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.16e-06,
+      "loss": 2.264,
+      "step": 408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.18e-06,
+      "loss": 2.4037,
+      "step": 409
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.2e-06,
+      "loss": 2.33,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.220000000000001e-06,
+      "loss": 2.3443,
+      "step": 411
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.24e-06,
+      "loss": 2.3843,
+      "step": 412
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.26e-06,
+      "loss": 2.3489,
+      "step": 413
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.28e-06,
+      "loss": 2.3777,
+      "step": 414
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.3e-06,
+      "loss": 2.368,
+      "step": 415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.32e-06,
+      "loss": 2.3814,
+      "step": 416
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.34e-06,
+      "loss": 2.3449,
+      "step": 417
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.36e-06,
+      "loss": 2.397,
+      "step": 418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.380000000000001e-06,
+      "loss": 2.3188,
+      "step": 419
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 2.3264,
+      "step": 420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.42e-06,
+      "loss": 2.3822,
+      "step": 421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.44e-06,
+      "loss": 2.3954,
+      "step": 422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.46e-06,
+      "loss": 2.3636,
+      "step": 423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.48e-06,
+      "loss": 2.3949,
+      "step": 424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.5e-06,
+      "loss": 2.434,
+      "step": 425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.52e-06,
+      "loss": 2.2858,
+      "step": 426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.540000000000001e-06,
+      "loss": 2.3239,
+      "step": 427
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.560000000000001e-06,
+      "loss": 2.371,
+      "step": 428
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.580000000000001e-06,
+      "loss": 2.3821,
+      "step": 429
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.6e-06,
+      "loss": 2.332,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.62e-06,
+      "loss": 2.347,
+      "step": 431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.64e-06,
+      "loss": 2.401,
+      "step": 432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.66e-06,
+      "loss": 2.2833,
+      "step": 433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.68e-06,
+      "loss": 2.3607,
+      "step": 434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.700000000000001e-06,
+      "loss": 2.4096,
+      "step": 435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.720000000000001e-06,
+      "loss": 2.325,
+      "step": 436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.740000000000001e-06,
+      "loss": 2.3061,
+      "step": 437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.76e-06,
+      "loss": 2.3256,
+      "step": 438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.78e-06,
+      "loss": 2.3427,
+      "step": 439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.8e-06,
+      "loss": 2.3972,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.82e-06,
+      "loss": 2.2532,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.84e-06,
+      "loss": 2.3241,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.860000000000002e-06,
+      "loss": 2.3457,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.880000000000001e-06,
+      "loss": 2.3249,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 2.3697,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.920000000000001e-06,
+      "loss": 2.3601,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.94e-06,
+      "loss": 2.2603,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.96e-06,
+      "loss": 2.2614,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 8.98e-06,
+      "loss": 2.3903,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9e-06,
+      "loss": 2.2095,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.020000000000002e-06,
+      "loss": 2.2615,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.040000000000002e-06,
+      "loss": 2.3594,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.060000000000001e-06,
+      "loss": 2.2643,
+      "step": 453
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.080000000000001e-06,
+      "loss": 2.373,
+      "step": 454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 2.3148,
+      "step": 455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.12e-06,
+      "loss": 2.422,
+      "step": 456
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.14e-06,
+      "loss": 2.4152,
+      "step": 457
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.16e-06,
+      "loss": 2.3618,
+      "step": 458
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.180000000000002e-06,
+      "loss": 2.3823,
+      "step": 459
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 2.3393,
+      "step": 460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.220000000000002e-06,
+      "loss": 2.3036,
+      "step": 461
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.240000000000001e-06,
+      "loss": 2.305,
+      "step": 462
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.260000000000001e-06,
+      "loss": 2.2981,
+      "step": 463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.280000000000001e-06,
+      "loss": 2.3284,
+      "step": 464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.3e-06,
+      "loss": 2.3543,
+      "step": 465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.32e-06,
+      "loss": 2.3098,
+      "step": 466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.340000000000002e-06,
+      "loss": 2.3526,
+      "step": 467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.360000000000002e-06,
+      "loss": 2.3317,
+      "step": 468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.38e-06,
+      "loss": 2.4108,
+      "step": 469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.4e-06,
+      "loss": 2.3344,
+      "step": 470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.42e-06,
+      "loss": 2.3459,
+      "step": 471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.440000000000001e-06,
+      "loss": 2.2248,
+      "step": 472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.460000000000001e-06,
+      "loss": 2.3495,
+      "step": 473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.48e-06,
+      "loss": 2.3382,
+      "step": 474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.5e-06,
+      "loss": 2.372,
+      "step": 475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.52e-06,
+      "loss": 2.3911,
+      "step": 476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.54e-06,
+      "loss": 2.4211,
+      "step": 477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.56e-06,
+      "loss": 2.2151,
+      "step": 478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.58e-06,
+      "loss": 2.3017,
+      "step": 479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 2.3212,
+      "step": 480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.620000000000001e-06,
+      "loss": 2.3456,
+      "step": 481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.640000000000001e-06,
+      "loss": 2.3515,
+      "step": 482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.66e-06,
+      "loss": 2.3565,
+      "step": 483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.68e-06,
+      "loss": 2.3116,
+      "step": 484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.7e-06,
+      "loss": 2.3044,
+      "step": 485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.72e-06,
+      "loss": 2.4271,
+      "step": 486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.74e-06,
+      "loss": 2.3439,
+      "step": 487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.760000000000001e-06,
+      "loss": 2.3342,
+      "step": 488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.780000000000001e-06,
+      "loss": 2.3475,
+      "step": 489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 2.3422,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.820000000000001e-06,
+      "loss": 2.3415,
+      "step": 491
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.84e-06,
+      "loss": 2.3085,
+      "step": 492
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86e-06,
+      "loss": 2.3502,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88e-06,
+      "loss": 2.3317,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.9e-06,
+      "loss": 2.2023,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 2.2907,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.940000000000001e-06,
+      "loss": 2.3587,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.960000000000001e-06,
+      "loss": 2.378,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.980000000000001e-06,
+      "loss": 2.38,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1e-05,
+      "loss": 2.2965,
+      "step": 500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999998645462876e-06,
+      "loss": 2.3523,
+      "step": 501
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99999458185223e-06,
+      "loss": 2.4007,
+      "step": 502
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999987809170272e-06,
+      "loss": 2.335,
+      "step": 503
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999978327420663e-06,
+      "loss": 2.2951,
+      "step": 504
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999966136608545e-06,
+      "loss": 2.4493,
+      "step": 505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999951236740525e-06,
+      "loss": 2.2988,
+      "step": 506
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999933627824672e-06,
+      "loss": 2.3326,
+      "step": 507
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99991330987053e-06,
+      "loss": 2.3771,
+      "step": 508
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999890282889104e-06,
+      "loss": 2.3267,
+      "step": 509
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999864546892874e-06,
+      "loss": 2.3771,
+      "step": 510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999836101895782e-06,
+      "loss": 2.3612,
+      "step": 511
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999804947913241e-06,
+      "loss": 2.3291,
+      "step": 512
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999771084962129e-06,
+      "loss": 2.3259,
+      "step": 513
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999734513060795e-06,
+      "loss": 2.3488,
+      "step": 514
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999695232229054e-06,
+      "loss": 2.327,
+      "step": 515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999653242488187e-06,
+      "loss": 2.3433,
+      "step": 516
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99960854386095e-06,
+      "loss": 2.3568,
+      "step": 517
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999561136371556e-06,
+      "loss": 2.166,
+      "step": 518
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999511020045692e-06,
+      "loss": 2.3752,
+      "step": 519
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999458194910512e-06,
+      "loss": 2.4264,
+      "step": 520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999402660994639e-06,
+      "loss": 2.2545,
+      "step": 521
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999344418328161e-06,
+      "loss": 2.3542,
+      "step": 522
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999283466942636e-06,
+      "loss": 2.2633,
+      "step": 523
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999219806871086e-06,
+      "loss": 2.3582,
+      "step": 524
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999153438148004e-06,
+      "loss": 2.2435,
+      "step": 525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99908436080935e-06,
+      "loss": 2.2561,
+      "step": 526
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.999012574892552e-06,
+      "loss": 2.3137,
+      "step": 527
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998938080436503e-06,
+      "loss": 2.3009,
+      "step": 528
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998860877481567e-06,
+      "loss": 2.4184,
+      "step": 529
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99878096606957e-06,
+      "loss": 2.3259,
+      "step": 530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998698346243814e-06,
+      "loss": 2.3993,
+      "step": 531
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99861301804906e-06,
+      "loss": 2.363,
+      "step": 532
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99852498153154e-06,
+      "loss": 2.3312,
+      "step": 533
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998434236738958e-06,
+      "loss": 2.2813,
+      "step": 534
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998340783720477e-06,
+      "loss": 2.2685,
+      "step": 535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99824462252673e-06,
+      "loss": 2.3794,
+      "step": 536
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.998145753209823e-06,
+      "loss": 2.383,
+      "step": 537
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.99804417582332e-06,
+      "loss": 2.3484,
+      "step": 538
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997939890422262e-06,
+      "loss": 2.2568,
+      "step": 539
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997832897063148e-06,
+      "loss": 2.3411,
+      "step": 540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997723195803952e-06,
+      "loss": 2.3657,
+      "step": 541
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997610786704111e-06,
+      "loss": 2.3292,
+      "step": 542
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997495669824525e-06,
+      "loss": 2.338,
+      "step": 543
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997377845227577e-06,
+      "loss": 2.4201,
+      "step": 544
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997257312977094e-06,
+      "loss": 2.3209,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997134073138389e-06,
+      "loss": 2.2948,
+      "step": 546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.997008125778234e-06,
+      "loss": 2.3215,
+      "step": 547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.996879470964869e-06,
+      "loss": 2.3853,
+      "step": 548
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.996748108768002e-06,
+      "loss": 2.3296,
+      "step": 549
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.996614039258803e-06,
+      "loss": 2.3179,
+      "step": 550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.996477262509919e-06,
+      "loss": 2.3494,
+      "step": 551
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.996337778595452e-06,
+      "loss": 2.3588,
+      "step": 552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.996195587590981e-06,
+      "loss": 2.4379,
+      "step": 553
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.996050689573545e-06,
+      "loss": 2.3278,
+      "step": 554
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.995903084621651e-06,
+      "loss": 2.3485,
+      "step": 555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.995752772815275e-06,
+      "loss": 2.4472,
+      "step": 556
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.995599754235857e-06,
+      "loss": 2.3863,
+      "step": 557
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.995444028966307e-06,
+      "loss": 2.4145,
+      "step": 558
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.995285597090997e-06,
+      "loss": 2.3113,
+      "step": 559
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.995124458695769e-06,
+      "loss": 2.3703,
+      "step": 560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.994960613867929e-06,
+      "loss": 2.306,
+      "step": 561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.99479406269625e-06,
+      "loss": 2.3326,
+      "step": 562
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.994624805270975e-06,
+      "loss": 2.2488,
+      "step": 563
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.994452841683808e-06,
+      "loss": 2.4193,
+      "step": 564
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.994278172027921e-06,
+      "loss": 2.322,
+      "step": 565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.994100796397954e-06,
+      "loss": 2.4212,
+      "step": 566
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.993920714890011e-06,
+      "loss": 2.3506,
+      "step": 567
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.993737927601663e-06,
+      "loss": 2.2288,
+      "step": 568
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.993552434631947e-06,
+      "loss": 2.3017,
+      "step": 569
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.993364236081367e-06,
+      "loss": 2.2607,
+      "step": 570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.993173332051888e-06,
+      "loss": 2.372,
+      "step": 571
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.992979722646948e-06,
+      "loss": 2.2749,
+      "step": 572
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.992783407971448e-06,
+      "loss": 2.33,
+      "step": 573
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.99258438813175e-06,
+      "loss": 2.3386,
+      "step": 574
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.99238266323569e-06,
+      "loss": 2.3194,
+      "step": 575
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.992178233392565e-06,
+      "loss": 2.2633,
+      "step": 576
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.991971098713136e-06,
+      "loss": 2.4658,
+      "step": 577
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.991761259309633e-06,
+      "loss": 2.3824,
+      "step": 578
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.99154871529575e-06,
+      "loss": 2.4297,
+      "step": 579
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.991333466786648e-06,
+      "loss": 2.3448,
+      "step": 580
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.991115513898947e-06,
+      "loss": 2.281,
+      "step": 581
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.990894856750744e-06,
+      "loss": 2.3974,
+      "step": 582
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.99067149546159e-06,
+      "loss": 2.3187,
+      "step": 583
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.990445430152507e-06,
+      "loss": 2.3347,
+      "step": 584
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.990216660945979e-06,
+      "loss": 2.3088,
+      "step": 585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.989985187965955e-06,
+      "loss": 2.3382,
+      "step": 586
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.989751011337856e-06,
+      "loss": 2.306,
+      "step": 587
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.98951413118856e-06,
+      "loss": 2.3033,
+      "step": 588
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.98927454764641e-06,
+      "loss": 2.2905,
+      "step": 589
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.989032260841216e-06,
+      "loss": 2.305,
+      "step": 590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.988787270904255e-06,
+      "loss": 2.2811,
+      "step": 591
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.988539577968265e-06,
+      "loss": 2.3013,
+      "step": 592
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.988289182167451e-06,
+      "loss": 2.3153,
+      "step": 593
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.988036083637478e-06,
+      "loss": 2.3226,
+      "step": 594
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.987780282515483e-06,
+      "loss": 2.3378,
+      "step": 595
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.98752177894006e-06,
+      "loss": 2.3896,
+      "step": 596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.987260573051268e-06,
+      "loss": 2.3887,
+      "step": 597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.986996664990637e-06,
+      "loss": 2.3781,
+      "step": 598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.986730054901153e-06,
+      "loss": 2.4036,
+      "step": 599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.98646074292727e-06,
+      "loss": 2.4033,
+      "step": 600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.986188729214908e-06,
+      "loss": 2.2803,
+      "step": 601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.985914013911442e-06,
+      "loss": 2.2635,
+      "step": 602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.985636597165724e-06,
+      "loss": 2.2875,
+      "step": 603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.985356479128056e-06,
+      "loss": 2.263,
+      "step": 604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.985073659950215e-06,
+      "loss": 2.2955,
+      "step": 605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.984788139785432e-06,
+      "loss": 2.3083,
+      "step": 606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.984499918788413e-06,
+      "loss": 2.3861,
+      "step": 607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.984208997115314e-06,
+      "loss": 2.4066,
+      "step": 608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.983915374923763e-06,
+      "loss": 2.3866,
+      "step": 609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.983619052372847e-06,
+      "loss": 2.3113,
+      "step": 610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.983320029623122e-06,
+      "loss": 2.3004,
+      "step": 611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.983018306836601e-06,
+      "loss": 2.1896,
+      "step": 612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.98271388417676e-06,
+      "loss": 2.3066,
+      "step": 613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.98240676180854e-06,
+      "loss": 2.4289,
+      "step": 614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.98209693989835e-06,
+      "loss": 2.4037,
+      "step": 615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.981784418614048e-06,
+      "loss": 2.2436,
+      "step": 616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.981469198124968e-06,
+      "loss": 2.2494,
+      "step": 617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.981151278601899e-06,
+      "loss": 2.2616,
+      "step": 618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.980830660217096e-06,
+      "loss": 2.3347,
+      "step": 619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.980507343144273e-06,
+      "loss": 2.3506,
+      "step": 620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.98018132755861e-06,
+      "loss": 2.303,
+      "step": 621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.979852613636745e-06,
+      "loss": 2.3265,
+      "step": 622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.979521201556783e-06,
+      "loss": 2.3444,
+      "step": 623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.979187091498283e-06,
+      "loss": 2.3664,
+      "step": 624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.978850283642276e-06,
+      "loss": 2.3271,
+      "step": 625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.978510778171247e-06,
+      "loss": 2.3328,
+      "step": 626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.978168575269148e-06,
+      "loss": 2.3107,
+      "step": 627
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.977823675121382e-06,
+      "loss": 2.2926,
+      "step": 628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.977476077914829e-06,
+      "loss": 2.3318,
+      "step": 629
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.97712578383782e-06,
+      "loss": 2.3327,
+      "step": 630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.97677279308015e-06,
+      "loss": 2.3096,
+      "step": 631
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.97641710583307e-06,
+      "loss": 2.2646,
+      "step": 632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.976058722289303e-06,
+      "loss": 2.2949,
+      "step": 633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.975697642643023e-06,
+      "loss": 2.4022,
+      "step": 634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.975333867089869e-06,
+      "loss": 2.3564,
+      "step": 635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.974967395826941e-06,
+      "loss": 2.2899,
+      "step": 636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.974598229052797e-06,
+      "loss": 2.3193,
+      "step": 637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.974226366967458e-06,
+      "loss": 2.308,
+      "step": 638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.973851809772404e-06,
+      "loss": 2.4376,
+      "step": 639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.973474557670574e-06,
+      "loss": 2.3628,
+      "step": 640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.973094610866373e-06,
+      "loss": 2.3946,
+      "step": 641
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.972711969565659e-06,
+      "loss": 2.3899,
+      "step": 642
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.972326633975754e-06,
+      "loss": 2.4102,
+      "step": 643
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.971938604305435e-06,
+      "loss": 2.3951,
+      "step": 644
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.971547880764946e-06,
+      "loss": 2.3302,
+      "step": 645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.971154463565987e-06,
+      "loss": 2.2062,
+      "step": 646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.970758352921715e-06,
+      "loss": 2.3419,
+      "step": 647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.97035954904675e-06,
+      "loss": 2.3466,
+      "step": 648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.96995805215717e-06,
+      "loss": 2.3287,
+      "step": 649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.969553862470509e-06,
+      "loss": 2.3923,
+      "step": 650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.969146980205768e-06,
+      "loss": 2.3743,
+      "step": 651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.968737405583398e-06,
+      "loss": 2.257,
+      "step": 652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.968325138825314e-06,
+      "loss": 2.3031,
+      "step": 653
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.96791018015489e-06,
+      "loss": 2.2627,
+      "step": 654
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.967492529796952e-06,
+      "loss": 2.2612,
+      "step": 655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.967072187977793e-06,
+      "loss": 2.3884,
+      "step": 656
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.966649154925161e-06,
+      "loss": 2.2971,
+      "step": 657
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.96622343086826e-06,
+      "loss": 2.3568,
+      "step": 658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.965795016037752e-06,
+      "loss": 2.4179,
+      "step": 659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.965363910665762e-06,
+      "loss": 2.3195,
+      "step": 660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.964930114985867e-06,
+      "loss": 2.3386,
+      "step": 661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.964493629233106e-06,
+      "loss": 2.2258,
+      "step": 662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.96405445364397e-06,
+      "loss": 2.284,
+      "step": 663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.963612588456413e-06,
+      "loss": 2.3382,
+      "step": 664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.963168033909843e-06,
+      "loss": 2.3628,
+      "step": 665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.962720790245128e-06,
+      "loss": 2.3344,
+      "step": 666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.96227085770459e-06,
+      "loss": 2.2771,
+      "step": 667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.961818236532012e-06,
+      "loss": 2.4291,
+      "step": 668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.961362926972627e-06,
+      "loss": 2.2767,
+      "step": 669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.96090492927313e-06,
+      "loss": 2.3663,
+      "step": 670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.960444243681671e-06,
+      "loss": 2.3815,
+      "step": 671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.959980870447853e-06,
+      "loss": 2.3856,
+      "step": 672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.959514809822746e-06,
+      "loss": 2.3358,
+      "step": 673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.959046062058863e-06,
+      "loss": 2.3496,
+      "step": 674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.958574627410181e-06,
+      "loss": 2.3298,
+      "step": 675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.958100506132127e-06,
+      "loss": 2.3422,
+      "step": 676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.957623698481589e-06,
+      "loss": 2.3896,
+      "step": 677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.957144204716908e-06,
+      "loss": 2.2982,
+      "step": 678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.956662025097883e-06,
+      "loss": 2.2686,
+      "step": 679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.956177159885765e-06,
+      "loss": 2.2914,
+      "step": 680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.95568960934326e-06,
+      "loss": 2.2846,
+      "step": 681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.95519937373453e-06,
+      "loss": 2.3958,
+      "step": 682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.95470645332519e-06,
+      "loss": 2.3494,
+      "step": 683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.95421084838232e-06,
+      "loss": 2.2013,
+      "step": 684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.953712559174435e-06,
+      "loss": 2.3175,
+      "step": 685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.953211585971525e-06,
+      "loss": 2.3382,
+      "step": 686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.952707929045018e-06,
+      "loss": 2.3724,
+      "step": 687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.952201588667805e-06,
+      "loss": 2.3169,
+      "step": 688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.95169256511423e-06,
+      "loss": 2.3628,
+      "step": 689
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.95118085866009e-06,
+      "loss": 2.2913,
+      "step": 690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.950666469582631e-06,
+      "loss": 2.2349,
+      "step": 691
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.950149398160562e-06,
+      "loss": 2.2255,
+      "step": 692
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.949629644674037e-06,
+      "loss": 2.3759,
+      "step": 693
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.949107209404664e-06,
+      "loss": 2.2567,
+      "step": 694
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.94858209263551e-06,
+      "loss": 2.3075,
+      "step": 695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.94805429465109e-06,
+      "loss": 2.4072,
+      "step": 696
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.947523815737374e-06,
+      "loss": 2.3166,
+      "step": 697
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.946990656181782e-06,
+      "loss": 2.3913,
+      "step": 698
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.946454816273186e-06,
+      "loss": 2.3294,
+      "step": 699
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.945916296301914e-06,
+      "loss": 2.3245,
+      "step": 700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.945375096559745e-06,
+      "loss": 2.3754,
+      "step": 701
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.944831217339906e-06,
+      "loss": 2.3678,
+      "step": 702
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.944284658937081e-06,
+      "loss": 2.3533,
+      "step": 703
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.943735421647404e-06,
+      "loss": 2.2929,
+      "step": 704
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.943183505768458e-06,
+      "loss": 2.3387,
+      "step": 705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.94262891159928e-06,
+      "loss": 2.2947,
+      "step": 706
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.94207163944036e-06,
+      "loss": 2.464,
+      "step": 707
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.941511689593634e-06,
+      "loss": 2.3153,
+      "step": 708
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.94094906236249e-06,
+      "loss": 2.3001,
+      "step": 709
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.94038375805177e-06,
+      "loss": 2.297,
+      "step": 710
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.939815776967761e-06,
+      "loss": 2.3978,
+      "step": 711
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.939245119418208e-06,
+      "loss": 2.3942,
+      "step": 712
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.938671785712298e-06,
+      "loss": 2.3648,
+      "step": 713
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.938095776160674e-06,
+      "loss": 2.2601,
+      "step": 714
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.937517091075426e-06,
+      "loss": 2.4053,
+      "step": 715
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.936935730770093e-06,
+      "loss": 2.3644,
+      "step": 716
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.936351695559666e-06,
+      "loss": 2.3887,
+      "step": 717
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.935764985760583e-06,
+      "loss": 2.3919,
+      "step": 718
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.935175601690734e-06,
+      "loss": 2.3805,
+      "step": 719
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.934583543669454e-06,
+      "loss": 2.3015,
+      "step": 720
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.933988812017529e-06,
+      "loss": 2.3203,
+      "step": 721
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.933391407057196e-06,
+      "loss": 2.2734,
+      "step": 722
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.932791329112134e-06,
+      "loss": 2.2958,
+      "step": 723
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.932188578507475e-06,
+      "loss": 2.3455,
+      "step": 724
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.931583155569802e-06,
+      "loss": 2.288,
+      "step": 725
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.930975060627137e-06,
+      "loss": 2.1854,
+      "step": 726
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.930364294008958e-06,
+      "loss": 2.3877,
+      "step": 727
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.929750856046187e-06,
+      "loss": 2.363,
+      "step": 728
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.929134747071193e-06,
+      "loss": 2.3073,
+      "step": 729
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.928515967417794e-06,
+      "loss": 2.4355,
+      "step": 730
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.927894517421252e-06,
+      "loss": 2.3079,
+      "step": 731
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.927270397418281e-06,
+      "loss": 2.323,
+      "step": 732
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.926643607747036e-06,
+      "loss": 2.3188,
+      "step": 733
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.92601414874712e-06,
+      "loss": 2.3749,
+      "step": 734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.925382020759587e-06,
+      "loss": 2.3334,
+      "step": 735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.924747224126932e-06,
+      "loss": 2.3462,
+      "step": 736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.924109759193095e-06,
+      "loss": 2.3092,
+      "step": 737
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.923469626303465e-06,
+      "loss": 2.4418,
+      "step": 738
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.922826825804877e-06,
+      "loss": 2.3663,
+      "step": 739
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.922181358045608e-06,
+      "loss": 2.4414,
+      "step": 740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.921533223375384e-06,
+      "loss": 2.4272,
+      "step": 741
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.920882422145372e-06,
+      "loss": 2.2187,
+      "step": 742
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.920228954708186e-06,
+      "loss": 2.2853,
+      "step": 743
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.919572821417887e-06,
+      "loss": 2.3968,
+      "step": 744
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.918914022629973e-06,
+      "loss": 2.2702,
+      "step": 745
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.918252558701397e-06,
+      "loss": 2.4135,
+      "step": 746
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.917588429990545e-06,
+      "loss": 2.345,
+      "step": 747
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.916921636857253e-06,
+      "loss": 2.3687,
+      "step": 748
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.9162521796628e-06,
+      "loss": 2.3237,
+      "step": 749
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.91558005876991e-06,
+      "loss": 2.3481,
+      "step": 750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.914905274542742e-06,
+      "loss": 2.3961,
+      "step": 751
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.91422782734691e-06,
+      "loss": 2.2943,
+      "step": 752
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.913547717549463e-06,
+      "loss": 2.3774,
+      "step": 753
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.912864945518894e-06,
+      "loss": 2.2964,
+      "step": 754
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.91217951162514e-06,
+      "loss": 2.3984,
+      "step": 755
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.911491416239578e-06,
+      "loss": 2.3301,
+      "step": 756
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.91080065973503e-06,
+      "loss": 2.3049,
+      "step": 757
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.910107242485757e-06,
+      "loss": 2.3447,
+      "step": 758
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.909411164867462e-06,
+      "loss": 2.3672,
+      "step": 759
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.908712427257291e-06,
+      "loss": 2.3283,
+      "step": 760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.908011030033832e-06,
+      "loss": 2.2809,
+      "step": 761
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.90730697357711e-06,
+      "loss": 2.2801,
+      "step": 762
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.906600258268597e-06,
+      "loss": 2.2642,
+      "step": 763
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.905890884491196e-06,
+      "loss": 2.4014,
+      "step": 764
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.90517885262926e-06,
+      "loss": 2.2448,
+      "step": 765
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.904464163068579e-06,
+      "loss": 2.3249,
+      "step": 766
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.90374681619638e-06,
+      "loss": 2.3419,
+      "step": 767
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.903026812401334e-06,
+      "loss": 2.2995,
+      "step": 768
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.902304152073547e-06,
+      "loss": 2.3082,
+      "step": 769
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.901578835604572e-06,
+      "loss": 2.4096,
+      "step": 770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.900850863387392e-06,
+      "loss": 2.3042,
+      "step": 771
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.900120235816436e-06,
+      "loss": 2.3604,
+      "step": 772
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.899386953287566e-06,
+      "loss": 2.3949,
+      "step": 773
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.898651016198087e-06,
+      "loss": 2.3702,
+      "step": 774
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.897912424946741e-06,
+      "loss": 2.1963,
+      "step": 775
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.897171179933706e-06,
+      "loss": 2.3165,
+      "step": 776
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.896427281560602e-06,
+      "loss": 2.4196,
+      "step": 777
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.895680730230483e-06,
+      "loss": 2.2851,
+      "step": 778
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.894931526347841e-06,
+      "loss": 2.3835,
+      "step": 779
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.894179670318607e-06,
+      "loss": 2.3046,
+      "step": 780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.893425162550148e-06,
+      "loss": 2.2061,
+      "step": 781
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.892668003451265e-06,
+      "loss": 2.3456,
+      "step": 782
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.891908193432199e-06,
+      "loss": 2.3072,
+      "step": 783
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.891145732904628e-06,
+      "loss": 2.3488,
+      "step": 784
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.890380622281663e-06,
+      "loss": 2.3561,
+      "step": 785
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.889612861977855e-06,
+      "loss": 2.3327,
+      "step": 786
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.888842452409183e-06,
+      "loss": 2.4257,
+      "step": 787
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.888069393993069e-06,
+      "loss": 2.3089,
+      "step": 788
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.887293687148369e-06,
+      "loss": 2.3042,
+      "step": 789
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.88651533229537e-06,
+      "loss": 2.2572,
+      "step": 790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.885734329855798e-06,
+      "loss": 2.3111,
+      "step": 791
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.884950680252811e-06,
+      "loss": 2.3124,
+      "step": 792
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.884164383911001e-06,
+      "loss": 2.28,
+      "step": 793
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.883375441256397e-06,
+      "loss": 2.356,
+      "step": 794
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.882583852716458e-06,
+      "loss": 2.3693,
+      "step": 795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.88178961872008e-06,
+      "loss": 2.301,
+      "step": 796
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.88099273969759e-06,
+      "loss": 2.3634,
+      "step": 797
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.88019321608075e-06,
+      "loss": 2.2633,
+      "step": 798
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.879391048302751e-06,
+      "loss": 2.302,
+      "step": 799
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.878586236798222e-06,
+      "loss": 2.34,
+      "step": 800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.87777878200322e-06,
+      "loss": 2.3744,
+      "step": 801
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.876968684355238e-06,
+      "loss": 2.4454,
+      "step": 802
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.876155944293198e-06,
+      "loss": 2.3517,
+      "step": 803
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.875340562257454e-06,
+      "loss": 2.258,
+      "step": 804
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.874522538689792e-06,
+      "loss": 2.392,
+      "step": 805
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.87370187403343e-06,
+      "loss": 2.4306,
+      "step": 806
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.872878568733018e-06,
+      "loss": 2.2242,
+      "step": 807
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.872052623234632e-06,
+      "loss": 2.3069,
+      "step": 808
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.871224037985782e-06,
+      "loss": 2.2744,
+      "step": 809
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.87039281343541e-06,
+      "loss": 2.2804,
+      "step": 810
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.869558950033883e-06,
+      "loss": 2.3656,
+      "step": 811
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.868722448233003e-06,
+      "loss": 2.3359,
+      "step": 812
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.867883308485997e-06,
+      "loss": 2.3166,
+      "step": 813
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.867041531247526e-06,
+      "loss": 2.3912,
+      "step": 814
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.866197116973675e-06,
+      "loss": 2.3347,
+      "step": 815
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.865350066121962e-06,
+      "loss": 2.3308,
+      "step": 816
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.86450037915133e-06,
+      "loss": 2.2719,
+      "step": 817
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.863648056522152e-06,
+      "loss": 2.358,
+      "step": 818
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.86279309869623e-06,
+      "loss": 2.3218,
+      "step": 819
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.861935506136794e-06,
+      "loss": 2.2996,
+      "step": 820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.861075279308499e-06,
+      "loss": 2.302,
+      "step": 821
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.860212418677427e-06,
+      "loss": 2.3561,
+      "step": 822
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.859346924711093e-06,
+      "loss": 2.3329,
+      "step": 823
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.85847879787843e-06,
+      "loss": 2.4233,
+      "step": 824
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.857608038649804e-06,
+      "loss": 2.4125,
+      "step": 825
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.856734647497006e-06,
+      "loss": 2.2609,
+      "step": 826
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.855858624893251e-06,
+      "loss": 2.3417,
+      "step": 827
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.854979971313183e-06,
+      "loss": 2.2611,
+      "step": 828
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.854098687232867e-06,
+      "loss": 2.3262,
+      "step": 829
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.853214773129796e-06,
+      "loss": 2.3183,
+      "step": 830
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.85232822948289e-06,
+      "loss": 2.3563,
+      "step": 831
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.85143905677249e-06,
+      "loss": 2.3671,
+      "step": 832
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.850547255480363e-06,
+      "loss": 2.2845,
+      "step": 833
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.849652826089702e-06,
+      "loss": 2.23,
+      "step": 834
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.848755769085118e-06,
+      "loss": 2.3121,
+      "step": 835
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.847856084952653e-06,
+      "loss": 2.3349,
+      "step": 836
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.846953774179771e-06,
+      "loss": 2.4567,
+      "step": 837
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.846048837255353e-06,
+      "loss": 2.4331,
+      "step": 838
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.845141274669709e-06,
+      "loss": 2.3052,
+      "step": 839
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.84423108691457e-06,
+      "loss": 2.4196,
+      "step": 840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.843318274483089e-06,
+      "loss": 2.3187,
+      "step": 841
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.842402837869843e-06,
+      "loss": 2.2247,
+      "step": 842
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.841484777570826e-06,
+      "loss": 2.4139,
+      "step": 843
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.840564094083461e-06,
+      "loss": 2.3334,
+      "step": 844
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.839640787906584e-06,
+      "loss": 2.2928,
+      "step": 845
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.838714859540459e-06,
+      "loss": 2.3358,
+      "step": 846
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.837786309486765e-06,
+      "loss": 2.2341,
+      "step": 847
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.836855138248604e-06,
+      "loss": 2.3585,
+      "step": 848
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.835921346330502e-06,
+      "loss": 2.3366,
+      "step": 849
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.834984934238399e-06,
+      "loss": 2.296,
+      "step": 850
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.834045902479657e-06,
+      "loss": 2.3214,
+      "step": 851
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.833104251563058e-06,
+      "loss": 2.3062,
+      "step": 852
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.8321599819988e-06,
+      "loss": 2.2161,
+      "step": 853
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.831213094298507e-06,
+      "loss": 2.3326,
+      "step": 854
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.83026358897521e-06,
+      "loss": 2.2953,
+      "step": 855
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.829311466543373e-06,
+      "loss": 2.3713,
+      "step": 856
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.828356727518864e-06,
+      "loss": 2.3441,
+      "step": 857
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.827399372418978e-06,
+      "loss": 2.2898,
+      "step": 858
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.826439401762424e-06,
+      "loss": 2.2921,
+      "step": 859
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.825476816069326e-06,
+      "loss": 2.3104,
+      "step": 860
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.82451161586123e-06,
+      "loss": 2.3188,
+      "step": 861
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.823543801661095e-06,
+      "loss": 2.294,
+      "step": 862
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.822573373993296e-06,
+      "loss": 2.3272,
+      "step": 863
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.821600333383626e-06,
+      "loss": 2.1933,
+      "step": 864
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.82062468035929e-06,
+      "loss": 2.3984,
+      "step": 865
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.819646415448918e-06,
+      "loss": 2.2543,
+      "step": 866
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.818665539182542e-06,
+      "loss": 2.3254,
+      "step": 867
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.817682052091618e-06,
+      "loss": 2.3672,
+      "step": 868
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.816695954709014e-06,
+      "loss": 2.3791,
+      "step": 869
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.815707247569013e-06,
+      "loss": 2.3698,
+      "step": 870
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.814715931207308e-06,
+      "loss": 2.2724,
+      "step": 871
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.813722006161013e-06,
+      "loss": 2.241,
+      "step": 872
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.81272547296865e-06,
+      "loss": 2.3857,
+      "step": 873
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.811726332170153e-06,
+      "loss": 2.3075,
+      "step": 874
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.810724584306873e-06,
+      "loss": 2.3993,
+      "step": 875
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.809720229921573e-06,
+      "loss": 2.2716,
+      "step": 876
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.808713269558425e-06,
+      "loss": 2.3154,
+      "step": 877
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.807703703763015e-06,
+      "loss": 2.3676,
+      "step": 878
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.806691533082346e-06,
+      "loss": 2.2391,
+      "step": 879
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.805676758064822e-06,
+      "loss": 2.3036,
+      "step": 880
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.804659379260262e-06,
+      "loss": 2.3049,
+      "step": 881
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.803639397219901e-06,
+      "loss": 2.3689,
+      "step": 882
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.802616812496377e-06,
+      "loss": 2.3252,
+      "step": 883
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.801591625643744e-06,
+      "loss": 2.2751,
+      "step": 884
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.800563837217465e-06,
+      "loss": 2.3555,
+      "step": 885
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.799533447774405e-06,
+      "loss": 2.3784,
+      "step": 886
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.798500457872851e-06,
+      "loss": 2.3479,
+      "step": 887
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.797464868072489e-06,
+      "loss": 2.369,
+      "step": 888
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.796426678934415e-06,
+      "loss": 2.3303,
+      "step": 889
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.795385891021139e-06,
+      "loss": 2.4153,
+      "step": 890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.794342504896574e-06,
+      "loss": 2.4077,
+      "step": 891
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.79329652112604e-06,
+      "loss": 2.3057,
+      "step": 892
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.792247940276272e-06,
+      "loss": 2.3546,
+      "step": 893
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.7911967629154e-06,
+      "loss": 2.2522,
+      "step": 894
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.790142989612971e-06,
+      "loss": 2.2483,
+      "step": 895
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.789086620939936e-06,
+      "loss": 2.4013,
+      "step": 896
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.78802765746865e-06,
+      "loss": 2.2875,
+      "step": 897
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.786966099772875e-06,
+      "loss": 2.356,
+      "step": 898
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.785901948427778e-06,
+      "loss": 2.337,
+      "step": 899
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.784835204009932e-06,
+      "loss": 2.3054,
+      "step": 900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.783765867097317e-06,
+      "loss": 2.3386,
+      "step": 901
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.782693938269313e-06,
+      "loss": 2.3238,
+      "step": 902
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.781619418106711e-06,
+      "loss": 2.3513,
+      "step": 903
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.780542307191698e-06,
+      "loss": 2.291,
+      "step": 904
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.77946260610787e-06,
+      "loss": 2.2621,
+      "step": 905
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.778380315440224e-06,
+      "loss": 2.3104,
+      "step": 906
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.777295435775165e-06,
+      "loss": 2.3081,
+      "step": 907
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.77620796770049e-06,
+      "loss": 2.2734,
+      "step": 908
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.775117911805413e-06,
+      "loss": 2.3296,
+      "step": 909
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.774025268680538e-06,
+      "loss": 2.3308,
+      "step": 910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.772930038917875e-06,
+      "loss": 2.4036,
+      "step": 911
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.77183222311084e-06,
+      "loss": 2.2547,
+      "step": 912
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.77073182185424e-06,
+      "loss": 2.2972,
+      "step": 913
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.769628835744292e-06,
+      "loss": 2.3421,
+      "step": 914
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.768523265378612e-06,
+      "loss": 2.2868,
+      "step": 915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.76741511135621e-06,
+      "loss": 2.3079,
+      "step": 916
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.766304374277504e-06,
+      "loss": 2.2583,
+      "step": 917
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.765191054744305e-06,
+      "loss": 2.328,
+      "step": 918
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.764075153359828e-06,
+      "loss": 2.3792,
+      "step": 919
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.762956670728685e-06,
+      "loss": 2.439,
+      "step": 920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.761835607456886e-06,
+      "loss": 2.3658,
+      "step": 921
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.760711964151842e-06,
+      "loss": 2.329,
+      "step": 922
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.759585741422353e-06,
+      "loss": 2.3344,
+      "step": 923
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.75845693987863e-06,
+      "loss": 2.3073,
+      "step": 924
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.757325560132272e-06,
+      "loss": 2.3415,
+      "step": 925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.756191602796276e-06,
+      "loss": 2.2839,
+      "step": 926
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.755055068485039e-06,
+      "loss": 2.4012,
+      "step": 927
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.753915957814353e-06,
+      "loss": 2.3357,
+      "step": 928
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.752774271401403e-06,
+      "loss": 2.2741,
+      "step": 929
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.75163000986477e-06,
+      "loss": 2.294,
+      "step": 930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.750483173824435e-06,
+      "loss": 2.3675,
+      "step": 931
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.74933376390177e-06,
+      "loss": 2.3465,
+      "step": 932
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.748181780719543e-06,
+      "loss": 2.3588,
+      "step": 933
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.747027224901914e-06,
+      "loss": 2.3207,
+      "step": 934
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.745870097074439e-06,
+      "loss": 2.3019,
+      "step": 935
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.744710397864068e-06,
+      "loss": 2.3587,
+      "step": 936
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.743548127899143e-06,
+      "loss": 2.2553,
+      "step": 937
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.742383287809399e-06,
+      "loss": 2.3449,
+      "step": 938
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.74121587822596e-06,
+      "loss": 2.3145,
+      "step": 939
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.740045899781353e-06,
+      "loss": 2.325,
+      "step": 940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.738873353109484e-06,
+      "loss": 2.2284,
+      "step": 941
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.73769823884566e-06,
+      "loss": 2.4078,
+      "step": 942
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.736520557626572e-06,
+      "loss": 2.3109,
+      "step": 943
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.735340310090306e-06,
+      "loss": 2.3206,
+      "step": 944
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.73415749687634e-06,
+      "loss": 2.398,
+      "step": 945
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.732972118625536e-06,
+      "loss": 2.3795,
+      "step": 946
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.731784175980155e-06,
+      "loss": 2.2815,
+      "step": 947
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.730593669583836e-06,
+      "loss": 2.3724,
+      "step": 948
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.729400600081615e-06,
+      "loss": 2.3538,
+      "step": 949
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.728204968119916e-06,
+      "loss": 2.274,
+      "step": 950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.727006774346551e-06,
+      "loss": 2.3272,
+      "step": 951
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.725806019410718e-06,
+      "loss": 2.3682,
+      "step": 952
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.724602703963001e-06,
+      "loss": 2.3919,
+      "step": 953
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.723396828655378e-06,
+      "loss": 2.3035,
+      "step": 954
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.72218839414121e-06,
+      "loss": 2.2889,
+      "step": 955
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.720977401075243e-06,
+      "loss": 2.2769,
+      "step": 956
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.719763850113611e-06,
+      "loss": 2.3777,
+      "step": 957
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.718547741913835e-06,
+      "loss": 2.3164,
+      "step": 958
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.71732907713482e-06,
+      "loss": 2.428,
+      "step": 959
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.716107856436855e-06,
+      "loss": 2.2796,
+      "step": 960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.71488408048162e-06,
+      "loss": 2.358,
+      "step": 961
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.713657749932172e-06,
+      "loss": 2.3065,
+      "step": 962
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.712428865452955e-06,
+      "loss": 2.2604,
+      "step": 963
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.711197427709796e-06,
+      "loss": 2.3231,
+      "step": 964
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.70996343736991e-06,
+      "loss": 2.2681,
+      "step": 965
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.708726895101886e-06,
+      "loss": 2.3443,
+      "step": 966
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.707487801575705e-06,
+      "loss": 2.2741,
+      "step": 967
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.706246157462726e-06,
+      "loss": 2.3052,
+      "step": 968
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.705001963435688e-06,
+      "loss": 2.3036,
+      "step": 969
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.703755220168715e-06,
+      "loss": 2.3585,
+      "step": 970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.702505928337312e-06,
+      "loss": 2.3155,
+      "step": 971
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.701254088618363e-06,
+      "loss": 2.3003,
+      "step": 972
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.699999701690133e-06,
+      "loss": 2.4173,
+      "step": 973
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.698742768232268e-06,
+      "loss": 2.2791,
+      "step": 974
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.697483288925791e-06,
+      "loss": 2.2699,
+      "step": 975
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.69622126445311e-06,
+      "loss": 2.311,
+      "step": 976
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.694956695498008e-06,
+      "loss": 2.3319,
+      "step": 977
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.693689582745645e-06,
+      "loss": 2.3124,
+      "step": 978
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.692419926882563e-06,
+      "loss": 2.3107,
+      "step": 979
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.691147728596682e-06,
+      "loss": 2.296,
+      "step": 980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.689872988577293e-06,
+      "loss": 2.257,
+      "step": 981
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.688595707515076e-06,
+      "loss": 2.3166,
+      "step": 982
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.687315886102075e-06,
+      "loss": 2.2938,
+      "step": 983
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.68603352503172e-06,
+      "loss": 2.4413,
+      "step": 984
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.68474862499881e-06,
+      "loss": 2.3555,
+      "step": 985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.683461186699526e-06,
+      "loss": 2.333,
+      "step": 986
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.682171210831418e-06,
+      "loss": 2.4089,
+      "step": 987
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.680878698093417e-06,
+      "loss": 2.3658,
+      "step": 988
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.679583649185824e-06,
+      "loss": 2.2351,
+      "step": 989
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.678286064810316e-06,
+      "loss": 2.2853,
+      "step": 990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.676985945669944e-06,
+      "loss": 2.3471,
+      "step": 991
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.675683292469131e-06,
+      "loss": 2.2639,
+      "step": 992
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.674378105913676e-06,
+      "loss": 2.3811,
+      "step": 993
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.673070386710746e-06,
+      "loss": 2.3365,
+      "step": 994
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.671760135568882e-06,
+      "loss": 2.3086,
+      "step": 995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.670447353198e-06,
+      "loss": 2.4175,
+      "step": 996
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.669132040309386e-06,
+      "loss": 2.2996,
+      "step": 997
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.667814197615691e-06,
+      "loss": 2.3031,
+      "step": 998
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.666493825830947e-06,
+      "loss": 2.3775,
+      "step": 999
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.665170925670549e-06,
+      "loss": 2.2996,
+      "step": 1000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.663845497851262e-06,
+      "loss": 2.3046,
+      "step": 1001
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.662517543091225e-06,
+      "loss": 2.3994,
+      "step": 1002
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.661187062109944e-06,
+      "loss": 2.2482,
+      "step": 1003
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.65985405562829e-06,
+      "loss": 2.2987,
+      "step": 1004
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.658518524368509e-06,
+      "loss": 2.2884,
+      "step": 1005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.657180469054213e-06,
+      "loss": 2.3809,
+      "step": 1006
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.655839890410375e-06,
+      "loss": 2.3467,
+      "step": 1007
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.654496789163344e-06,
+      "loss": 2.3809,
+      "step": 1008
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.653151166040833e-06,
+      "loss": 2.2587,
+      "step": 1009
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.651803021771918e-06,
+      "loss": 2.2177,
+      "step": 1010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.650452357087043e-06,
+      "loss": 2.3117,
+      "step": 1011
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.649099172718022e-06,
+      "loss": 2.3746,
+      "step": 1012
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.647743469398028e-06,
+      "loss": 2.243,
+      "step": 1013
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.6463852478616e-06,
+      "loss": 2.2649,
+      "step": 1014
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.645024508844646e-06,
+      "loss": 2.3783,
+      "step": 1015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.643661253084431e-06,
+      "loss": 2.4245,
+      "step": 1016
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.64229548131959e-06,
+      "loss": 2.4282,
+      "step": 1017
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.640927194290117e-06,
+      "loss": 2.2407,
+      "step": 1018
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.63955639273737e-06,
+      "loss": 2.3082,
+      "step": 1019
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.63818307740407e-06,
+      "loss": 2.3836,
+      "step": 1020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.6368072490343e-06,
+      "loss": 2.3989,
+      "step": 1021
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.635428908373504e-06,
+      "loss": 2.3841,
+      "step": 1022
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.634048056168489e-06,
+      "loss": 2.2486,
+      "step": 1023
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.632664693167417e-06,
+      "loss": 2.2052,
+      "step": 1024
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.631278820119819e-06,
+      "loss": 2.3124,
+      "step": 1025
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.62989043777658e-06,
+      "loss": 2.3287,
+      "step": 1026
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.628499546889946e-06,
+      "loss": 2.3631,
+      "step": 1027
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.627106148213521e-06,
+      "loss": 2.3459,
+      "step": 1028
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.625710242502272e-06,
+      "loss": 2.3831,
+      "step": 1029
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.62431183051252e-06,
+      "loss": 2.2728,
+      "step": 1030
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.622910913001946e-06,
+      "loss": 2.4274,
+      "step": 1031
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.621507490729585e-06,
+      "loss": 2.2688,
+      "step": 1032
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.620101564455836e-06,
+      "loss": 2.3668,
+      "step": 1033
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.618693134942448e-06,
+      "loss": 2.3075,
+      "step": 1034
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.61728220295253e-06,
+      "loss": 2.3481,
+      "step": 1035
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.615868769250547e-06,
+      "loss": 2.3431,
+      "step": 1036
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.614452834602318e-06,
+      "loss": 2.2966,
+      "step": 1037
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.613034399775015e-06,
+      "loss": 2.2948,
+      "step": 1038
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.61161346553717e-06,
+      "loss": 2.3404,
+      "step": 1039
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.610190032658664e-06,
+      "loss": 2.2778,
+      "step": 1040
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.608764101910737e-06,
+      "loss": 2.3068,
+      "step": 1041
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.607335674065976e-06,
+      "loss": 2.3719,
+      "step": 1042
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.605904749898326e-06,
+      "loss": 2.3411,
+      "step": 1043
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.604471330183082e-06,
+      "loss": 2.3783,
+      "step": 1044
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.603035415696896e-06,
+      "loss": 2.3377,
+      "step": 1045
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.601597007217762e-06,
+      "loss": 2.3524,
+      "step": 1046
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.600156105525036e-06,
+      "loss": 2.3163,
+      "step": 1047
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.598712711399416e-06,
+      "loss": 2.2502,
+      "step": 1048
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.597266825622957e-06,
+      "loss": 2.3616,
+      "step": 1049
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.595818448979061e-06,
+      "loss": 2.268,
+      "step": 1050
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.59436758225248e-06,
+      "loss": 2.302,
+      "step": 1051
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.592914226229315e-06,
+      "loss": 2.2977,
+      "step": 1052
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.591458381697013e-06,
+      "loss": 2.2923,
+      "step": 1053
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.590000049444377e-06,
+      "loss": 2.3398,
+      "step": 1054
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.588539230261551e-06,
+      "loss": 2.446,
+      "step": 1055
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.587075924940027e-06,
+      "loss": 2.2905,
+      "step": 1056
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.58561013427265e-06,
+      "loss": 2.3147,
+      "step": 1057
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.584141859053603e-06,
+      "loss": 2.2852,
+      "step": 1058
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.582671100078418e-06,
+      "loss": 2.3202,
+      "step": 1059
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.581197858143977e-06,
+      "loss": 2.248,
+      "step": 1060
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.579722134048506e-06,
+      "loss": 2.2558,
+      "step": 1061
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.578243928591571e-06,
+      "loss": 2.3054,
+      "step": 1062
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.576763242574085e-06,
+      "loss": 2.2735,
+      "step": 1063
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.57528007679831e-06,
+      "loss": 2.4088,
+      "step": 1064
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.573794432067844e-06,
+      "loss": 2.3146,
+      "step": 1065
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.572306309187632e-06,
+      "loss": 2.3069,
+      "step": 1066
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.57081570896396e-06,
+      "loss": 2.3035,
+      "step": 1067
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.569322632204458e-06,
+      "loss": 2.3037,
+      "step": 1068
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.5678270797181e-06,
+      "loss": 2.3012,
+      "step": 1069
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.566329052315195e-06,
+      "loss": 2.2956,
+      "step": 1070
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.564828550807398e-06,
+      "loss": 2.2537,
+      "step": 1071
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.563325576007702e-06,
+      "loss": 2.3017,
+      "step": 1072
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.561820128730441e-06,
+      "loss": 2.3756,
+      "step": 1073
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.560312209791291e-06,
+      "loss": 2.236,
+      "step": 1074
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.558801820007262e-06,
+      "loss": 2.2154,
+      "step": 1075
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.557288960196708e-06,
+      "loss": 2.2787,
+      "step": 1076
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.555773631179317e-06,
+      "loss": 2.4056,
+      "step": 1077
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.554255833776119e-06,
+      "loss": 2.349,
+      "step": 1078
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.552735568809477e-06,
+      "loss": 2.4015,
+      "step": 1079
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.551212837103092e-06,
+      "loss": 2.2478,
+      "step": 1080
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.549687639482007e-06,
+      "loss": 2.2563,
+      "step": 1081
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.548159976772593e-06,
+      "loss": 2.3361,
+      "step": 1082
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.546629849802562e-06,
+      "loss": 2.2702,
+      "step": 1083
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.545097259400958e-06,
+      "loss": 2.353,
+      "step": 1084
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.543562206398164e-06,
+      "loss": 2.4004,
+      "step": 1085
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.542024691625891e-06,
+      "loss": 2.2875,
+      "step": 1086
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.540484715917191e-06,
+      "loss": 2.3151,
+      "step": 1087
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.538942280106443e-06,
+      "loss": 2.3667,
+      "step": 1088
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.537397385029363e-06,
+      "loss": 2.2732,
+      "step": 1089
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.535850031522998e-06,
+      "loss": 2.3448,
+      "step": 1090
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.534300220425726e-06,
+      "loss": 2.283,
+      "step": 1091
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.532747952577259e-06,
+      "loss": 2.3489,
+      "step": 1092
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.531193228818638e-06,
+      "loss": 2.3481,
+      "step": 1093
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.529636049992235e-06,
+      "loss": 2.3285,
+      "step": 1094
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.528076416941754e-06,
+      "loss": 2.3308,
+      "step": 1095
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.526514330512225e-06,
+      "loss": 2.3315,
+      "step": 1096
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.524949791550014e-06,
+      "loss": 2.3021,
+      "step": 1097
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.523382800902805e-06,
+      "loss": 2.3102,
+      "step": 1098
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.521813359419623e-06,
+      "loss": 2.3619,
+      "step": 1099
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.520241467950812e-06,
+      "loss": 2.2957,
+      "step": 1100
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.518667127348045e-06,
+      "loss": 2.3305,
+      "step": 1101
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.517090338464325e-06,
+      "loss": 2.2089,
+      "step": 1102
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.51551110215398e-06,
+      "loss": 2.3419,
+      "step": 1103
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.513929419272662e-06,
+      "loss": 2.2009,
+      "step": 1104
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.51234529067735e-06,
+      "loss": 2.3071,
+      "step": 1105
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.510758717226351e-06,
+      "loss": 2.2753,
+      "step": 1106
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.509169699779292e-06,
+      "loss": 2.3116,
+      "step": 1107
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.507578239197126e-06,
+      "loss": 2.3429,
+      "step": 1108
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.505984336342131e-06,
+      "loss": 2.2965,
+      "step": 1109
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.504387992077906e-06,
+      "loss": 2.2252,
+      "step": 1110
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.502789207269375e-06,
+      "loss": 2.3405,
+      "step": 1111
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.501187982782785e-06,
+      "loss": 2.3335,
+      "step": 1112
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.499584319485702e-06,
+      "loss": 2.2503,
+      "step": 1113
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.497978218247013e-06,
+      "loss": 2.2795,
+      "step": 1114
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.496369679936928e-06,
+      "loss": 2.3413,
+      "step": 1115
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.494758705426978e-06,
+      "loss": 2.2703,
+      "step": 1116
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.493145295590011e-06,
+      "loss": 2.336,
+      "step": 1117
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.4915294513002e-06,
+      "loss": 2.3579,
+      "step": 1118
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.48991117343303e-06,
+      "loss": 2.3754,
+      "step": 1119
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.48829046286531e-06,
+      "loss": 2.3401,
+      "step": 1120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.486667320475163e-06,
+      "loss": 2.2823,
+      "step": 1121
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.485041747142034e-06,
+      "loss": 2.2471,
+      "step": 1122
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.483413743746683e-06,
+      "loss": 2.3648,
+      "step": 1123
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.481783311171183e-06,
+      "loss": 2.3219,
+      "step": 1124
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.48015045029893e-06,
+      "loss": 2.2575,
+      "step": 1125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.47851516201463e-06,
+      "loss": 2.3542,
+      "step": 1126
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.476877447204308e-06,
+      "loss": 2.2441,
+      "step": 1127
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.475237306755303e-06,
+      "loss": 2.2681,
+      "step": 1128
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.473594741556265e-06,
+      "loss": 2.3889,
+      "step": 1129
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.471949752497161e-06,
+      "loss": 2.3018,
+      "step": 1130
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.470302340469271e-06,
+      "loss": 2.2496,
+      "step": 1131
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.468652506365186e-06,
+      "loss": 2.4439,
+      "step": 1132
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.467000251078813e-06,
+      "loss": 2.3726,
+      "step": 1133
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.465345575505367e-06,
+      "loss": 2.3745,
+      "step": 1134
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.463688480541377e-06,
+      "loss": 2.3268,
+      "step": 1135
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.46202896708468e-06,
+      "loss": 2.2799,
+      "step": 1136
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.460367036034426e-06,
+      "loss": 2.3129,
+      "step": 1137
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.458702688291072e-06,
+      "loss": 2.3309,
+      "step": 1138
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.45703592475639e-06,
+      "loss": 2.3804,
+      "step": 1139
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.455366746333454e-06,
+      "loss": 2.3348,
+      "step": 1140
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.453695153926653e-06,
+      "loss": 2.201,
+      "step": 1141
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.452021148441676e-06,
+      "loss": 2.2394,
+      "step": 1142
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.450344730785528e-06,
+      "loss": 2.3001,
+      "step": 1143
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.448665901866514e-06,
+      "loss": 2.2607,
+      "step": 1144
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.44698466259425e-06,
+      "loss": 2.3553,
+      "step": 1145
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.445301013879658e-06,
+      "loss": 2.3857,
+      "step": 1146
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.44361495663496e-06,
+      "loss": 2.2438,
+      "step": 1147
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.441926491773692e-06,
+      "loss": 2.2722,
+      "step": 1148
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.440235620210682e-06,
+      "loss": 2.313,
+      "step": 1149
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.438542342862077e-06,
+      "loss": 2.4031,
+      "step": 1150
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.436846660645316e-06,
+      "loss": 2.3507,
+      "step": 1151
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.435148574479144e-06,
+      "loss": 2.2429,
+      "step": 1152
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.433448085283614e-06,
+      "loss": 2.3524,
+      "step": 1153
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.43174519398007e-06,
+      "loss": 2.3422,
+      "step": 1154
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.430039901491167e-06,
+      "loss": 2.2758,
+      "step": 1155
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.428332208740858e-06,
+      "loss": 2.2672,
+      "step": 1156
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.426622116654396e-06,
+      "loss": 2.315,
+      "step": 1157
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.424909626158333e-06,
+      "loss": 2.3702,
+      "step": 1158
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.423194738180523e-06,
+      "loss": 2.3185,
+      "step": 1159
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.421477453650118e-06,
+      "loss": 2.3753,
+      "step": 1160
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.419757773497568e-06,
+      "loss": 2.3671,
+      "step": 1161
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.418035698654621e-06,
+      "loss": 2.4144,
+      "step": 1162
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.416311230054325e-06,
+      "loss": 2.3721,
+      "step": 1163
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.41458436863102e-06,
+      "loss": 2.2927,
+      "step": 1164
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.412855115320344e-06,
+      "loss": 2.2721,
+      "step": 1165
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.411123471059234e-06,
+      "loss": 2.3926,
+      "step": 1166
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.409389436785919e-06,
+      "loss": 2.2929,
+      "step": 1167
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.407653013439927e-06,
+      "loss": 2.3242,
+      "step": 1168
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.405914201962078e-06,
+      "loss": 2.3575,
+      "step": 1169
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.404173003294486e-06,
+      "loss": 2.2617,
+      "step": 1170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.402429418380555e-06,
+      "loss": 2.3435,
+      "step": 1171
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.400683448164988e-06,
+      "loss": 2.3014,
+      "step": 1172
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.398935093593776e-06,
+      "loss": 2.3962,
+      "step": 1173
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.397184355614206e-06,
+      "loss": 2.3175,
+      "step": 1174
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.39543123517485e-06,
+      "loss": 2.3022,
+      "step": 1175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.393675733225578e-06,
+      "loss": 2.2721,
+      "step": 1176
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.391917850717546e-06,
+      "loss": 2.3346,
+      "step": 1177
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.390157588603201e-06,
+      "loss": 2.2921,
+      "step": 1178
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.388394947836278e-06,
+      "loss": 2.3618,
+      "step": 1179
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.386629929371804e-06,
+      "loss": 2.1966,
+      "step": 1180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.38486253416609e-06,
+      "loss": 2.309,
+      "step": 1181
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.38309276317674e-06,
+      "loss": 2.2478,
+      "step": 1182
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.381320617362638e-06,
+      "loss": 2.2622,
+      "step": 1183
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.379546097683963e-06,
+      "loss": 2.313,
+      "step": 1184
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.377769205102171e-06,
+      "loss": 2.2858,
+      "step": 1185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.375989940580014e-06,
+      "loss": 2.3971,
+      "step": 1186
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.374208305081524e-06,
+      "loss": 2.3369,
+      "step": 1187
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.372424299572014e-06,
+      "loss": 2.2525,
+      "step": 1188
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.370637925018086e-06,
+      "loss": 2.327,
+      "step": 1189
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.368849182387625e-06,
+      "loss": 2.3625,
+      "step": 1190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.367058072649796e-06,
+      "loss": 2.3142,
+      "step": 1191
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.365264596775052e-06,
+      "loss": 2.3143,
+      "step": 1192
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.363468755735123e-06,
+      "loss": 2.4135,
+      "step": 1193
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.361670550503025e-06,
+      "loss": 2.3697,
+      "step": 1194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.359869982053049e-06,
+      "loss": 2.2994,
+      "step": 1195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.35806705136077e-06,
+      "loss": 2.3508,
+      "step": 1196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.356261759403044e-06,
+      "loss": 2.343,
+      "step": 1197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.354454107158003e-06,
+      "loss": 2.3233,
+      "step": 1198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.352644095605064e-06,
+      "loss": 2.2226,
+      "step": 1199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.350831725724915e-06,
+      "loss": 2.3018,
+      "step": 1200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.349016998499525e-06,
+      "loss": 2.3738,
+      "step": 1201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.347199914912141e-06,
+      "loss": 2.3234,
+      "step": 1202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.345380475947285e-06,
+      "loss": 2.3287,
+      "step": 1203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.343558682590757e-06,
+      "loss": 2.3231,
+      "step": 1204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.34173453582963e-06,
+      "loss": 2.3846,
+      "step": 1205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.339908036652256e-06,
+      "loss": 2.3311,
+      "step": 1206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.338079186048258e-06,
+      "loss": 2.3682,
+      "step": 1207
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.336247985008533e-06,
+      "loss": 2.3093,
+      "step": 1208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.334414434525257e-06,
+      "loss": 2.3139,
+      "step": 1209
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.332578535591873e-06,
+      "loss": 2.2892,
+      "step": 1210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.330740289203098e-06,
+      "loss": 2.2941,
+      "step": 1211
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.328899696354919e-06,
+      "loss": 2.2929,
+      "step": 1212
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.3270567580446e-06,
+      "loss": 2.3272,
+      "step": 1213
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.32521147527067e-06,
+      "loss": 2.3272,
+      "step": 1214
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.323363849032933e-06,
+      "loss": 2.3322,
+      "step": 1215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.321513880332458e-06,
+      "loss": 2.315,
+      "step": 1216
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.319661570171585e-06,
+      "loss": 2.3287,
+      "step": 1217
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.317806919553927e-06,
+      "loss": 2.2461,
+      "step": 1218
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.315949929484358e-06,
+      "loss": 2.2518,
+      "step": 1219
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.314090600969023e-06,
+      "loss": 2.2386,
+      "step": 1220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.312228935015336e-06,
+      "loss": 2.3358,
+      "step": 1221
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.310364932631972e-06,
+      "loss": 2.3503,
+      "step": 1222
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.308498594828879e-06,
+      "loss": 2.3112,
+      "step": 1223
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.306629922617262e-06,
+      "loss": 2.4245,
+      "step": 1224
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.304758917009598e-06,
+      "loss": 2.242,
+      "step": 1225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.302885579019626e-06,
+      "loss": 2.229,
+      "step": 1226
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.301009909662349e-06,
+      "loss": 2.2696,
+      "step": 1227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.29913190995403e-06,
+      "loss": 2.2703,
+      "step": 1228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.2972515809122e-06,
+      "loss": 2.2481,
+      "step": 1229
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.295368923555645e-06,
+      "loss": 2.4422,
+      "step": 1230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.29348393890442e-06,
+      "loss": 2.2738,
+      "step": 1231
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.291596627979836e-06,
+      "loss": 2.2678,
+      "step": 1232
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.289706991804468e-06,
+      "loss": 2.2917,
+      "step": 1233
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.28781503140215e-06,
+      "loss": 2.3833,
+      "step": 1234
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.285920747797969e-06,
+      "loss": 2.3588,
+      "step": 1235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.28402414201828e-06,
+      "loss": 2.3361,
+      "step": 1236
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.282125215090694e-06,
+      "loss": 2.3254,
+      "step": 1237
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.280223968044073e-06,
+      "loss": 2.3393,
+      "step": 1238
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.278320401908543e-06,
+      "loss": 2.2792,
+      "step": 1239
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.276414517715484e-06,
+      "loss": 2.2894,
+      "step": 1240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.274506316497534e-06,
+      "loss": 2.2444,
+      "step": 1241
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.272595799288585e-06,
+      "loss": 2.2549,
+      "step": 1242
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.27068296712378e-06,
+      "loss": 2.3458,
+      "step": 1243
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.268767821039522e-06,
+      "loss": 2.3912,
+      "step": 1244
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.266850362073465e-06,
+      "loss": 2.3201,
+      "step": 1245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.264930591264519e-06,
+      "loss": 2.2698,
+      "step": 1246
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.263008509652843e-06,
+      "loss": 2.4124,
+      "step": 1247
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.261084118279846e-06,
+      "loss": 2.249,
+      "step": 1248
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.259157418188198e-06,
+      "loss": 2.344,
+      "step": 1249
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.25722841042181e-06,
+      "loss": 2.2351,
+      "step": 1250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.255297096025848e-06,
+      "loss": 2.2392,
+      "step": 1251
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.253363476046727e-06,
+      "loss": 2.3493,
+      "step": 1252
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.251427551532108e-06,
+      "loss": 2.2538,
+      "step": 1253
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.249489323530907e-06,
+      "loss": 2.3775,
+      "step": 1254
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.247548793093287e-06,
+      "loss": 2.2748,
+      "step": 1255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.24560596127065e-06,
+      "loss": 2.3074,
+      "step": 1256
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.243660829115654e-06,
+      "loss": 2.3148,
+      "step": 1257
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.241713397682203e-06,
+      "loss": 2.3482,
+      "step": 1258
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.23976366802544e-06,
+      "loss": 2.3347,
+      "step": 1259
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.23781164120176e-06,
+      "loss": 2.3004,
+      "step": 1260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.235857318268801e-06,
+      "loss": 2.2993,
+      "step": 1261
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.233900700285441e-06,
+      "loss": 2.2536,
+      "step": 1262
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.231941788311806e-06,
+      "loss": 2.2856,
+      "step": 1263
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.229980583409265e-06,
+      "loss": 2.3464,
+      "step": 1264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.228017086640429e-06,
+      "loss": 2.3245,
+      "step": 1265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.226051299069145e-06,
+      "loss": 2.2527,
+      "step": 1266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.224083221760509e-06,
+      "loss": 2.326,
+      "step": 1267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.222112855780857e-06,
+      "loss": 2.2434,
+      "step": 1268
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.220140202197757e-06,
+      "loss": 2.3063,
+      "step": 1269
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.218165262080024e-06,
+      "loss": 2.2576,
+      "step": 1270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.21618803649771e-06,
+      "loss": 2.3128,
+      "step": 1271
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.214208526522108e-06,
+      "loss": 2.3134,
+      "step": 1272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.212226733225744e-06,
+      "loss": 2.2248,
+      "step": 1273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.21024265768238e-06,
+      "loss": 2.3631,
+      "step": 1274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.208256300967023e-06,
+      "loss": 2.2334,
+      "step": 1275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.206267664155906e-06,
+      "loss": 2.2623,
+      "step": 1276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.204276748326504e-06,
+      "loss": 2.3627,
+      "step": 1277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.202283554557526e-06,
+      "loss": 2.3705,
+      "step": 1278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.200288083928913e-06,
+      "loss": 2.2841,
+      "step": 1279
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.198290337521837e-06,
+      "loss": 2.3023,
+      "step": 1280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.196290316418712e-06,
+      "loss": 2.2311,
+      "step": 1281
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.194288021703178e-06,
+      "loss": 2.2697,
+      "step": 1282
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.192283454460105e-06,
+      "loss": 2.2628,
+      "step": 1283
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.1902766157756e-06,
+      "loss": 2.3442,
+      "step": 1284
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.188267506736996e-06,
+      "loss": 2.2463,
+      "step": 1285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.186256128432861e-06,
+      "loss": 2.331,
+      "step": 1286
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.184242481952988e-06,
+      "loss": 2.2494,
+      "step": 1287
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.1822265683884e-06,
+      "loss": 2.4334,
+      "step": 1288
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.180208388831351e-06,
+      "loss": 2.297,
+      "step": 1289
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.178187944375318e-06,
+      "loss": 2.3365,
+      "step": 1290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.17616523611501e-06,
+      "loss": 2.3137,
+      "step": 1291
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.174140265146356e-06,
+      "loss": 2.2468,
+      "step": 1292
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.172113032566522e-06,
+      "loss": 2.2794,
+      "step": 1293
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.170083539473889e-06,
+      "loss": 2.3011,
+      "step": 1294
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.168051786968064e-06,
+      "loss": 2.1853,
+      "step": 1295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.166017776149888e-06,
+      "loss": 2.284,
+      "step": 1296
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.163981508121411e-06,
+      "loss": 2.2322,
+      "step": 1297
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.161942983985916e-06,
+      "loss": 2.3012,
+      "step": 1298
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.159902204847907e-06,
+      "loss": 2.3314,
+      "step": 1299
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.157859171813107e-06,
+      "loss": 2.3287,
+      "step": 1300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.155813885988462e-06,
+      "loss": 2.2674,
+      "step": 1301
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.153766348482139e-06,
+      "loss": 2.2461,
+      "step": 1302
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.151716560403521e-06,
+      "loss": 2.3517,
+      "step": 1303
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.149664522863218e-06,
+      "loss": 2.3017,
+      "step": 1304
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.14761023697305e-06,
+      "loss": 2.3108,
+      "step": 1305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.145553703846064e-06,
+      "loss": 2.382,
+      "step": 1306
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.143494924596517e-06,
+      "loss": 2.3686,
+      "step": 1307
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.141433900339887e-06,
+      "loss": 2.2641,
+      "step": 1308
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.139370632192868e-06,
+      "loss": 2.331,
+      "step": 1309
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.13730512127337e-06,
+      "loss": 2.2941,
+      "step": 1310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.135237368700518e-06,
+      "loss": 2.2874,
+      "step": 1311
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.133167375594647e-06,
+      "loss": 2.3309,
+      "step": 1312
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.131095143077313e-06,
+      "loss": 2.4294,
+      "step": 1313
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.129020672271283e-06,
+      "loss": 2.3562,
+      "step": 1314
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.126943964300535e-06,
+      "loss": 2.4134,
+      "step": 1315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.124865020290261e-06,
+      "loss": 2.1851,
+      "step": 1316
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.122783841366863e-06,
+      "loss": 2.2367,
+      "step": 1317
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.120700428657954e-06,
+      "loss": 2.3768,
+      "step": 1318
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.118614783292359e-06,
+      "loss": 2.3058,
+      "step": 1319
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.11652690640011e-06,
+      "loss": 2.2673,
+      "step": 1320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.114436799112453e-06,
+      "loss": 2.3617,
+      "step": 1321
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.112344462561836e-06,
+      "loss": 2.3665,
+      "step": 1322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.11024989788192e-06,
+      "loss": 2.3348,
+      "step": 1323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.10815310620757e-06,
+      "loss": 2.3648,
+      "step": 1324
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.106054088674861e-06,
+      "loss": 2.361,
+      "step": 1325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.103952846421068e-06,
+      "loss": 2.235,
+      "step": 1326
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.101849380584677e-06,
+      "loss": 2.2711,
+      "step": 1327
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.099743692305379e-06,
+      "loss": 2.3269,
+      "step": 1328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.097635782724066e-06,
+      "loss": 2.2548,
+      "step": 1329
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.095525652982833e-06,
+      "loss": 2.1959,
+      "step": 1330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.093413304224982e-06,
+      "loss": 2.2552,
+      "step": 1331
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.091298737595015e-06,
+      "loss": 2.3549,
+      "step": 1332
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.089181954238633e-06,
+      "loss": 2.3892,
+      "step": 1333
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.087062955302742e-06,
+      "loss": 2.4109,
+      "step": 1334
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.08494174193545e-06,
+      "loss": 2.2893,
+      "step": 1335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.082818315286054e-06,
+      "loss": 2.2751,
+      "step": 1336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.080692676505067e-06,
+      "loss": 2.3335,
+      "step": 1337
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.078564826744187e-06,
+      "loss": 2.3012,
+      "step": 1338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.076434767156317e-06,
+      "loss": 2.2512,
+      "step": 1339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.074302498895553e-06,
+      "loss": 2.3624,
+      "step": 1340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.07216802311719e-06,
+      "loss": 2.3061,
+      "step": 1341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.070031340977717e-06,
+      "loss": 2.2958,
+      "step": 1342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.067892453634824e-06,
+      "loss": 2.2715,
+      "step": 1343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.065751362247389e-06,
+      "loss": 2.2783,
+      "step": 1344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.063608067975486e-06,
+      "loss": 2.2505,
+      "step": 1345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.061462571980388e-06,
+      "loss": 2.2839,
+      "step": 1346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.059314875424553e-06,
+      "loss": 2.2137,
+      "step": 1347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.057164979471636e-06,
+      "loss": 2.3973,
+      "step": 1348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.055012885286483e-06,
+      "loss": 2.3203,
+      "step": 1349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.05285859403513e-06,
+      "loss": 2.4051,
+      "step": 1350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.050702106884806e-06,
+      "loss": 2.3046,
+      "step": 1351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.048543425003924e-06,
+      "loss": 2.3595,
+      "step": 1352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.046382549562092e-06,
+      "loss": 2.3117,
+      "step": 1353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.044219481730105e-06,
+      "loss": 2.2367,
+      "step": 1354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.042054222679944e-06,
+      "loss": 2.2948,
+      "step": 1355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.039886773584779e-06,
+      "loss": 2.2445,
+      "step": 1356
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.037717135618967e-06,
+      "loss": 2.4166,
+      "step": 1357
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.035545309958048e-06,
+      "loss": 2.3515,
+      "step": 1358
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.03337129777875e-06,
+      "loss": 2.3021,
+      "step": 1359
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.031195100258988e-06,
+      "loss": 2.2224,
+      "step": 1360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.029016718577855e-06,
+      "loss": 2.2485,
+      "step": 1361
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.02683615391563e-06,
+      "loss": 2.3368,
+      "step": 1362
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.024653407453776e-06,
+      "loss": 2.2119,
+      "step": 1363
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.02246848037494e-06,
+      "loss": 2.3733,
+      "step": 1364
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.020281373862947e-06,
+      "loss": 2.3468,
+      "step": 1365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.018092089102802e-06,
+      "loss": 2.3202,
+      "step": 1366
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.01590062728069e-06,
+      "loss": 2.335,
+      "step": 1367
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.013706989583984e-06,
+      "loss": 2.2939,
+      "step": 1368
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.011511177201226e-06,
+      "loss": 2.3684,
+      "step": 1369
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.009313191322138e-06,
+      "loss": 2.3136,
+      "step": 1370
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.007113033137622e-06,
+      "loss": 2.2628,
+      "step": 1371
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.00491070383976e-06,
+      "loss": 2.2427,
+      "step": 1372
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.002706204621802e-06,
+      "loss": 2.3267,
+      "step": 1373
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.000499536678182e-06,
+      "loss": 2.2148,
+      "step": 1374
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.998290701204503e-06,
+      "loss": 2.2874,
+      "step": 1375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.996079699397547e-06,
+      "loss": 2.3237,
+      "step": 1376
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.993866532455267e-06,
+      "loss": 2.2244,
+      "step": 1377
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.991651201576787e-06,
+      "loss": 2.323,
+      "step": 1378
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.98943370796241e-06,
+      "loss": 2.3785,
+      "step": 1379
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.987214052813605e-06,
+      "loss": 2.251,
+      "step": 1380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.984992237333015e-06,
+      "loss": 2.4431,
+      "step": 1381
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.98276826272445e-06,
+      "loss": 2.242,
+      "step": 1382
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.980542130192896e-06,
+      "loss": 2.375,
+      "step": 1383
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.978313840944504e-06,
+      "loss": 2.2978,
+      "step": 1384
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.976083396186592e-06,
+      "loss": 2.3489,
+      "step": 1385
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.97385079712765e-06,
+      "loss": 2.3065,
+      "step": 1386
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.971616044977334e-06,
+      "loss": 2.2981,
+      "step": 1387
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.969379140946464e-06,
+      "loss": 2.3462,
+      "step": 1388
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.967140086247028e-06,
+      "loss": 2.3966,
+      "step": 1389
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.964898882092179e-06,
+      "loss": 2.2984,
+      "step": 1390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.962655529696236e-06,
+      "loss": 2.2373,
+      "step": 1391
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.960410030274681e-06,
+      "loss": 2.2861,
+      "step": 1392
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.958162385044157e-06,
+      "loss": 2.234,
+      "step": 1393
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.955912595222474e-06,
+      "loss": 2.4036,
+      "step": 1394
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.953660662028598e-06,
+      "loss": 2.426,
+      "step": 1395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.951406586682663e-06,
+      "loss": 2.2833,
+      "step": 1396
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.949150370405958e-06,
+      "loss": 2.3081,
+      "step": 1397
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.946892014420936e-06,
+      "loss": 2.2456,
+      "step": 1398
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.94463151995121e-06,
+      "loss": 2.3629,
+      "step": 1399
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.942368888221544e-06,
+      "loss": 2.2686,
+      "step": 1400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.94010412045787e-06,
+      "loss": 2.2829,
+      "step": 1401
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.937837217887273e-06,
+      "loss": 2.3534,
+      "step": 1402
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.935568181737992e-06,
+      "loss": 2.3577,
+      "step": 1403
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.933297013239423e-06,
+      "loss": 2.2843,
+      "step": 1404
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.931023713622125e-06,
+      "loss": 2.2815,
+      "step": 1405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.9287482841178e-06,
+      "loss": 2.3069,
+      "step": 1406
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.92647072595931e-06,
+      "loss": 2.2674,
+      "step": 1407
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.924191040380672e-06,
+      "loss": 2.3558,
+      "step": 1408
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.921909228617053e-06,
+      "loss": 2.2676,
+      "step": 1409
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.91962529190477e-06,
+      "loss": 2.29,
+      "step": 1410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.9173392314813e-06,
+      "loss": 2.2791,
+      "step": 1411
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.915051048585257e-06,
+      "loss": 2.3156,
+      "step": 1412
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.912760744456416e-06,
+      "loss": 2.1909,
+      "step": 1413
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.910468320335699e-06,
+      "loss": 2.2972,
+      "step": 1414
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.908173777465172e-06,
+      "loss": 2.3049,
+      "step": 1415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.905877117088055e-06,
+      "loss": 2.3641,
+      "step": 1416
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.903578340448714e-06,
+      "loss": 2.3247,
+      "step": 1417
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.901277448792656e-06,
+      "loss": 2.2484,
+      "step": 1418
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.898974443366541e-06,
+      "loss": 2.2724,
+      "step": 1419
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.896669325418172e-06,
+      "loss": 2.3093,
+      "step": 1420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.894362096196497e-06,
+      "loss": 2.3061,
+      "step": 1421
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.892052756951604e-06,
+      "loss": 2.2743,
+      "step": 1422
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.889741308934728e-06,
+      "loss": 2.3662,
+      "step": 1423
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.887427753398249e-06,
+      "loss": 2.2697,
+      "step": 1424
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.88511209159568e-06,
+      "loss": 2.2913,
+      "step": 1425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.882794324781686e-06,
+      "loss": 2.3317,
+      "step": 1426
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.880474454212066e-06,
+      "loss": 2.26,
+      "step": 1427
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.87815248114376e-06,
+      "loss": 2.2303,
+      "step": 1428
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.875828406834848e-06,
+      "loss": 2.3417,
+      "step": 1429
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.873502232544547e-06,
+      "loss": 2.3481,
+      "step": 1430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.871173959533212e-06,
+      "loss": 2.4239,
+      "step": 1431
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.86884358906234e-06,
+      "loss": 2.3031,
+      "step": 1432
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.866511122394554e-06,
+      "loss": 2.3374,
+      "step": 1433
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.864176560793626e-06,
+      "loss": 2.3734,
+      "step": 1434
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.861839905524452e-06,
+      "loss": 2.186,
+      "step": 1435
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.859501157853067e-06,
+      "loss": 2.3123,
+      "step": 1436
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.85716031904664e-06,
+      "loss": 2.335,
+      "step": 1437
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.85481739037347e-06,
+      "loss": 2.2999,
+      "step": 1438
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.852472373102994e-06,
+      "loss": 2.3382,
+      "step": 1439
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.850125268505775e-06,
+      "loss": 2.3207,
+      "step": 1440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.84777607785351e-06,
+      "loss": 2.2747,
+      "step": 1441
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.845424802419025e-06,
+      "loss": 2.2664,
+      "step": 1442
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.843071443476274e-06,
+      "loss": 2.2899,
+      "step": 1443
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.840716002300347e-06,
+      "loss": 2.2725,
+      "step": 1444
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.838358480167453e-06,
+      "loss": 2.2709,
+      "step": 1445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.83599887835493e-06,
+      "loss": 2.3065,
+      "step": 1446
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.833637198141251e-06,
+      "loss": 2.4568,
+      "step": 1447
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.83127344080601e-06,
+      "loss": 2.3176,
+      "step": 1448
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.828907607629917e-06,
+      "loss": 2.2982,
+      "step": 1449
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.826539699894824e-06,
+      "loss": 2.4182,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.824169718883697e-06,
+      "loss": 2.3327,
+      "step": 1451
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.821797665880626e-06,
+      "loss": 2.3289,
+      "step": 1452
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.819423542170824e-06,
+      "loss": 2.2234,
+      "step": 1453
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.817047349040626e-06,
+      "loss": 2.3441,
+      "step": 1454
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.81466908777749e-06,
+      "loss": 2.2874,
+      "step": 1455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.812288759669994e-06,
+      "loss": 2.3341,
+      "step": 1456
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.809906366007832e-06,
+      "loss": 2.3259,
+      "step": 1457
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.807521908081823e-06,
+      "loss": 2.3984,
+      "step": 1458
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.8051353871839e-06,
+      "loss": 2.2912,
+      "step": 1459
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.802746804607119e-06,
+      "loss": 2.3146,
+      "step": 1460
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.800356161645646e-06,
+      "loss": 2.3134,
+      "step": 1461
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.797963459594768e-06,
+      "loss": 2.2918,
+      "step": 1462
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.795568699750886e-06,
+      "loss": 2.3115,
+      "step": 1463
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.793171883411515e-06,
+      "loss": 2.3348,
+      "step": 1464
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.790773011875288e-06,
+      "loss": 2.2885,
+      "step": 1465
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.788372086441949e-06,
+      "loss": 2.312,
+      "step": 1466
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.785969108412357e-06,
+      "loss": 2.374,
+      "step": 1467
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.783564079088478e-06,
+      "loss": 2.3357,
+      "step": 1468
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.781156999773392e-06,
+      "loss": 2.355,
+      "step": 1469
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.778747871771293e-06,
+      "loss": 2.3247,
+      "step": 1470
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.77633669638748e-06,
+      "loss": 2.2436,
+      "step": 1471
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.773923474928365e-06,
+      "loss": 2.3358,
+      "step": 1472
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.771508208701469e-06,
+      "loss": 2.2679,
+      "step": 1473
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.769090899015415e-06,
+      "loss": 2.241,
+      "step": 1474
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.766671547179939e-06,
+      "loss": 2.223,
+      "step": 1475
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.764250154505884e-06,
+      "loss": 2.3545,
+      "step": 1476
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.761826722305193e-06,
+      "loss": 2.2603,
+      "step": 1477
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.759401251890921e-06,
+      "loss": 2.3809,
+      "step": 1478
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.756973744577221e-06,
+      "loss": 2.1561,
+      "step": 1479
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.754544201679354e-06,
+      "loss": 2.3777,
+      "step": 1480
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.752112624513682e-06,
+      "loss": 2.2837,
+      "step": 1481
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.749679014397669e-06,
+      "loss": 2.1957,
+      "step": 1482
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.747243372649883e-06,
+      "loss": 2.2731,
+      "step": 1483
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.744805700589989e-06,
+      "loss": 2.2183,
+      "step": 1484
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.742365999538755e-06,
+      "loss": 2.3241,
+      "step": 1485
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.739924270818046e-06,
+      "loss": 2.2428,
+      "step": 1486
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.737480515750827e-06,
+      "loss": 2.2455,
+      "step": 1487
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.735034735661162e-06,
+      "loss": 2.3297,
+      "step": 1488
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.732586931874212e-06,
+      "loss": 2.2946,
+      "step": 1489
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.730137105716231e-06,
+      "loss": 2.2166,
+      "step": 1490
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.72768525851457e-06,
+      "loss": 2.4585,
+      "step": 1491
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.72523139159768e-06,
+      "loss": 2.3683,
+      "step": 1492
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.722775506295103e-06,
+      "loss": 2.2555,
+      "step": 1493
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.72031760393747e-06,
+      "loss": 2.2878,
+      "step": 1494
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.717857685856512e-06,
+      "loss": 2.3987,
+      "step": 1495
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.715395753385048e-06,
+      "loss": 2.2986,
+      "step": 1496
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.712931807856988e-06,
+      "loss": 2.2573,
+      "step": 1497
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.710465850607336e-06,
+      "loss": 2.3882,
+      "step": 1498
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.707997882972186e-06,
+      "loss": 2.3822,
+      "step": 1499
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.705527906288718e-06,
+      "loss": 2.3332,
+      "step": 1500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.7030559218952e-06,
+      "loss": 2.2804,
+      "step": 1501
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.700581931130992e-06,
+      "loss": 2.2952,
+      "step": 1502
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.69810593533654e-06,
+      "loss": 2.342,
+      "step": 1503
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.695627935853372e-06,
+      "loss": 2.1959,
+      "step": 1504
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.693147934024107e-06,
+      "loss": 2.3128,
+      "step": 1505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.690665931192447e-06,
+      "loss": 2.2957,
+      "step": 1506
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.688181928703179e-06,
+      "loss": 2.4012,
+      "step": 1507
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.68569592790217e-06,
+      "loss": 2.2865,
+      "step": 1508
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.683207930136372e-06,
+      "loss": 2.2266,
+      "step": 1509
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.680717936753822e-06,
+      "loss": 2.2883,
+      "step": 1510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.678225949103634e-06,
+      "loss": 2.3273,
+      "step": 1511
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.675731968536004e-06,
+      "loss": 2.2913,
+      "step": 1512
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.673235996402204e-06,
+      "loss": 2.3961,
+      "step": 1513
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.670738034054596e-06,
+      "loss": 2.3068,
+      "step": 1514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.668238082846607e-06,
+      "loss": 2.2784,
+      "step": 1515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.66573614413275e-06,
+      "loss": 2.2796,
+      "step": 1516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.663232219268613e-06,
+      "loss": 2.2903,
+      "step": 1517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.66072630961086e-06,
+      "loss": 2.2903,
+      "step": 1518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.658218416517227e-06,
+      "loss": 2.3005,
+      "step": 1519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.655708541346532e-06,
+      "loss": 2.3177,
+      "step": 1520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.653196685458661e-06,
+      "loss": 2.2346,
+      "step": 1521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.650682850214572e-06,
+      "loss": 2.335,
+      "step": 1522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.648167036976303e-06,
+      "loss": 2.3887,
+      "step": 1523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.645649247106956e-06,
+      "loss": 2.2896,
+      "step": 1524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.643129481970705e-06,
+      "loss": 2.2834,
+      "step": 1525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.640607742932802e-06,
+      "loss": 2.3397,
+      "step": 1526
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.638084031359558e-06,
+      "loss": 2.303,
+      "step": 1527
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.635558348618359e-06,
+      "loss": 2.2766,
+      "step": 1528
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.633030696077658e-06,
+      "loss": 2.3061,
+      "step": 1529
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.630501075106971e-06,
+      "loss": 2.3324,
+      "step": 1530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.62796948707689e-06,
+      "loss": 2.3542,
+      "step": 1531
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.625435933359063e-06,
+      "loss": 2.3395,
+      "step": 1532
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.622900415326209e-06,
+      "loss": 2.3095,
+      "step": 1533
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.620362934352109e-06,
+      "loss": 2.3598,
+      "step": 1534
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.617823491811605e-06,
+      "loss": 2.2988,
+      "step": 1535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.61528208908061e-06,
+      "loss": 2.3248,
+      "step": 1536
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.612738727536087e-06,
+      "loss": 2.305,
+      "step": 1537
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.610193408556073e-06,
+      "loss": 2.26,
+      "step": 1538
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.607646133519658e-06,
+      "loss": 2.2776,
+      "step": 1539
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.605096903806991e-06,
+      "loss": 2.3671,
+      "step": 1540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.602545720799287e-06,
+      "loss": 2.308,
+      "step": 1541
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.59999258587881e-06,
+      "loss": 2.278,
+      "step": 1542
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.59743750042889e-06,
+      "loss": 2.3007,
+      "step": 1543
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.594880465833908e-06,
+      "loss": 2.3055,
+      "step": 1544
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.592321483479305e-06,
+      "loss": 2.2714,
+      "step": 1545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.589760554751573e-06,
+      "loss": 2.3349,
+      "step": 1546
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.587197681038266e-06,
+      "loss": 2.2145,
+      "step": 1547
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.584632863727982e-06,
+      "loss": 2.4354,
+      "step": 1548
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.58206610421038e-06,
+      "loss": 2.3104,
+      "step": 1549
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.579497403876166e-06,
+      "loss": 2.2533,
+      "step": 1550
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.576926764117104e-06,
+      "loss": 2.3432,
+      "step": 1551
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.574354186326001e-06,
+      "loss": 2.385,
+      "step": 1552
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.571779671896721e-06,
+      "loss": 2.2505,
+      "step": 1553
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.56920322222417e-06,
+      "loss": 2.327,
+      "step": 1554
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.566624838704309e-06,
+      "loss": 2.2762,
+      "step": 1555
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.564044522734147e-06,
+      "loss": 2.2881,
+      "step": 1556
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.561462275711734e-06,
+      "loss": 2.2824,
+      "step": 1557
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.55887809903617e-06,
+      "loss": 2.2495,
+      "step": 1558
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.556291994107603e-06,
+      "loss": 2.2535,
+      "step": 1559
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.55370396232722e-06,
+      "loss": 2.3213,
+      "step": 1560
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.551114005097256e-06,
+      "loss": 2.3362,
+      "step": 1561
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.548522123820989e-06,
+      "loss": 2.3648,
+      "step": 1562
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.545928319902739e-06,
+      "loss": 2.3816,
+      "step": 1563
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.543332594747864e-06,
+      "loss": 2.2506,
+      "step": 1564
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.540734949762773e-06,
+      "loss": 2.2287,
+      "step": 1565
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.538135386354903e-06,
+      "loss": 2.3021,
+      "step": 1566
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.535533905932739e-06,
+      "loss": 2.2951,
+      "step": 1567
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.532930509905799e-06,
+      "loss": 2.3609,
+      "step": 1568
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.530325199684645e-06,
+      "loss": 2.3293,
+      "step": 1569
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.527717976680868e-06,
+      "loss": 2.3271,
+      "step": 1570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.525108842307107e-06,
+      "loss": 2.4045,
+      "step": 1571
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.522497797977026e-06,
+      "loss": 2.3673,
+      "step": 1572
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.519884845105325e-06,
+      "loss": 2.342,
+      "step": 1573
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.517269985107743e-06,
+      "loss": 2.2772,
+      "step": 1574
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.514653219401052e-06,
+      "loss": 2.2926,
+      "step": 1575
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.512034549403053e-06,
+      "loss": 2.2553,
+      "step": 1576
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.50941397653258e-06,
+      "loss": 2.2814,
+      "step": 1577
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.506791502209497e-06,
+      "loss": 2.3583,
+      "step": 1578
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.5041671278547e-06,
+      "loss": 2.3136,
+      "step": 1579
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.501540854890117e-06,
+      "loss": 2.3254,
+      "step": 1580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.498912684738699e-06,
+      "loss": 2.348,
+      "step": 1581
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.496282618824427e-06,
+      "loss": 2.3065,
+      "step": 1582
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.493650658572312e-06,
+      "loss": 2.2881,
+      "step": 1583
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.491016805408387e-06,
+      "loss": 2.2096,
+      "step": 1584
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.488381060759716e-06,
+      "loss": 2.2811,
+      "step": 1585
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.485743426054381e-06,
+      "loss": 2.3124,
+      "step": 1586
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.483103902721494e-06,
+      "loss": 2.3407,
+      "step": 1587
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.480462492191187e-06,
+      "loss": 2.2749,
+      "step": 1588
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.477819195894615e-06,
+      "loss": 2.2224,
+      "step": 1589
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.475174015263954e-06,
+      "loss": 2.2764,
+      "step": 1590
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.472526951732405e-06,
+      "loss": 2.3216,
+      "step": 1591
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.469878006734186e-06,
+      "loss": 2.2662,
+      "step": 1592
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.467227181704533e-06,
+      "loss": 2.272,
+      "step": 1593
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.464574478079703e-06,
+      "loss": 2.3034,
+      "step": 1594
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.461919897296971e-06,
+      "loss": 2.3194,
+      "step": 1595
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.459263440794627e-06,
+      "loss": 2.2977,
+      "step": 1596
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.45660511001198e-06,
+      "loss": 2.2801,
+      "step": 1597
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.453944906389353e-06,
+      "loss": 2.3048,
+      "step": 1598
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.451282831368083e-06,
+      "loss": 2.32,
+      "step": 1599
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.448618886390523e-06,
+      "loss": 2.3204,
+      "step": 1600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.445953072900034e-06,
+      "loss": 2.3274,
+      "step": 1601
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.443285392340999e-06,
+      "loss": 2.3106,
+      "step": 1602
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.440615846158802e-06,
+      "loss": 2.3501,
+      "step": 1603
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.437944435799847e-06,
+      "loss": 2.3007,
+      "step": 1604
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.435271162711542e-06,
+      "loss": 2.2879,
+      "step": 1605
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.432596028342304e-06,
+      "loss": 2.3046,
+      "step": 1606
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.429919034141564e-06,
+      "loss": 2.2836,
+      "step": 1607
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.427240181559754e-06,
+      "loss": 2.2448,
+      "step": 1608
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.424559472048319e-06,
+      "loss": 2.2727,
+      "step": 1609
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.421876907059706e-06,
+      "loss": 2.2703,
+      "step": 1610
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.419192488047369e-06,
+      "loss": 2.3755,
+      "step": 1611
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.416506216465766e-06,
+      "loss": 2.3198,
+      "step": 1612
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.413818093770355e-06,
+      "loss": 2.3314,
+      "step": 1613
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.411128121417609e-06,
+      "loss": 2.3077,
+      "step": 1614
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.408436300864986e-06,
+      "loss": 2.1735,
+      "step": 1615
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.405742633570961e-06,
+      "loss": 2.2792,
+      "step": 1616
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.403047120995e-06,
+      "loss": 2.2147,
+      "step": 1617
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.400349764597572e-06,
+      "loss": 2.2663,
+      "step": 1618
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.397650565840146e-06,
+      "loss": 2.2851,
+      "step": 1619
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.394949526185185e-06,
+      "loss": 2.2566,
+      "step": 1620
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.392246647096156e-06,
+      "loss": 2.3935,
+      "step": 1621
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.389541930037516e-06,
+      "loss": 2.2424,
+      "step": 1622
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.386835376474725e-06,
+      "loss": 2.2418,
+      "step": 1623
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.384126987874227e-06,
+      "loss": 2.304,
+      "step": 1624
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.381416765703474e-06,
+      "loss": 2.313,
+      "step": 1625
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.378704711430902e-06,
+      "loss": 2.2557,
+      "step": 1626
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.375990826525941e-06,
+      "loss": 2.2162,
+      "step": 1627
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.373275112459016e-06,
+      "loss": 2.3055,
+      "step": 1628
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.370557570701541e-06,
+      "loss": 2.3362,
+      "step": 1629
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.36783820272592e-06,
+      "loss": 2.3485,
+      "step": 1630
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.365117010005547e-06,
+      "loss": 2.2757,
+      "step": 1631
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.362393994014804e-06,
+      "loss": 2.2647,
+      "step": 1632
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.359669156229062e-06,
+      "loss": 2.2687,
+      "step": 1633
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.35694249812468e-06,
+      "loss": 2.2788,
+      "step": 1634
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.354214021179e-06,
+      "loss": 2.2616,
+      "step": 1635
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.351483726870351e-06,
+      "loss": 2.2737,
+      "step": 1636
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.348751616678051e-06,
+      "loss": 2.3175,
+      "step": 1637
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.34601769208239e-06,
+      "loss": 2.3183,
+      "step": 1638
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.343281954564657e-06,
+      "loss": 2.2726,
+      "step": 1639
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.340544405607112e-06,
+      "loss": 2.4226,
+      "step": 1640
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.337805046693e-06,
+      "loss": 2.332,
+      "step": 1641
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.335063879306544e-06,
+      "loss": 2.3202,
+      "step": 1642
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.332320904932951e-06,
+      "loss": 2.3515,
+      "step": 1643
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.329576125058406e-06,
+      "loss": 2.2466,
+      "step": 1644
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.326829541170072e-06,
+      "loss": 2.2513,
+      "step": 1645
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.324081154756088e-06,
+      "loss": 2.35,
+      "step": 1646
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.321330967305568e-06,
+      "loss": 2.3324,
+      "step": 1647
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.31857898030861e-06,
+      "loss": 2.3045,
+      "step": 1648
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.315825195256276e-06,
+      "loss": 2.287,
+      "step": 1649
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.31306961364061e-06,
+      "loss": 2.3076,
+      "step": 1650
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.310312236954628e-06,
+      "loss": 2.2966,
+      "step": 1651
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.307553066692314e-06,
+      "loss": 2.2564,
+      "step": 1652
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.304792104348634e-06,
+      "loss": 2.3609,
+      "step": 1653
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.302029351419511e-06,
+      "loss": 2.3325,
+      "step": 1654
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.29926480940185e-06,
+      "loss": 2.3019,
+      "step": 1655
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.29649847979352e-06,
+      "loss": 2.3137,
+      "step": 1656
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.29373036409336e-06,
+      "loss": 2.2215,
+      "step": 1657
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.290960463801174e-06,
+      "loss": 2.3733,
+      "step": 1658
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.288188780417738e-06,
+      "loss": 2.2004,
+      "step": 1659
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.28541531544479e-06,
+      "loss": 2.3232,
+      "step": 1660
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.282640070385033e-06,
+      "loss": 2.2639,
+      "step": 1661
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.279863046742138e-06,
+      "loss": 2.3515,
+      "step": 1662
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.277084246020737e-06,
+      "loss": 2.3774,
+      "step": 1663
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.274303669726427e-06,
+      "loss": 2.2858,
+      "step": 1664
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.271521319365761e-06,
+      "loss": 2.3251,
+      "step": 1665
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.268737196446264e-06,
+      "loss": 2.3049,
+      "step": 1666
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.265951302476408e-06,
+      "loss": 2.2706,
+      "step": 1667
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.263163638965639e-06,
+      "loss": 2.2805,
+      "step": 1668
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.26037420742435e-06,
+      "loss": 2.296,
+      "step": 1669
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.257583009363899e-06,
+      "loss": 2.1748,
+      "step": 1670
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.254790046296595e-06,
+      "loss": 2.3427,
+      "step": 1671
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.251995319735711e-06,
+      "loss": 2.2857,
+      "step": 1672
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.249198831195468e-06,
+      "loss": 2.2987,
+      "step": 1673
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.246400582191045e-06,
+      "loss": 2.2138,
+      "step": 1674
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.243600574238577e-06,
+      "loss": 2.2833,
+      "step": 1675
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.24079880885515e-06,
+      "loss": 2.3076,
+      "step": 1676
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.237995287558801e-06,
+      "loss": 2.2946,
+      "step": 1677
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.235190011868522e-06,
+      "loss": 2.2956,
+      "step": 1678
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.232382983304245e-06,
+      "loss": 2.268,
+      "step": 1679
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.22957420338687e-06,
+      "loss": 2.3297,
+      "step": 1680
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.226763673638231e-06,
+      "loss": 2.2467,
+      "step": 1681
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.223951395581114e-06,
+      "loss": 2.3081,
+      "step": 1682
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.221137370739255e-06,
+      "loss": 2.3089,
+      "step": 1683
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.21832160063733e-06,
+      "loss": 2.2988,
+      "step": 1684
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.215504086800972e-06,
+      "loss": 2.3437,
+      "step": 1685
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.212684830756745e-06,
+      "loss": 2.3431,
+      "step": 1686
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.20986383403217e-06,
+      "loss": 2.3567,
+      "step": 1687
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.207041098155701e-06,
+      "loss": 2.2134,
+      "step": 1688
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.204216624656738e-06,
+      "loss": 2.2678,
+      "step": 1689
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.201390415065626e-06,
+      "loss": 2.31,
+      "step": 1690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.198562470913643e-06,
+      "loss": 2.2739,
+      "step": 1691
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.195732793733014e-06,
+      "loss": 2.2527,
+      "step": 1692
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.192901385056899e-06,
+      "loss": 2.3084,
+      "step": 1693
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.1900682464194e-06,
+      "loss": 2.2878,
+      "step": 1694
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.18723337935555e-06,
+      "loss": 2.2809,
+      "step": 1695
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.184396785401322e-06,
+      "loss": 2.3151,
+      "step": 1696
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.181558466093628e-06,
+      "loss": 2.3102,
+      "step": 1697
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.17871842297031e-06,
+      "loss": 2.2206,
+      "step": 1698
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.175876657570142e-06,
+      "loss": 2.2214,
+      "step": 1699
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.173033171432841e-06,
+      "loss": 2.2798,
+      "step": 1700
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.170187966099045e-06,
+      "loss": 2.3492,
+      "step": 1701
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.167341043110332e-06,
+      "loss": 2.3087,
+      "step": 1702
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.164492404009206e-06,
+      "loss": 2.2892,
+      "step": 1703
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.1616420503391e-06,
+      "loss": 2.2874,
+      "step": 1704
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.15878998364438e-06,
+      "loss": 2.3907,
+      "step": 1705
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.155936205470338e-06,
+      "loss": 2.3322,
+      "step": 1706
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.153080717363193e-06,
+      "loss": 2.2456,
+      "step": 1707
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.15022352087009e-06,
+      "loss": 2.3185,
+      "step": 1708
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.1473646175391e-06,
+      "loss": 2.3583,
+      "step": 1709
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.144504008919224e-06,
+      "loss": 2.4012,
+      "step": 1710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.141641696560376e-06,
+      "loss": 2.2644,
+      "step": 1711
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.138777682013403e-06,
+      "loss": 2.1765,
+      "step": 1712
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.13591196683007e-06,
+      "loss": 2.3476,
+      "step": 1713
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.133044552563065e-06,
+      "loss": 2.327,
+      "step": 1714
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.130175440765993e-06,
+      "loss": 2.2523,
+      "step": 1715
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.127304632993383e-06,
+      "loss": 2.3566,
+      "step": 1716
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.124432130800682e-06,
+      "loss": 2.2313,
+      "step": 1717
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.12155793574425e-06,
+      "loss": 2.2918,
+      "step": 1718
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.118682049381377e-06,
+      "loss": 2.2942,
+      "step": 1719
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.115804473270253e-06,
+      "loss": 2.2957,
+      "step": 1720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.112925208969996e-06,
+      "loss": 2.2216,
+      "step": 1721
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.110044258040631e-06,
+      "loss": 2.3843,
+      "step": 1722
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.107161622043103e-06,
+      "loss": 2.2539,
+      "step": 1723
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.104277302539264e-06,
+      "loss": 2.385,
+      "step": 1724
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.101391301091883e-06,
+      "loss": 2.2631,
+      "step": 1725
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.09850361926464e-06,
+      "loss": 2.2904,
+      "step": 1726
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.09561425862212e-06,
+      "loss": 2.2228,
+      "step": 1727
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.092723220729826e-06,
+      "loss": 2.343,
+      "step": 1728
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.089830507154159e-06,
+      "loss": 2.2969,
+      "step": 1729
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.086936119462441e-06,
+      "loss": 2.1749,
+      "step": 1730
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.08404005922289e-06,
+      "loss": 2.4112,
+      "step": 1731
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.081142328004638e-06,
+      "loss": 2.2755,
+      "step": 1732
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.078242927377714e-06,
+      "loss": 2.4117,
+      "step": 1733
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.07534185891306e-06,
+      "loss": 2.2949,
+      "step": 1734
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.072439124182517e-06,
+      "loss": 2.3394,
+      "step": 1735
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.069534724758828e-06,
+      "loss": 2.1994,
+      "step": 1736
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.066628662215642e-06,
+      "loss": 2.2709,
+      "step": 1737
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.063720938127505e-06,
+      "loss": 2.2489,
+      "step": 1738
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.06081155406987e-06,
+      "loss": 2.2775,
+      "step": 1739
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.057900511619077e-06,
+      "loss": 2.34,
+      "step": 1740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.054987812352377e-06,
+      "loss": 2.3526,
+      "step": 1741
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.052073457847913e-06,
+      "loss": 2.3093,
+      "step": 1742
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.049157449684723e-06,
+      "loss": 2.2613,
+      "step": 1743
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.04623978944275e-06,
+      "loss": 2.2729,
+      "step": 1744
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.043320478702816e-06,
+      "loss": 2.3662,
+      "step": 1745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.040399519046656e-06,
+      "loss": 2.2924,
+      "step": 1746
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.037476912056883e-06,
+      "loss": 2.2332,
+      "step": 1747
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.034552659317012e-06,
+      "loss": 2.3613,
+      "step": 1748
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.031626762411446e-06,
+      "loss": 2.3639,
+      "step": 1749
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.028699222925479e-06,
+      "loss": 2.2905,
+      "step": 1750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.025770042445295e-06,
+      "loss": 2.2132,
+      "step": 1751
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.02283922255797e-06,
+      "loss": 2.3068,
+      "step": 1752
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.019906764851463e-06,
+      "loss": 2.3033,
+      "step": 1753
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.016972670914624e-06,
+      "loss": 2.3464,
+      "step": 1754
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.01403694233719e-06,
+      "loss": 2.2323,
+      "step": 1755
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.011099580709777e-06,
+      "loss": 2.3458,
+      "step": 1756
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.0081605876239e-06,
+      "loss": 2.2339,
+      "step": 1757
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.00521996467194e-06,
+      "loss": 2.1688,
+      "step": 1758
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.002277713447176e-06,
+      "loss": 2.2594,
+      "step": 1759
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.999333835543764e-06,
+      "loss": 2.362,
+      "step": 1760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.996388332556735e-06,
+      "loss": 2.3117,
+      "step": 1761
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.993441206082012e-06,
+      "loss": 2.2602,
+      "step": 1762
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.99049245771639e-06,
+      "loss": 2.3219,
+      "step": 1763
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.987542089057543e-06,
+      "loss": 2.2371,
+      "step": 1764
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.984590101704026e-06,
+      "loss": 2.2296,
+      "step": 1765
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.98163649725527e-06,
+      "loss": 2.1818,
+      "step": 1766
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.978681277311582e-06,
+      "loss": 2.3093,
+      "step": 1767
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.975724443474143e-06,
+      "loss": 2.3077,
+      "step": 1768
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.97276599734501e-06,
+      "loss": 2.4012,
+      "step": 1769
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.969805940527112e-06,
+      "loss": 2.3573,
+      "step": 1770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.966844274624255e-06,
+      "loss": 2.3921,
+      "step": 1771
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.963881001241106e-06,
+      "loss": 2.2505,
+      "step": 1772
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.960916121983221e-06,
+      "loss": 2.3285,
+      "step": 1773
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.95794963845701e-06,
+      "loss": 2.2698,
+      "step": 1774
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.954981552269755e-06,
+      "loss": 2.3099,
+      "step": 1775
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.952011865029614e-06,
+      "loss": 2.2218,
+      "step": 1776
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.949040578345606e-06,
+      "loss": 2.2478,
+      "step": 1777
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.946067693827618e-06,
+      "loss": 2.3581,
+      "step": 1778
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.943093213086404e-06,
+      "loss": 2.3068,
+      "step": 1779
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.94011713773358e-06,
+      "loss": 2.2485,
+      "step": 1780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.93713946938163e-06,
+      "loss": 2.3019,
+      "step": 1781
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.934160209643897e-06,
+      "loss": 2.2979,
+      "step": 1782
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.931179360134588e-06,
+      "loss": 2.3282,
+      "step": 1783
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.928196922468772e-06,
+      "loss": 2.3109,
+      "step": 1784
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.92521289826238e-06,
+      "loss": 2.368,
+      "step": 1785
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.9222272891322e-06,
+      "loss": 2.3244,
+      "step": 1786
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.919240096695877e-06,
+      "loss": 2.3401,
+      "step": 1787
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.916251322571918e-06,
+      "loss": 2.3296,
+      "step": 1788
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.913260968379687e-06,
+      "loss": 2.2345,
+      "step": 1789
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.910269035739398e-06,
+      "loss": 2.3176,
+      "step": 1790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.907275526272127e-06,
+      "loss": 2.3272,
+      "step": 1791
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.904280441599802e-06,
+      "loss": 2.2694,
+      "step": 1792
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.901283783345204e-06,
+      "loss": 2.3929,
+      "step": 1793
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.898285553131967e-06,
+      "loss": 2.3338,
+      "step": 1794
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.895285752584577e-06,
+      "loss": 2.2242,
+      "step": 1795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.892284383328367e-06,
+      "loss": 2.3234,
+      "step": 1796
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.88928144698953e-06,
+      "loss": 2.3216,
+      "step": 1797
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.886276945195098e-06,
+      "loss": 2.2411,
+      "step": 1798
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.883270879572954e-06,
+      "loss": 2.195,
+      "step": 1799
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.88026325175183e-06,
+      "loss": 2.3413,
+      "step": 1800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.877254063361301e-06,
+      "loss": 2.2159,
+      "step": 1801
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.874243316031795e-06,
+      "loss": 2.24,
+      "step": 1802
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.871231011394575e-06,
+      "loss": 2.2249,
+      "step": 1803
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.868217151081757e-06,
+      "loss": 2.2656,
+      "step": 1804
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.86520173672629e-06,
+      "loss": 2.2384,
+      "step": 1805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.862184769961972e-06,
+      "loss": 2.2824,
+      "step": 1806
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.85916625242344e-06,
+      "loss": 2.3403,
+      "step": 1807
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.856146185746175e-06,
+      "loss": 2.2463,
+      "step": 1808
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.853124571566491e-06,
+      "loss": 2.254,
+      "step": 1809
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.850101411521544e-06,
+      "loss": 2.2214,
+      "step": 1810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.847076707249328e-06,
+      "loss": 2.1786,
+      "step": 1811
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.844050460388672e-06,
+      "loss": 2.234,
+      "step": 1812
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.84102267257924e-06,
+      "loss": 2.3383,
+      "step": 1813
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.837993345461535e-06,
+      "loss": 2.2332,
+      "step": 1814
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.834962480676888e-06,
+      "loss": 2.3035,
+      "step": 1815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.83193007986747e-06,
+      "loss": 2.2634,
+      "step": 1816
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.82889614467628e-06,
+      "loss": 2.3629,
+      "step": 1817
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.825860676747146e-06,
+      "loss": 2.2711,
+      "step": 1818
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.822823677724734e-06,
+      "loss": 2.3559,
+      "step": 1819
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.819785149254534e-06,
+      "loss": 2.3124,
+      "step": 1820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.816745092982863e-06,
+      "loss": 2.3232,
+      "step": 1821
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.813703510556874e-06,
+      "loss": 2.2779,
+      "step": 1822
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.810660403624535e-06,
+      "loss": 2.2456,
+      "step": 1823
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.807615773834653e-06,
+      "loss": 2.2928,
+      "step": 1824
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.804569622836848e-06,
+      "loss": 2.3251,
+      "step": 1825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.801521952281574e-06,
+      "loss": 2.305,
+      "step": 1826
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.798472763820102e-06,
+      "loss": 2.3043,
+      "step": 1827
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.795422059104528e-06,
+      "loss": 2.3232,
+      "step": 1828
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.79236983978777e-06,
+      "loss": 2.2496,
+      "step": 1829
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.789316107523564e-06,
+      "loss": 2.2864,
+      "step": 1830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.786260863966468e-06,
+      "loss": 2.2916,
+      "step": 1831
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.78320411077186e-06,
+      "loss": 2.3494,
+      "step": 1832
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.780145849595932e-06,
+      "loss": 2.2686,
+      "step": 1833
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.777086082095698e-06,
+      "loss": 2.271,
+      "step": 1834
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.774024809928984e-06,
+      "loss": 2.2969,
+      "step": 1835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.77096203475443e-06,
+      "loss": 2.3604,
+      "step": 1836
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.767897758231498e-06,
+      "loss": 2.2909,
+      "step": 1837
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.764831982020455e-06,
+      "loss": 2.3376,
+      "step": 1838
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.761764707782386e-06,
+      "loss": 2.1701,
+      "step": 1839
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.758695937179185e-06,
+      "loss": 2.2787,
+      "step": 1840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.755625671873556e-06,
+      "loss": 2.3707,
+      "step": 1841
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.752553913529019e-06,
+      "loss": 2.2895,
+      "step": 1842
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.749480663809891e-06,
+      "loss": 2.3095,
+      "step": 1843
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.746405924381312e-06,
+      "loss": 2.3281,
+      "step": 1844
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.743329696909217e-06,
+      "loss": 2.2854,
+      "step": 1845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.740251983060352e-06,
+      "loss": 2.3032,
+      "step": 1846
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.737172784502269e-06,
+      "loss": 2.2708,
+      "step": 1847
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.734092102903323e-06,
+      "loss": 2.2553,
+      "step": 1848
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.731009939932674e-06,
+      "loss": 2.3417,
+      "step": 1849
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.727926297260282e-06,
+      "loss": 2.2869,
+      "step": 1850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.724841176556913e-06,
+      "loss": 2.2763,
+      "step": 1851
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.721754579494127e-06,
+      "loss": 2.2724,
+      "step": 1852
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.718666507744293e-06,
+      "loss": 2.2927,
+      "step": 1853
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.71557696298057e-06,
+      "loss": 2.392,
+      "step": 1854
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.712485946876924e-06,
+      "loss": 2.4016,
+      "step": 1855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.709393461108108e-06,
+      "loss": 2.2547,
+      "step": 1856
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.70629950734968e-06,
+      "loss": 2.3427,
+      "step": 1857
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.703204087277989e-06,
+      "loss": 2.2447,
+      "step": 1858
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.70010720257018e-06,
+      "loss": 2.247,
+      "step": 1859
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.69700885490419e-06,
+      "loss": 2.2188,
+      "step": 1860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.693909045958751e-06,
+      "loss": 2.2962,
+      "step": 1861
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.690807777413387e-06,
+      "loss": 2.3202,
+      "step": 1862
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.687705050948408e-06,
+      "loss": 2.1824,
+      "step": 1863
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.68460086824492e-06,
+      "loss": 2.3836,
+      "step": 1864
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.681495230984813e-06,
+      "loss": 2.3605,
+      "step": 1865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.678388140850769e-06,
+      "loss": 2.2938,
+      "step": 1866
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.675279599526256e-06,
+      "loss": 2.2717,
+      "step": 1867
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.672169608695526e-06,
+      "loss": 2.2392,
+      "step": 1868
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.66905817004362e-06,
+      "loss": 2.342,
+      "step": 1869
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.66594528525636e-06,
+      "loss": 2.1736,
+      "step": 1870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.662830956020356e-06,
+      "loss": 2.2992,
+      "step": 1871
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.659715184022993e-06,
+      "loss": 2.3774,
+      "step": 1872
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.656597970952447e-06,
+      "loss": 2.2525,
+      "step": 1873
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.65347931849767e-06,
+      "loss": 2.3332,
+      "step": 1874
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.65035922834839e-06,
+      "loss": 2.2894,
+      "step": 1875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.647237702195123e-06,
+      "loss": 2.3197,
+      "step": 1876
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.644114741729156e-06,
+      "loss": 2.3167,
+      "step": 1877
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.640990348642557e-06,
+      "loss": 2.3387,
+      "step": 1878
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.637864524628167e-06,
+      "loss": 2.2979,
+      "step": 1879
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.634737271379603e-06,
+      "loss": 2.3518,
+      "step": 1880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.63160859059126e-06,
+      "loss": 2.203,
+      "step": 1881
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.628478483958299e-06,
+      "loss": 2.2308,
+      "step": 1882
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.625346953176664e-06,
+      "loss": 2.3267,
+      "step": 1883
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.622213999943062e-06,
+      "loss": 2.3048,
+      "step": 1884
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.619079625954975e-06,
+      "loss": 2.2979,
+      "step": 1885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.61594383291065e-06,
+      "loss": 2.4306,
+      "step": 1886
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.61280662250911e-06,
+      "loss": 2.3561,
+      "step": 1887
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.609667996450141e-06,
+      "loss": 2.3231,
+      "step": 1888
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.606527956434298e-06,
+      "loss": 2.249,
+      "step": 1889
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.603386504162897e-06,
+      "loss": 2.2245,
+      "step": 1890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.600243641338028e-06,
+      "loss": 2.3394,
+      "step": 1891
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.597099369662539e-06,
+      "loss": 2.274,
+      "step": 1892
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.593953690840046e-06,
+      "loss": 2.2925,
+      "step": 1893
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.590806606574921e-06,
+      "loss": 2.2356,
+      "step": 1894
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.5876581185723005e-06,
+      "loss": 2.356,
+      "step": 1895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.584508228538086e-06,
+      "loss": 2.3597,
+      "step": 1896
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.58135693817893e-06,
+      "loss": 2.284,
+      "step": 1897
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.57820424920225e-06,
+      "loss": 2.2978,
+      "step": 1898
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.575050163316223e-06,
+      "loss": 2.2496,
+      "step": 1899
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.571894682229776e-06,
+      "loss": 2.3533,
+      "step": 1900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.568737807652596e-06,
+      "loss": 2.3003,
+      "step": 1901
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.565579541295125e-06,
+      "loss": 2.3273,
+      "step": 1902
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.562419884868559e-06,
+      "loss": 2.4081,
+      "step": 1903
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.559258840084848e-06,
+      "loss": 2.2733,
+      "step": 1904
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.55609640865669e-06,
+      "loss": 2.2333,
+      "step": 1905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.552932592297537e-06,
+      "loss": 2.2472,
+      "step": 1906
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.549767392721595e-06,
+      "loss": 2.2687,
+      "step": 1907
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.546600811643816e-06,
+      "loss": 2.2459,
+      "step": 1908
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.543432850779897e-06,
+      "loss": 2.3596,
+      "step": 1909
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.540263511846291e-06,
+      "loss": 2.2939,
+      "step": 1910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.537092796560188e-06,
+      "loss": 2.2798,
+      "step": 1911
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.533920706639531e-06,
+      "loss": 2.3245,
+      "step": 1912
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.530747243803006e-06,
+      "loss": 2.2571,
+      "step": 1913
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.5275724097700435e-06,
+      "loss": 2.2171,
+      "step": 1914
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.524396206260811e-06,
+      "loss": 2.3388,
+      "step": 1915
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.521218634996226e-06,
+      "loss": 2.3513,
+      "step": 1916
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.518039697697945e-06,
+      "loss": 2.1597,
+      "step": 1917
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.514859396088361e-06,
+      "loss": 2.334,
+      "step": 1918
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.5116777318906105e-06,
+      "loss": 2.243,
+      "step": 1919
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.508494706828564e-06,
+      "loss": 2.1989,
+      "step": 1920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.505310322626836e-06,
+      "loss": 2.3844,
+      "step": 1921
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.502124581010766e-06,
+      "loss": 2.2325,
+      "step": 1922
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.498937483706443e-06,
+      "loss": 2.2538,
+      "step": 1923
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.495749032440681e-06,
+      "loss": 2.3273,
+      "step": 1924
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.49255922894103e-06,
+      "loss": 2.2646,
+      "step": 1925
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.489368074935773e-06,
+      "loss": 2.3451,
+      "step": 1926
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.486175572153926e-06,
+      "loss": 2.2869,
+      "step": 1927
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.4829817223252324e-06,
+      "loss": 2.2857,
+      "step": 1928
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.479786527180168e-06,
+      "loss": 2.3149,
+      "step": 1929
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.476589988449939e-06,
+      "loss": 2.3404,
+      "step": 1930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 7.473392107866474e-06,
+      "loss": 2.3806,
+      "step": 1931
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.470192887162435e-06,
+      "loss": 2.3249,
+      "step": 1932
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.466992328071207e-06,
+      "loss": 2.2616,
+      "step": 1933
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.463790432326898e-06,
+      "loss": 2.394,
+      "step": 1934
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.460587201664346e-06,
+      "loss": 2.3028,
+      "step": 1935
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.4573826378191075e-06,
+      "loss": 2.2269,
+      "step": 1936
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.454176742527462e-06,
+      "loss": 2.3517,
+      "step": 1937
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.450969517526412e-06,
+      "loss": 2.3195,
+      "step": 1938
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.4477609645536795e-06,
+      "loss": 2.3413,
+      "step": 1939
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.444551085347708e-06,
+      "loss": 2.2547,
+      "step": 1940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.441339881647654e-06,
+      "loss": 2.3574,
+      "step": 1941
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.4381273551933965e-06,
+      "loss": 2.2803,
+      "step": 1942
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.434913507725532e-06,
+      "loss": 2.3048,
+      "step": 1943
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.431698340985369e-06,
+      "loss": 2.293,
+      "step": 1944
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.428481856714934e-06,
+      "loss": 2.2169,
+      "step": 1945
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.4252640566569655e-06,
+      "loss": 2.261,
+      "step": 1946
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.422044942554913e-06,
+      "loss": 2.3062,
+      "step": 1947
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.418824516152942e-06,
+      "loss": 2.2521,
+      "step": 1948
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.415602779195932e-06,
+      "loss": 2.3229,
+      "step": 1949
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.41237973342946e-06,
+      "loss": 2.2916,
+      "step": 1950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.409155380599824e-06,
+      "loss": 2.2924,
+      "step": 1951
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.405929722454026e-06,
+      "loss": 2.2954,
+      "step": 1952
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.402702760739775e-06,
+      "loss": 2.246,
+      "step": 1953
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.399474497205489e-06,
+      "loss": 2.2951,
+      "step": 1954
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.396244933600285e-06,
+      "loss": 2.2201,
+      "step": 1955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.393014071673992e-06,
+      "loss": 2.4037,
+      "step": 1956
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.3897819131771365e-06,
+      "loss": 2.2104,
+      "step": 1957
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.386548459860952e-06,
+      "loss": 2.3252,
+      "step": 1958
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.383313713477371e-06,
+      "loss": 2.3467,
+      "step": 1959
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.3800776757790275e-06,
+      "loss": 2.1773,
+      "step": 1960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.376840348519253e-06,
+      "loss": 2.2739,
+      "step": 1961
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.373601733452081e-06,
+      "loss": 2.3233,
+      "step": 1962
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.370361832332241e-06,
+      "loss": 2.1815,
+      "step": 1963
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.3671206469151605e-06,
+      "loss": 2.2591,
+      "step": 1964
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.363878178956959e-06,
+      "loss": 2.3153,
+      "step": 1965
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.360634430214456e-06,
+      "loss": 2.3169,
+      "step": 1966
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.357389402445163e-06,
+      "loss": 2.336,
+      "step": 1967
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.354143097407283e-06,
+      "loss": 2.243,
+      "step": 1968
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.350895516859713e-06,
+      "loss": 2.2705,
+      "step": 1969
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.34764666256204e-06,
+      "loss": 2.4319,
+      "step": 1970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.3443965362745425e-06,
+      "loss": 2.2811,
+      "step": 1971
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.341145139758185e-06,
+      "loss": 2.2997,
+      "step": 1972
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.337892474774625e-06,
+      "loss": 2.3856,
+      "step": 1973
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.334638543086203e-06,
+      "loss": 2.2449,
+      "step": 1974
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.3313833464559495e-06,
+      "loss": 2.4458,
+      "step": 1975
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.328126886647575e-06,
+      "loss": 2.3083,
+      "step": 1976
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.324869165425481e-06,
+      "loss": 2.2943,
+      "step": 1977
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.321610184554747e-06,
+      "loss": 2.3393,
+      "step": 1978
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 7.318349945801138e-06,
+      "loss": 2.2822,
+      "step": 1979
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.3150884509311025e-06,
+      "loss": 2.3196,
+      "step": 1980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.311825701711761e-06,
+      "loss": 2.3509,
+      "step": 1981
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.308561699910923e-06,
+      "loss": 2.1981,
+      "step": 1982
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.305296447297075e-06,
+      "loss": 2.265,
+      "step": 1983
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.302029945639377e-06,
+      "loss": 2.3062,
+      "step": 1984
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.298762196707668e-06,
+      "loss": 2.3455,
+      "step": 1985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.295493202272462e-06,
+      "loss": 2.3966,
+      "step": 1986
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.292222964104952e-06,
+      "loss": 2.3254,
+      "step": 1987
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.288951483976998e-06,
+      "loss": 2.2466,
+      "step": 1988
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.285678763661139e-06,
+      "loss": 2.3995,
+      "step": 1989
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.282404804930583e-06,
+      "loss": 2.2505,
+      "step": 1990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.279129609559208e-06,
+      "loss": 2.245,
+      "step": 1991
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.275853179321565e-06,
+      "loss": 2.2154,
+      "step": 1992
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.2725755159928715e-06,
+      "loss": 2.1794,
+      "step": 1993
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.269296621349016e-06,
+      "loss": 2.4047,
+      "step": 1994
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.26601649716655e-06,
+      "loss": 2.3329,
+      "step": 1995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.262735145222696e-06,
+      "loss": 2.3302,
+      "step": 1996
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.259452567295335e-06,
+      "loss": 2.2233,
+      "step": 1997
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.256168765163021e-06,
+      "loss": 2.2989,
+      "step": 1998
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.252883740604966e-06,
+      "loss": 2.2725,
+      "step": 1999
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.249597495401044e-06,
+      "loss": 2.2677,
+      "step": 2000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.2463100313317905e-06,
+      "loss": 2.3123,
+      "step": 2001
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.243021350178403e-06,
+      "loss": 2.2966,
+      "step": 2002
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.239731453722739e-06,
+      "loss": 2.326,
+      "step": 2003
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.236440343747313e-06,
+      "loss": 2.2815,
+      "step": 2004
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.233148022035297e-06,
+      "loss": 2.3572,
+      "step": 2005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.229854490370518e-06,
+      "loss": 2.3432,
+      "step": 2006
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.226559750537462e-06,
+      "loss": 2.2319,
+      "step": 2007
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.223263804321269e-06,
+      "loss": 2.2718,
+      "step": 2008
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.219966653507728e-06,
+      "loss": 2.357,
+      "step": 2009
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.21666829988329e-06,
+      "loss": 2.3287,
+      "step": 2010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.213368745235047e-06,
+      "loss": 2.2915,
+      "step": 2011
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.21006799135075e-06,
+      "loss": 2.1565,
+      "step": 2012
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.206766040018792e-06,
+      "loss": 2.2859,
+      "step": 2013
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.203462893028224e-06,
+      "loss": 2.3878,
+      "step": 2014
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.200158552168739e-06,
+      "loss": 2.3557,
+      "step": 2015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.196853019230676e-06,
+      "loss": 2.205,
+      "step": 2016
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.193546296005023e-06,
+      "loss": 2.2214,
+      "step": 2017
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.190238384283413e-06,
+      "loss": 2.3371,
+      "step": 2018
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.186929285858119e-06,
+      "loss": 2.2831,
+      "step": 2019
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.183619002522062e-06,
+      "loss": 2.3087,
+      "step": 2020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.180307536068803e-06,
+      "loss": 2.1996,
+      "step": 2021
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.176994888292539e-06,
+      "loss": 2.2917,
+      "step": 2022
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.173681060988117e-06,
+      "loss": 2.2272,
+      "step": 2023
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.170366055951017e-06,
+      "loss": 2.3204,
+      "step": 2024
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.167049874977356e-06,
+      "loss": 2.3047,
+      "step": 2025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 7.163732519863893e-06,
+      "loss": 2.3091,
+      "step": 2026
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.160413992408017e-06,
+      "loss": 2.3469,
+      "step": 2027
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.157094294407757e-06,
+      "loss": 2.2664,
+      "step": 2028
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.1537734276617744e-06,
+      "loss": 2.4448,
+      "step": 2029
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.1504513939693646e-06,
+      "loss": 2.213,
+      "step": 2030
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.147128195130455e-06,
+      "loss": 2.3229,
+      "step": 2031
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.143803832945602e-06,
+      "loss": 2.2522,
+      "step": 2032
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.1404783092159964e-06,
+      "loss": 2.3757,
+      "step": 2033
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.137151625743457e-06,
+      "loss": 2.3526,
+      "step": 2034
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.1338237843304285e-06,
+      "loss": 2.3023,
+      "step": 2035
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.130494786779987e-06,
+      "loss": 2.3289,
+      "step": 2036
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.127164634895829e-06,
+      "loss": 2.2313,
+      "step": 2037
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.1238333304822835e-06,
+      "loss": 2.3648,
+      "step": 2038
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.120500875344299e-06,
+      "loss": 2.2518,
+      "step": 2039
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.117167271287453e-06,
+      "loss": 2.2605,
+      "step": 2040
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.113832520117935e-06,
+      "loss": 2.3191,
+      "step": 2041
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.110496623642569e-06,
+      "loss": 2.2611,
+      "step": 2042
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.1071595836687875e-06,
+      "loss": 2.3368,
+      "step": 2043
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.103821402004654e-06,
+      "loss": 2.2279,
+      "step": 2044
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.100482080458841e-06,
+      "loss": 2.326,
+      "step": 2045
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.097141620840646e-06,
+      "loss": 2.2892,
+      "step": 2046
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0938000249599745e-06,
+      "loss": 2.291,
+      "step": 2047
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.090457294627358e-06,
+      "loss": 2.3159,
+      "step": 2048
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0871134316539336e-06,
+      "loss": 2.3575,
+      "step": 2049
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.083768437851458e-06,
+      "loss": 2.319,
+      "step": 2050
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.080422315032298e-06,
+      "loss": 2.2372,
+      "step": 2051
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0770750650094335e-06,
+      "loss": 2.3058,
+      "step": 2052
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.073726689596452e-06,
+      "loss": 2.3104,
+      "step": 2053
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0703771906075544e-06,
+      "loss": 2.2528,
+      "step": 2054
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.06702656985755e-06,
+      "loss": 2.266,
+      "step": 2055
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.063674829161854e-06,
+      "loss": 2.3051,
+      "step": 2056
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.060321970336488e-06,
+      "loss": 2.3287,
+      "step": 2057
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.056967995198083e-06,
+      "loss": 2.2819,
+      "step": 2058
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.053612905563869e-06,
+      "loss": 2.2326,
+      "step": 2059
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0502567032516885e-06,
+      "loss": 2.2718,
+      "step": 2060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0468993900799785e-06,
+      "loss": 2.2538,
+      "step": 2061
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.043540967867782e-06,
+      "loss": 2.3503,
+      "step": 2062
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.040181438434741e-06,
+      "loss": 2.2962,
+      "step": 2063
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.036820803601099e-06,
+      "loss": 2.3166,
+      "step": 2064
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.033459065187698e-06,
+      "loss": 2.3304,
+      "step": 2065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0300962250159784e-06,
+      "loss": 2.3638,
+      "step": 2066
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.026732284907976e-06,
+      "loss": 2.3355,
+      "step": 2067
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.023367246686323e-06,
+      "loss": 2.2801,
+      "step": 2068
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.0200011121742494e-06,
+      "loss": 2.3325,
+      "step": 2069
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.016633883195574e-06,
+      "loss": 2.4091,
+      "step": 2070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.013265561574716e-06,
+      "loss": 2.2725,
+      "step": 2071
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.009896149136675e-06,
+      "loss": 2.2963,
+      "step": 2072
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.006525647707054e-06,
+      "loss": 2.3321,
+      "step": 2073
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 7.003154059112039e-06,
+      "loss": 2.2467,
+      "step": 2074
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.999781385178409e-06,
+      "loss": 2.2979,
+      "step": 2075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.996407627733526e-06,
+      "loss": 2.3202,
+      "step": 2076
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.993032788605342e-06,
+      "loss": 2.3147,
+      "step": 2077
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.989656869622397e-06,
+      "loss": 2.3302,
+      "step": 2078
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.98627987261381e-06,
+      "loss": 2.2701,
+      "step": 2079
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.982901799409294e-06,
+      "loss": 2.1644,
+      "step": 2080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.979522651839136e-06,
+      "loss": 2.3341,
+      "step": 2081
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.976142431734208e-06,
+      "loss": 2.234,
+      "step": 2082
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.972761140925963e-06,
+      "loss": 2.2304,
+      "step": 2083
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.969378781246436e-06,
+      "loss": 2.2252,
+      "step": 2084
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.96599535452824e-06,
+      "loss": 2.2079,
+      "step": 2085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.962610862604565e-06,
+      "loss": 2.3386,
+      "step": 2086
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.9592253073091795e-06,
+      "loss": 2.2823,
+      "step": 2087
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.955838690476426e-06,
+      "loss": 2.2765,
+      "step": 2088
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.952451013941225e-06,
+      "loss": 2.2318,
+      "step": 2089
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.9490622795390706e-06,
+      "loss": 2.2943,
+      "step": 2090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.945672489106029e-06,
+      "loss": 2.3001,
+      "step": 2091
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.942281644478739e-06,
+      "loss": 2.2355,
+      "step": 2092
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.93888974749441e-06,
+      "loss": 2.1719,
+      "step": 2093
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.935496799990822e-06,
+      "loss": 2.2753,
+      "step": 2094
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.932102803806324e-06,
+      "loss": 2.364,
+      "step": 2095
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.928707760779838e-06,
+      "loss": 2.2894,
+      "step": 2096
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.925311672750842e-06,
+      "loss": 2.1996,
+      "step": 2097
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.921914541559391e-06,
+      "loss": 2.2922,
+      "step": 2098
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.9185163690461e-06,
+      "loss": 2.3452,
+      "step": 2099
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.915117157052149e-06,
+      "loss": 2.3365,
+      "step": 2100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.911716907419284e-06,
+      "loss": 2.3487,
+      "step": 2101
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.9083156219898065e-06,
+      "loss": 2.3193,
+      "step": 2102
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.9049133026065865e-06,
+      "loss": 2.3322,
+      "step": 2103
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.90150995111305e-06,
+      "loss": 2.2742,
+      "step": 2104
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.898105569353184e-06,
+      "loss": 2.317,
+      "step": 2105
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.894700159171535e-06,
+      "loss": 2.3253,
+      "step": 2106
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.891293722413201e-06,
+      "loss": 2.2502,
+      "step": 2107
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.887886260923841e-06,
+      "loss": 2.3067,
+      "step": 2108
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.88447777654967e-06,
+      "loss": 2.2478,
+      "step": 2109
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.881068271137455e-06,
+      "loss": 2.2933,
+      "step": 2110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.877657746534514e-06,
+      "loss": 2.2274,
+      "step": 2111
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.874246204588724e-06,
+      "loss": 2.2752,
+      "step": 2112
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.8708336471485045e-06,
+      "loss": 2.2744,
+      "step": 2113
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.867420076062834e-06,
+      "loss": 2.3217,
+      "step": 2114
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.864005493181235e-06,
+      "loss": 2.331,
+      "step": 2115
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.860589900353778e-06,
+      "loss": 2.3605,
+      "step": 2116
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.857173299431084e-06,
+      "loss": 2.2554,
+      "step": 2117
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.853755692264313e-06,
+      "loss": 2.1988,
+      "step": 2118
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.850337080705181e-06,
+      "loss": 2.307,
+      "step": 2119
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.84691746660594e-06,
+      "loss": 2.3117,
+      "step": 2120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.843496851819389e-06,
+      "loss": 2.3574,
+      "step": 2121
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.840075238198867e-06,
+      "loss": 2.255,
+      "step": 2122
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.836652627598255e-06,
+      "loss": 2.3132,
+      "step": 2123
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.833229021871974e-06,
+      "loss": 2.2256,
+      "step": 2124
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.829804422874984e-06,
+      "loss": 2.348,
+      "step": 2125
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.826378832462786e-06,
+      "loss": 2.3334,
+      "step": 2126
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.822952252491413e-06,
+      "loss": 2.4063,
+      "step": 2127
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.819524684817439e-06,
+      "loss": 2.2793,
+      "step": 2128
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.816096131297969e-06,
+      "loss": 2.2865,
+      "step": 2129
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.8126665937906445e-06,
+      "loss": 2.278,
+      "step": 2130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.8092360741536435e-06,
+      "loss": 2.2875,
+      "step": 2131
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.805804574245667e-06,
+      "loss": 2.2699,
+      "step": 2132
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.8023720959259565e-06,
+      "loss": 2.2376,
+      "step": 2133
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.798938641054278e-06,
+      "loss": 2.2948,
+      "step": 2134
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.795504211490929e-06,
+      "loss": 2.2686,
+      "step": 2135
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.792068809096735e-06,
+      "loss": 2.3763,
+      "step": 2136
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.788632435733047e-06,
+      "loss": 2.3165,
+      "step": 2137
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.785195093261741e-06,
+      "loss": 2.2659,
+      "step": 2138
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.781756783545225e-06,
+      "loss": 2.3434,
+      "step": 2139
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.778317508446424e-06,
+      "loss": 2.2731,
+      "step": 2140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.774877269828788e-06,
+      "loss": 2.2307,
+      "step": 2141
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.77143606955629e-06,
+      "loss": 2.3113,
+      "step": 2142
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.767993909493421e-06,
+      "loss": 2.2742,
+      "step": 2143
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.764550791505198e-06,
+      "loss": 2.2486,
+      "step": 2144
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.761106717457151e-06,
+      "loss": 2.266,
+      "step": 2145
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.757661689215333e-06,
+      "loss": 2.2905,
+      "step": 2146
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.754215708646308e-06,
+      "loss": 2.33,
+      "step": 2147
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.750768777617161e-06,
+      "loss": 2.2025,
+      "step": 2148
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.747320897995493e-06,
+      "loss": 2.2617,
+      "step": 2149
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.743872071649411e-06,
+      "loss": 2.2583,
+      "step": 2150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.740422300447547e-06,
+      "loss": 2.3009,
+      "step": 2151
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.736971586259032e-06,
+      "loss": 2.3306,
+      "step": 2152
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.733519930953518e-06,
+      "loss": 2.3163,
+      "step": 2153
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.73006733640116e-06,
+      "loss": 2.3539,
+      "step": 2154
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.726613804472628e-06,
+      "loss": 2.2594,
+      "step": 2155
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.723159337039097e-06,
+      "loss": 2.3056,
+      "step": 2156
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.719703935972247e-06,
+      "loss": 2.3448,
+      "step": 2157
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.716247603144267e-06,
+      "loss": 2.232,
+      "step": 2158
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.712790340427847e-06,
+      "loss": 2.2493,
+      "step": 2159
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.7093321496961865e-06,
+      "loss": 2.2816,
+      "step": 2160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.705873032822981e-06,
+      "loss": 2.282,
+      "step": 2161
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.702412991682436e-06,
+      "loss": 2.2574,
+      "step": 2162
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.698952028149248e-06,
+      "loss": 2.3196,
+      "step": 2163
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.695490144098622e-06,
+      "loss": 2.3369,
+      "step": 2164
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.692027341406258e-06,
+      "loss": 2.321,
+      "step": 2165
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.688563621948353e-06,
+      "loss": 2.3608,
+      "step": 2166
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.685098987601601e-06,
+      "loss": 2.318,
+      "step": 2167
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.681633440243194e-06,
+      "loss": 2.3308,
+      "step": 2168
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.678166981750815e-06,
+      "loss": 2.3658,
+      "step": 2169
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.674699614002644e-06,
+      "loss": 2.3888,
+      "step": 2170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.671231338877353e-06,
+      "loss": 2.2603,
+      "step": 2171
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.667762158254104e-06,
+      "loss": 2.3254,
+      "step": 2172
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.66429207401255e-06,
+      "loss": 2.2979,
+      "step": 2173
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.660821088032836e-06,
+      "loss": 2.2886,
+      "step": 2174
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.657349202195591e-06,
+      "loss": 2.1994,
+      "step": 2175
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.653876418381937e-06,
+      "loss": 2.2524,
+      "step": 2176
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.650402738473481e-06,
+      "loss": 2.2828,
+      "step": 2177
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.646928164352309e-06,
+      "loss": 2.3358,
+      "step": 2178
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.643452697901001e-06,
+      "loss": 2.2336,
+      "step": 2179
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.639976341002614e-06,
+      "loss": 2.2149,
+      "step": 2180
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.636499095540693e-06,
+      "loss": 2.3199,
+      "step": 2181
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.633020963399259e-06,
+      "loss": 2.2866,
+      "step": 2182
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.629541946462816e-06,
+      "loss": 2.3342,
+      "step": 2183
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.626062046616346e-06,
+      "loss": 2.3191,
+      "step": 2184
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.622581265745314e-06,
+      "loss": 2.2501,
+      "step": 2185
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.619099605735652e-06,
+      "loss": 2.2456,
+      "step": 2186
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.615617068473786e-06,
+      "loss": 2.3378,
+      "step": 2187
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.612133655846593e-06,
+      "loss": 2.2818,
+      "step": 2188
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.6086493697414464e-06,
+      "loss": 2.3026,
+      "step": 2189
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.6051642120461846e-06,
+      "loss": 2.2859,
+      "step": 2190
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.601678184649112e-06,
+      "loss": 2.3524,
+      "step": 2191
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.598191289439016e-06,
+      "loss": 2.2544,
+      "step": 2192
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.594703528305145e-06,
+      "loss": 2.2893,
+      "step": 2193
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.591214903137221e-06,
+      "loss": 2.2434,
+      "step": 2194
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.58772541582543e-06,
+      "loss": 2.2809,
+      "step": 2195
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.5842350682604314e-06,
+      "loss": 2.3384,
+      "step": 2196
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.580743862333348e-06,
+      "loss": 2.3111,
+      "step": 2197
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.577251799935763e-06,
+      "loss": 2.2383,
+      "step": 2198
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.573758882959731e-06,
+      "loss": 2.304,
+      "step": 2199
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.570265113297765e-06,
+      "loss": 2.1944,
+      "step": 2200
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.566770492842842e-06,
+      "loss": 2.3254,
+      "step": 2201
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.563275023488399e-06,
+      "loss": 2.3088,
+      "step": 2202
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.5597787071283325e-06,
+      "loss": 2.3414,
+      "step": 2203
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.556281545656999e-06,
+      "loss": 2.3155,
+      "step": 2204
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.552783540969212e-06,
+      "loss": 2.2126,
+      "step": 2205
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.549284694960244e-06,
+      "loss": 2.3231,
+      "step": 2206
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.545785009525821e-06,
+      "loss": 2.3574,
+      "step": 2207
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.542284486562125e-06,
+      "loss": 2.238,
+      "step": 2208
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.538783127965789e-06,
+      "loss": 2.3125,
+      "step": 2209
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.535280935633903e-06,
+      "loss": 2.242,
+      "step": 2210
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.531777911464007e-06,
+      "loss": 2.3454,
+      "step": 2211
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.528274057354091e-06,
+      "loss": 2.3748,
+      "step": 2212
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.524769375202595e-06,
+      "loss": 2.3711,
+      "step": 2213
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.521263866908409e-06,
+      "loss": 2.2692,
+      "step": 2214
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.5177575343708674e-06,
+      "loss": 2.3286,
+      "step": 2215
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.514250379489754e-06,
+      "loss": 2.2849,
+      "step": 2216
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.510742404165299e-06,
+      "loss": 2.2788,
+      "step": 2217
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.507233610298174e-06,
+      "loss": 2.1983,
+      "step": 2218
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.503723999789496e-06,
+      "loss": 2.2153,
+      "step": 2219
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.500213574540823e-06,
+      "loss": 2.3177,
+      "step": 2220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.496702336454157e-06,
+      "loss": 2.3456,
+      "step": 2221
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.493190287431941e-06,
+      "loss": 2.329,
+      "step": 2222
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.489677429377049e-06,
+      "loss": 2.2437,
+      "step": 2223
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.486163764192805e-06,
+      "loss": 2.3722,
+      "step": 2224
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.482649293782963e-06,
+      "loss": 2.2486,
+      "step": 2225
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.479134020051717e-06,
+      "loss": 2.3327,
+      "step": 2226
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.475617944903693e-06,
+      "loss": 2.3435,
+      "step": 2227
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.472101070243952e-06,
+      "loss": 2.2893,
+      "step": 2228
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.46858339797799e-06,
+      "loss": 2.2279,
+      "step": 2229
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.465064930011733e-06,
+      "loss": 2.2268,
+      "step": 2230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.4615456682515425e-06,
+      "loss": 2.2748,
+      "step": 2231
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.458025614604203e-06,
+      "loss": 2.3347,
+      "step": 2232
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.454504770976934e-06,
+      "loss": 2.2016,
+      "step": 2233
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.450983139277378e-06,
+      "loss": 2.3182,
+      "step": 2234
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.447460721413612e-06,
+      "loss": 2.1879,
+      "step": 2235
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.44393751929413e-06,
+      "loss": 2.2899,
+      "step": 2236
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.440413534827857e-06,
+      "loss": 2.1342,
+      "step": 2237
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.436888769924142e-06,
+      "loss": 2.2715,
+      "step": 2238
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.433363226492752e-06,
+      "loss": 2.334,
+      "step": 2239
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.42983690644388e-06,
+      "loss": 2.2984,
+      "step": 2240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.426309811688136e-06,
+      "loss": 2.3039,
+      "step": 2241
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.422781944136558e-06,
+      "loss": 2.3825,
+      "step": 2242
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.419253305700594e-06,
+      "loss": 2.316,
+      "step": 2243
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.415723898292111e-06,
+      "loss": 2.1997,
+      "step": 2244
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.412193723823395e-06,
+      "loss": 2.2975,
+      "step": 2245
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.408662784207149e-06,
+      "loss": 2.3531,
+      "step": 2246
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.405131081356487e-06,
+      "loss": 2.2264,
+      "step": 2247
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.401598617184939e-06,
+      "loss": 2.2197,
+      "step": 2248
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.398065393606445e-06,
+      "loss": 2.3088,
+      "step": 2249
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.39453141253536e-06,
+      "loss": 2.208,
+      "step": 2250
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.390996675886445e-06,
+      "loss": 2.2532,
+      "step": 2251
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.387461185574875e-06,
+      "loss": 2.3115,
+      "step": 2252
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.3839249435162305e-06,
+      "loss": 2.3188,
+      "step": 2253
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.3803879516265e-06,
+      "loss": 2.3605,
+      "step": 2254
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.376850211822077e-06,
+      "loss": 2.3206,
+      "step": 2255
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.373311726019763e-06,
+      "loss": 2.2563,
+      "step": 2256
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.369772496136762e-06,
+      "loss": 2.3413,
+      "step": 2257
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.366232524090681e-06,
+      "loss": 2.2139,
+      "step": 2258
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.362691811799527e-06,
+      "loss": 2.2825,
+      "step": 2259
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.3591503611817155e-06,
+      "loss": 2.348,
+      "step": 2260
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.355608174156053e-06,
+      "loss": 2.2988,
+      "step": 2261
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.3520652526417506e-06,
+      "loss": 2.2947,
+      "step": 2262
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.348521598558417e-06,
+      "loss": 2.2833,
+      "step": 2263
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.344977213826053e-06,
+      "loss": 2.35,
+      "step": 2264
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 6.341432100365063e-06,
+      "loss": 2.32,
+      "step": 2265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.337886260096239e-06,
+      "loss": 2.267,
+      "step": 2266
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.334339694940772e-06,
+      "loss": 2.2698,
+      "step": 2267
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.330792406820242e-06,
+      "loss": 2.3423,
+      "step": 2268
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.327244397656624e-06,
+      "loss": 2.3004,
+      "step": 2269
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.323695669372281e-06,
+      "loss": 2.36,
+      "step": 2270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.320146223889966e-06,
+      "loss": 2.2495,
+      "step": 2271
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.316596063132823e-06,
+      "loss": 2.2443,
+      "step": 2272
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.313045189024382e-06,
+      "loss": 2.2842,
+      "step": 2273
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.309493603488557e-06,
+      "loss": 2.212,
+      "step": 2274
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.30594130844965e-06,
+      "loss": 2.3522,
+      "step": 2275
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.302388305832351e-06,
+      "loss": 2.2281,
+      "step": 2276
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.298834597561725e-06,
+      "loss": 2.2339,
+      "step": 2277
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.295280185563226e-06,
+      "loss": 2.3272,
+      "step": 2278
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.291725071762689e-06,
+      "loss": 2.2691,
+      "step": 2279
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.288169258086323e-06,
+      "loss": 2.3277,
+      "step": 2280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.284612746460724e-06,
+      "loss": 2.2013,
+      "step": 2281
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.281055538812861e-06,
+      "loss": 2.2341,
+      "step": 2282
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.277497637070087e-06,
+      "loss": 2.2385,
+      "step": 2283
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.273939043160118e-06,
+      "loss": 2.3941,
+      "step": 2284
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.270379759011058e-06,
+      "loss": 2.2473,
+      "step": 2285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.266819786551378e-06,
+      "loss": 2.3609,
+      "step": 2286
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.263259127709924e-06,
+      "loss": 2.281,
+      "step": 2287
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.259697784415919e-06,
+      "loss": 2.3355,
+      "step": 2288
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.256135758598943e-06,
+      "loss": 2.2434,
+      "step": 2289
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.252573052188961e-06,
+      "loss": 2.146,
+      "step": 2290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.249009667116297e-06,
+      "loss": 2.3426,
+      "step": 2291
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.2454456053116485e-06,
+      "loss": 2.2786,
+      "step": 2292
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.241880868706075e-06,
+      "loss": 2.2686,
+      "step": 2293
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.238315459231005e-06,
+      "loss": 2.2872,
+      "step": 2294
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.234749378818228e-06,
+      "loss": 2.2858,
+      "step": 2295
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.231182629399901e-06,
+      "loss": 2.2988,
+      "step": 2296
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.2276152129085445e-06,
+      "loss": 2.247,
+      "step": 2297
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.224047131277033e-06,
+      "loss": 2.2533,
+      "step": 2298
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.22047838643861e-06,
+      "loss": 2.346,
+      "step": 2299
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.21690898032687e-06,
+      "loss": 2.3372,
+      "step": 2300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.213338914875775e-06,
+      "loss": 2.3126,
+      "step": 2301
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.209768192019635e-06,
+      "loss": 2.3137,
+      "step": 2302
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.206196813693125e-06,
+      "loss": 2.248,
+      "step": 2303
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.202624781831269e-06,
+      "loss": 2.2415,
+      "step": 2304
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.1990520983694454e-06,
+      "loss": 2.2661,
+      "step": 2305
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.19547876524339e-06,
+      "loss": 2.3738,
+      "step": 2306
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.191904784389184e-06,
+      "loss": 2.2423,
+      "step": 2307
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.188330157743268e-06,
+      "loss": 2.3063,
+      "step": 2308
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.184754887242424e-06,
+      "loss": 2.2933,
+      "step": 2309
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.181178974823789e-06,
+      "loss": 2.2832,
+      "step": 2310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.177602422424844e-06,
+      "loss": 2.2122,
+      "step": 2311
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 6.174025231983417e-06,
+      "loss": 2.2538,
+      "step": 2312
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.170447405437688e-06,
+      "loss": 2.3363,
+      "step": 2313
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.166868944726171e-06,
+      "loss": 2.2846,
+      "step": 2314
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.163289851787732e-06,
+      "loss": 2.2223,
+      "step": 2315
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.159710128561574e-06,
+      "loss": 2.1964,
+      "step": 2316
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.156129776987249e-06,
+      "loss": 2.2682,
+      "step": 2317
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.1525487990046405e-06,
+      "loss": 2.2834,
+      "step": 2318
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.148967196553976e-06,
+      "loss": 2.2806,
+      "step": 2319
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.145384971575823e-06,
+      "loss": 2.2236,
+      "step": 2320
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.141802126011081e-06,
+      "loss": 2.249,
+      "step": 2321
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.138218661800994e-06,
+      "loss": 2.2585,
+      "step": 2322
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.1346345808871325e-06,
+      "loss": 2.3297,
+      "step": 2323
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.131049885211404e-06,
+      "loss": 2.3494,
+      "step": 2324
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.127464576716052e-06,
+      "loss": 2.3033,
+      "step": 2325
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.123878657343648e-06,
+      "loss": 2.1956,
+      "step": 2326
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.120292129037098e-06,
+      "loss": 2.3841,
+      "step": 2327
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.116704993739635e-06,
+      "loss": 2.2815,
+      "step": 2328
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.113117253394825e-06,
+      "loss": 2.332,
+      "step": 2329
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.109528909946554e-06,
+      "loss": 2.3301,
+      "step": 2330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.105939965339045e-06,
+      "loss": 2.3387,
+      "step": 2331
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.102350421516837e-06,
+      "loss": 2.3281,
+      "step": 2332
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.098760280424801e-06,
+      "loss": 2.3272,
+      "step": 2333
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.095169544008129e-06,
+      "loss": 2.3195,
+      "step": 2334
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.0915782142123344e-06,
+      "loss": 2.2543,
+      "step": 2335
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.087986292983253e-06,
+      "loss": 2.282,
+      "step": 2336
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.084393782267039e-06,
+      "loss": 2.2416,
+      "step": 2337
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.080800684010172e-06,
+      "loss": 2.2745,
+      "step": 2338
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.077207000159444e-06,
+      "loss": 2.3139,
+      "step": 2339
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.073612732661966e-06,
+      "loss": 2.3022,
+      "step": 2340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.070017883465166e-06,
+      "loss": 2.2669,
+      "step": 2341
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.066422454516786e-06,
+      "loss": 2.2993,
+      "step": 2342
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.062826447764883e-06,
+      "loss": 2.2739,
+      "step": 2343
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.05922986515783e-06,
+      "loss": 2.3034,
+      "step": 2344
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.055632708644304e-06,
+      "loss": 2.1791,
+      "step": 2345
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.052034980173298e-06,
+      "loss": 2.3201,
+      "step": 2346
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.04843668169412e-06,
+      "loss": 2.2874,
+      "step": 2347
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.044837815156377e-06,
+      "loss": 2.2547,
+      "step": 2348
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.041238382509989e-06,
+      "loss": 2.3946,
+      "step": 2349
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.037638385705182e-06,
+      "loss": 2.2955,
+      "step": 2350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.034037826692489e-06,
+      "loss": 2.2728,
+      "step": 2351
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.030436707422746e-06,
+      "loss": 2.3129,
+      "step": 2352
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.026835029847093e-06,
+      "loss": 2.2566,
+      "step": 2353
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.02323279591697e-06,
+      "loss": 2.3066,
+      "step": 2354
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.019630007584123e-06,
+      "loss": 2.2335,
+      "step": 2355
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.016026666800597e-06,
+      "loss": 2.2955,
+      "step": 2356
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.0124227755187335e-06,
+      "loss": 2.231,
+      "step": 2357
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.008818335691178e-06,
+      "loss": 2.3059,
+      "step": 2358
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.005213349270864e-06,
+      "loss": 2.3042,
+      "step": 2359
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 6.001607818211032e-06,
+      "loss": 2.3006,
+      "step": 2360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.998001744465207e-06,
+      "loss": 2.2765,
+      "step": 2361
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.994395129987217e-06,
+      "loss": 2.2532,
+      "step": 2362
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.990787976731179e-06,
+      "loss": 2.289,
+      "step": 2363
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.987180286651503e-06,
+      "loss": 2.3031,
+      "step": 2364
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.983572061702886e-06,
+      "loss": 2.3336,
+      "step": 2365
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.979963303840319e-06,
+      "loss": 2.3507,
+      "step": 2366
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.976354015019083e-06,
+      "loss": 2.2161,
+      "step": 2367
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.972744197194739e-06,
+      "loss": 2.3216,
+      "step": 2368
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.969133852323147e-06,
+      "loss": 2.2547,
+      "step": 2369
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.965522982360441e-06,
+      "loss": 2.2797,
+      "step": 2370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.961911589263043e-06,
+      "loss": 2.2463,
+      "step": 2371
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.958299674987663e-06,
+      "loss": 2.1958,
+      "step": 2372
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.954687241491286e-06,
+      "loss": 2.3483,
+      "step": 2373
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.951074290731187e-06,
+      "loss": 2.2909,
+      "step": 2374
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.947460824664911e-06,
+      "loss": 2.3,
+      "step": 2375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.943846845250292e-06,
+      "loss": 2.3793,
+      "step": 2376
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.940232354445433e-06,
+      "loss": 2.3051,
+      "step": 2377
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.936617354208723e-06,
+      "loss": 2.3728,
+      "step": 2378
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.933001846498823e-06,
+      "loss": 2.1934,
+      "step": 2379
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.929385833274665e-06,
+      "loss": 2.3002,
+      "step": 2380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.925769316495461e-06,
+      "loss": 2.2859,
+      "step": 2381
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.922152298120694e-06,
+      "loss": 2.248,
+      "step": 2382
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.9185347801101165e-06,
+      "loss": 2.3613,
+      "step": 2383
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.914916764423756e-06,
+      "loss": 2.1931,
+      "step": 2384
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.911298253021904e-06,
+      "loss": 2.2978,
+      "step": 2385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.907679247865126e-06,
+      "loss": 2.3432,
+      "step": 2386
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.904059750914252e-06,
+      "loss": 2.2076,
+      "step": 2387
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.9004397641303805e-06,
+      "loss": 2.2914,
+      "step": 2388
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.896819289474872e-06,
+      "loss": 2.2497,
+      "step": 2389
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.8931983289093556e-06,
+      "loss": 2.3701,
+      "step": 2390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.889576884395718e-06,
+      "loss": 2.3245,
+      "step": 2391
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.885954957896115e-06,
+      "loss": 2.3598,
+      "step": 2392
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.8823325513729575e-06,
+      "loss": 2.241,
+      "step": 2393
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.878709666788923e-06,
+      "loss": 2.301,
+      "step": 2394
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.87508630610694e-06,
+      "loss": 2.2561,
+      "step": 2395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.871462471290202e-06,
+      "loss": 2.28,
+      "step": 2396
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.867838164302157e-06,
+      "loss": 2.3105,
+      "step": 2397
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.8642133871065035e-06,
+      "loss": 2.2782,
+      "step": 2398
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.860588141667205e-06,
+      "loss": 2.2603,
+      "step": 2399
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.856962429948472e-06,
+      "loss": 2.1824,
+      "step": 2400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.853336253914767e-06,
+      "loss": 2.3232,
+      "step": 2401
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.849709615530807e-06,
+      "loss": 2.2573,
+      "step": 2402
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.846082516761558e-06,
+      "loss": 2.3406,
+      "step": 2403
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.842454959572239e-06,
+      "loss": 2.318,
+      "step": 2404
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.8388269459283105e-06,
+      "loss": 2.3211,
+      "step": 2405
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.8351984777954875e-06,
+      "loss": 2.3036,
+      "step": 2406
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.831569557139725e-06,
+      "loss": 2.1965,
+      "step": 2407
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.827940185927227e-06,
+      "loss": 2.2706,
+      "step": 2408
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.824310366124442e-06,
+      "loss": 2.2895,
+      "step": 2409
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.820680099698058e-06,
+      "loss": 2.3549,
+      "step": 2410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.817049388615009e-06,
+      "loss": 2.2604,
+      "step": 2411
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.813418234842467e-06,
+      "loss": 2.3326,
+      "step": 2412
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.809786640347847e-06,
+      "loss": 2.3163,
+      "step": 2413
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.806154607098799e-06,
+      "loss": 2.3285,
+      "step": 2414
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.802522137063214e-06,
+      "loss": 2.3816,
+      "step": 2415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7988892322092175e-06,
+      "loss": 2.2253,
+      "step": 2416
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.795255894505171e-06,
+      "loss": 2.2826,
+      "step": 2417
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.791622125919669e-06,
+      "loss": 2.3238,
+      "step": 2418
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.787987928421546e-06,
+      "loss": 2.3487,
+      "step": 2419
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.78435330397986e-06,
+      "loss": 2.2813,
+      "step": 2420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.780718254563905e-06,
+      "loss": 2.367,
+      "step": 2421
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.777082782143208e-06,
+      "loss": 2.2947,
+      "step": 2422
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7734468886875175e-06,
+      "loss": 2.2092,
+      "step": 2423
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.769810576166818e-06,
+      "loss": 2.3149,
+      "step": 2424
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.766173846551316e-06,
+      "loss": 2.3045,
+      "step": 2425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.762536701811448e-06,
+      "loss": 2.3499,
+      "step": 2426
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.758899143917868e-06,
+      "loss": 2.3302,
+      "step": 2427
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.755261174841461e-06,
+      "loss": 2.3503,
+      "step": 2428
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.751622796553336e-06,
+      "loss": 2.2999,
+      "step": 2429
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.747984011024818e-06,
+      "loss": 2.2488,
+      "step": 2430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.744344820227454e-06,
+      "loss": 2.2078,
+      "step": 2431
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.740705226133013e-06,
+      "loss": 2.2805,
+      "step": 2432
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.737065230713481e-06,
+      "loss": 2.2989,
+      "step": 2433
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.733424835941059e-06,
+      "loss": 2.1926,
+      "step": 2434
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.729784043788174e-06,
+      "loss": 2.3111,
+      "step": 2435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.726142856227453e-06,
+      "loss": 2.2397,
+      "step": 2436
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7225012752317486e-06,
+      "loss": 2.3044,
+      "step": 2437
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.718859302774125e-06,
+      "loss": 2.2842,
+      "step": 2438
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.715216940827856e-06,
+      "loss": 2.282,
+      "step": 2439
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.711574191366427e-06,
+      "loss": 2.2788,
+      "step": 2440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.707931056363534e-06,
+      "loss": 2.2515,
+      "step": 2441
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.704287537793081e-06,
+      "loss": 2.3198,
+      "step": 2442
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.700643637629181e-06,
+      "loss": 2.4304,
+      "step": 2443
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.696999357846154e-06,
+      "loss": 2.2632,
+      "step": 2444
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.693354700418524e-06,
+      "loss": 2.2791,
+      "step": 2445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.689709667321021e-06,
+      "loss": 2.2868,
+      "step": 2446
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.686064260528577e-06,
+      "loss": 2.2569,
+      "step": 2447
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.682418482016329e-06,
+      "loss": 2.2626,
+      "step": 2448
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.6787723337596124e-06,
+      "loss": 2.3109,
+      "step": 2449
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.675125817733966e-06,
+      "loss": 2.2346,
+      "step": 2450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.671478935915127e-06,
+      "loss": 2.2277,
+      "step": 2451
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.667831690279027e-06,
+      "loss": 2.2902,
+      "step": 2452
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.664184082801799e-06,
+      "loss": 2.2629,
+      "step": 2453
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.660536115459773e-06,
+      "loss": 2.2492,
+      "step": 2454
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.6568877902294705e-06,
+      "loss": 2.3019,
+      "step": 2455
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.6532391090876084e-06,
+      "loss": 2.2892,
+      "step": 2456
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.649590074011094e-06,
+      "loss": 2.3232,
+      "step": 2457
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.645940686977033e-06,
+      "loss": 2.3123,
+      "step": 2458
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.642290949962714e-06,
+      "loss": 2.3151,
+      "step": 2459
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.6386408649456205e-06,
+      "loss": 2.2745,
+      "step": 2460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.634990433903423e-06,
+      "loss": 2.3105,
+      "step": 2461
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.631339658813977e-06,
+      "loss": 2.3219,
+      "step": 2462
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.627688541655329e-06,
+      "loss": 2.3508,
+      "step": 2463
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.624037084405708e-06,
+      "loss": 2.2657,
+      "step": 2464
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.62038528904353e-06,
+      "loss": 2.3035,
+      "step": 2465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.616733157547385e-06,
+      "loss": 2.2664,
+      "step": 2466
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.613080691896059e-06,
+      "loss": 2.2963,
+      "step": 2467
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.609427894068508e-06,
+      "loss": 2.3391,
+      "step": 2468
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.6057747660438745e-06,
+      "loss": 2.2482,
+      "step": 2469
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.602121309801478e-06,
+      "loss": 2.323,
+      "step": 2470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.598467527320812e-06,
+      "loss": 2.3005,
+      "step": 2471
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.594813420581554e-06,
+      "loss": 2.2468,
+      "step": 2472
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.5911589915635504e-06,
+      "loss": 2.2355,
+      "step": 2473
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.5875042422468265e-06,
+      "loss": 2.2346,
+      "step": 2474
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.5838491746115795e-06,
+      "loss": 2.2705,
+      "step": 2475
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.580193790638181e-06,
+      "loss": 2.2608,
+      "step": 2476
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.576538092307168e-06,
+      "loss": 2.2977,
+      "step": 2477
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.572882081599255e-06,
+      "loss": 2.41,
+      "step": 2478
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.569225760495324e-06,
+      "loss": 2.2424,
+      "step": 2479
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.5655691309764225e-06,
+      "loss": 2.1826,
+      "step": 2480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.561912195023766e-06,
+      "loss": 2.1958,
+      "step": 2481
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.558254954618738e-06,
+      "loss": 2.3319,
+      "step": 2482
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.554597411742886e-06,
+      "loss": 2.3113,
+      "step": 2483
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.550939568377919e-06,
+      "loss": 2.3258,
+      "step": 2484
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.547281426505714e-06,
+      "loss": 2.3114,
+      "step": 2485
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.543622988108304e-06,
+      "loss": 2.2487,
+      "step": 2486
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.539964255167884e-06,
+      "loss": 2.3067,
+      "step": 2487
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.536305229666815e-06,
+      "loss": 2.263,
+      "step": 2488
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.5326459135876064e-06,
+      "loss": 2.3824,
+      "step": 2489
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.528986308912933e-06,
+      "loss": 2.2656,
+      "step": 2490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.52532641762562e-06,
+      "loss": 2.1935,
+      "step": 2491
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.5216662417086556e-06,
+      "loss": 2.2219,
+      "step": 2492
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.51800578314517e-06,
+      "loss": 2.3177,
+      "step": 2493
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.514345043918461e-06,
+      "loss": 2.3236,
+      "step": 2494
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.51068402601197e-06,
+      "loss": 2.2375,
+      "step": 2495
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.507022731409289e-06,
+      "loss": 2.2845,
+      "step": 2496
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.503361162094163e-06,
+      "loss": 2.3073,
+      "step": 2497
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.499699320050483e-06,
+      "loss": 2.3278,
+      "step": 2498
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.496037207262293e-06,
+      "loss": 2.3175,
+      "step": 2499
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.492374825713776e-06,
+      "loss": 2.3589,
+      "step": 2500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.488712177389267e-06,
+      "loss": 2.2213,
+      "step": 2501
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.485049264273241e-06,
+      "loss": 2.377,
+      "step": 2502
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.481386088350321e-06,
+      "loss": 2.2185,
+      "step": 2503
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.4777226516052705e-06,
+      "loss": 2.2841,
+      "step": 2504
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.474058956022994e-06,
+      "loss": 2.3787,
+      "step": 2505
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.4703950035885345e-06,
+      "loss": 2.2996,
+      "step": 2506
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.4667307962870756e-06,
+      "loss": 2.2205,
+      "step": 2507
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.46306633610394e-06,
+      "loss": 2.3105,
+      "step": 2508
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.459401625024588e-06,
+      "loss": 2.2831,
+      "step": 2509
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.455736665034612e-06,
+      "loss": 2.3558,
+      "step": 2510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.4520714581197445e-06,
+      "loss": 2.2659,
+      "step": 2511
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.4484060062658466e-06,
+      "loss": 2.2319,
+      "step": 2512
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.444740311458915e-06,
+      "loss": 2.2849,
+      "step": 2513
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.441074375685078e-06,
+      "loss": 2.2957,
+      "step": 2514
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.437408200930595e-06,
+      "loss": 2.2101,
+      "step": 2515
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.433741789181854e-06,
+      "loss": 2.2736,
+      "step": 2516
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.430075142425369e-06,
+      "loss": 2.3778,
+      "step": 2517
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.4264082626477845e-06,
+      "loss": 2.1938,
+      "step": 2518
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.42274115183587e-06,
+      "loss": 2.3311,
+      "step": 2519
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.419073811976525e-06,
+      "loss": 2.3453,
+      "step": 2520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.4154062450567634e-06,
+      "loss": 2.2515,
+      "step": 2521
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.41173845306373e-06,
+      "loss": 2.2019,
+      "step": 2522
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.408070437984687e-06,
+      "loss": 2.2618,
+      "step": 2523
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.404402201807022e-06,
+      "loss": 2.2525,
+      "step": 2524
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.400733746518236e-06,
+      "loss": 2.2842,
+      "step": 2525
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.397065074105959e-06,
+      "loss": 2.2711,
+      "step": 2526
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.393396186557926e-06,
+      "loss": 2.3253,
+      "step": 2527
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.389727085861997e-06,
+      "loss": 2.3176,
+      "step": 2528
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.386057774006145e-06,
+      "loss": 2.2937,
+      "step": 2529
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.382388252978457e-06,
+      "loss": 2.2452,
+      "step": 2530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.378718524767139e-06,
+      "loss": 2.1959,
+      "step": 2531
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.375048591360496e-06,
+      "loss": 2.3519,
+      "step": 2532
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.371378454746959e-06,
+      "loss": 2.3294,
+      "step": 2533
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.3677081169150595e-06,
+      "loss": 2.3727,
+      "step": 2534
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.364037579853441e-06,
+      "loss": 2.2265,
+      "step": 2535
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.360366845550856e-06,
+      "loss": 2.2686,
+      "step": 2536
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.356695915996162e-06,
+      "loss": 2.2666,
+      "step": 2537
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.353024793178325e-06,
+      "loss": 2.2172,
+      "step": 2538
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.349353479086411e-06,
+      "loss": 2.3085,
+      "step": 2539
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.345681975709595e-06,
+      "loss": 2.347,
+      "step": 2540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.342010285037151e-06,
+      "loss": 2.2909,
+      "step": 2541
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.3383384090584555e-06,
+      "loss": 2.303,
+      "step": 2542
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.334666349762985e-06,
+      "loss": 2.176,
+      "step": 2543
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.330994109140315e-06,
+      "loss": 2.2965,
+      "step": 2544
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.327321689180124e-06,
+      "loss": 2.2984,
+      "step": 2545
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.323649091872179e-06,
+      "loss": 2.2911,
+      "step": 2546
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.319976319206352e-06,
+      "loss": 2.3039,
+      "step": 2547
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.316303373172601e-06,
+      "loss": 2.2548,
+      "step": 2548
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.312630255760986e-06,
+      "loss": 2.2355,
+      "step": 2549
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.308956968961655e-06,
+      "loss": 2.2563,
+      "step": 2550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.305283514764852e-06,
+      "loss": 2.3179,
+      "step": 2551
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.301609895160906e-06,
+      "loss": 2.2554,
+      "step": 2552
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.297936112140239e-06,
+      "loss": 2.268,
+      "step": 2553
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.294262167693363e-06,
+      "loss": 2.2758,
+      "step": 2554
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.290588063810874e-06,
+      "loss": 2.2404,
+      "step": 2555
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.2869138024834585e-06,
+      "loss": 2.3189,
+      "step": 2556
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.283239385701882e-06,
+      "loss": 2.2842,
+      "step": 2557
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.2795648154570016e-06,
+      "loss": 2.3114,
+      "step": 2558
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.2758900937397504e-06,
+      "loss": 2.18,
+      "step": 2559
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.27221522254115e-06,
+      "loss": 2.3575,
+      "step": 2560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.268540203852301e-06,
+      "loss": 2.3317,
+      "step": 2561
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.2648650396643806e-06,
+      "loss": 2.3192,
+      "step": 2562
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.261189731968649e-06,
+      "loss": 2.3208,
+      "step": 2563
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.257514282756441e-06,
+      "loss": 2.1276,
+      "step": 2564
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.253838694019171e-06,
+      "loss": 2.234,
+      "step": 2565
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.250162967748327e-06,
+      "loss": 2.2172,
+      "step": 2566
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.246487105935473e-06,
+      "loss": 2.1789,
+      "step": 2567
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.242811110572243e-06,
+      "loss": 2.2288,
+      "step": 2568
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.239134983650346e-06,
+      "loss": 2.2925,
+      "step": 2569
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.2354587271615664e-06,
+      "loss": 2.3138,
+      "step": 2570
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.231782343097751e-06,
+      "loss": 2.2919,
+      "step": 2571
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.228105833450819e-06,
+      "loss": 2.2943,
+      "step": 2572
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.224429200212761e-06,
+      "loss": 2.2489,
+      "step": 2573
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.220752445375627e-06,
+      "loss": 2.2497,
+      "step": 2574
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.217075570931541e-06,
+      "loss": 2.2191,
+      "step": 2575
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.213398578872688e-06,
+      "loss": 2.2998,
+      "step": 2576
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.209721471191315e-06,
+      "loss": 2.3186,
+      "step": 2577
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.206044249879734e-06,
+      "loss": 2.317,
+      "step": 2578
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.202366916930319e-06,
+      "loss": 2.2738,
+      "step": 2579
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.198689474335503e-06,
+      "loss": 2.3147,
+      "step": 2580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.19501192408778e-06,
+      "loss": 2.3086,
+      "step": 2581
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.1913342681797e-06,
+      "loss": 2.3692,
+      "step": 2582
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.187656508603872e-06,
+      "loss": 2.2679,
+      "step": 2583
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.183978647352961e-06,
+      "loss": 2.3208,
+      "step": 2584
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.180300686419686e-06,
+      "loss": 2.2967,
+      "step": 2585
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.176622627796822e-06,
+      "loss": 2.2703,
+      "step": 2586
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.172944473477196e-06,
+      "loss": 2.345,
+      "step": 2587
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.1692662254536865e-06,
+      "loss": 2.2955,
+      "step": 2588
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.165587885719221e-06,
+      "loss": 2.2315,
+      "step": 2589
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.161909456266781e-06,
+      "loss": 2.1915,
+      "step": 2590
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.158230939089392e-06,
+      "loss": 2.3008,
+      "step": 2591
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.154552336180132e-06,
+      "loss": 2.3151,
+      "step": 2592
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.1508736495321185e-06,
+      "loss": 2.2168,
+      "step": 2593
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.147194881138522e-06,
+      "loss": 2.2299,
+      "step": 2594
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.143516032992555e-06,
+      "loss": 2.1429,
+      "step": 2595
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.139837107087468e-06,
+      "loss": 2.1047,
+      "step": 2596
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.136158105416562e-06,
+      "loss": 2.2766,
+      "step": 2597
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.132479029973169e-06,
+      "loss": 2.3076,
+      "step": 2598
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.128799882750674e-06,
+      "loss": 2.1961,
+      "step": 2599
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.125120665742486e-06,
+      "loss": 2.3207,
+      "step": 2600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.121441380942066e-06,
+      "loss": 2.3628,
+      "step": 2601
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.117762030342902e-06,
+      "loss": 2.2723,
+      "step": 2602
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.11408261593852e-06,
+      "loss": 2.2937,
+      "step": 2603
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.110403139722484e-06,
+      "loss": 2.1539,
+      "step": 2604
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.106723603688386e-06,
+      "loss": 2.2648,
+      "step": 2605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.103044009829855e-06,
+      "loss": 2.2352,
+      "step": 2606
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.099364360140549e-06,
+      "loss": 2.263,
+      "step": 2607
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.0956846566141595e-06,
+      "loss": 2.215,
+      "step": 2608
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.092004901244398e-06,
+      "loss": 2.2412,
+      "step": 2609
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.0883250960250176e-06,
+      "loss": 2.3381,
+      "step": 2610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.084645242949787e-06,
+      "loss": 2.2021,
+      "step": 2611
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.080965344012509e-06,
+      "loss": 2.2899,
+      "step": 2612
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.077285401207004e-06,
+      "loss": 2.3078,
+      "step": 2613
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.073605416527119e-06,
+      "loss": 2.382,
+      "step": 2614
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.069925391966728e-06,
+      "loss": 2.3289,
+      "step": 2615
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.066245329519721e-06,
+      "loss": 2.182,
+      "step": 2616
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.0625652311800125e-06,
+      "loss": 2.2549,
+      "step": 2617
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.0588850989415295e-06,
+      "loss": 2.2441,
+      "step": 2618
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.055204934798227e-06,
+      "loss": 2.2953,
+      "step": 2619
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.05152474074407e-06,
+      "loss": 2.2921,
+      "step": 2620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.047844518773044e-06,
+      "loss": 2.2929,
+      "step": 2621
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.044164270879149e-06,
+      "loss": 2.2391,
+      "step": 2622
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.040483999056393e-06,
+      "loss": 2.2793,
+      "step": 2623
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.036803705298809e-06,
+      "loss": 2.2114,
+      "step": 2624
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.033123391600426e-06,
+      "loss": 2.245,
+      "step": 2625
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.029443059955302e-06,
+      "loss": 2.3314,
+      "step": 2626
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.025762712357488e-06,
+      "loss": 2.2981,
+      "step": 2627
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.0220823508010555e-06,
+      "loss": 2.2508,
+      "step": 2628
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.0184019772800765e-06,
+      "loss": 2.3059,
+      "step": 2629
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.014721593788632e-06,
+      "loss": 2.3578,
+      "step": 2630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.011041202320811e-06,
+      "loss": 2.2926,
+      "step": 2631
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.007360804870703e-06,
+      "loss": 2.2573,
+      "step": 2632
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.0036804034324e-06,
+      "loss": 2.2362,
+      "step": 2633
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5e-06,
+      "loss": 2.2787,
+      "step": 2634
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.996319596567601e-06,
+      "loss": 2.2045,
+      "step": 2635
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.992639195129298e-06,
+      "loss": 2.2698,
+      "step": 2636
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.98895879767919e-06,
+      "loss": 2.2403,
+      "step": 2637
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.985278406211369e-06,
+      "loss": 2.2298,
+      "step": 2638
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.981598022719926e-06,
+      "loss": 2.3189,
+      "step": 2639
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.977917649198945e-06,
+      "loss": 2.2133,
+      "step": 2640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.974237287642513e-06,
+      "loss": 2.2448,
+      "step": 2641
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9705569400447e-06,
+      "loss": 2.287,
+      "step": 2642
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.966876608399573e-06,
+      "loss": 2.3466,
+      "step": 2643
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.963196294701194e-06,
+      "loss": 2.2699,
+      "step": 2644
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.9595160009436075e-06,
+      "loss": 2.1965,
+      "step": 2645
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.955835729120853e-06,
+      "loss": 2.2832,
+      "step": 2646
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.952155481226956e-06,
+      "loss": 2.2444,
+      "step": 2647
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.948475259255931e-06,
+      "loss": 2.2219,
+      "step": 2648
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.944795065201775e-06,
+      "loss": 2.2336,
+      "step": 2649
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.941114901058472e-06,
+      "loss": 2.3432,
+      "step": 2650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.937434768819989e-06,
+      "loss": 2.3156,
+      "step": 2651
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.933754670480281e-06,
+      "loss": 2.2735,
+      "step": 2652
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.930074608033274e-06,
+      "loss": 2.2329,
+      "step": 2653
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.926394583472881e-06,
+      "loss": 2.2788,
+      "step": 2654
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.922714598792998e-06,
+      "loss": 2.379,
+      "step": 2655
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.919034655987493e-06,
+      "loss": 2.3807,
+      "step": 2656
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.915354757050214e-06,
+      "loss": 2.3005,
+      "step": 2657
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.911674903974983e-06,
+      "loss": 2.2792,
+      "step": 2658
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.907995098755603e-06,
+      "loss": 2.2199,
+      "step": 2659
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.904315343385844e-06,
+      "loss": 2.2817,
+      "step": 2660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.900635639859452e-06,
+      "loss": 2.2792,
+      "step": 2661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.8969559901701466e-06,
+      "loss": 2.2979,
+      "step": 2662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.8932763963116155e-06,
+      "loss": 2.178,
+      "step": 2663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.889596860277519e-06,
+      "loss": 2.2957,
+      "step": 2664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.885917384061481e-06,
+      "loss": 2.2518,
+      "step": 2665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.882237969657099e-06,
+      "loss": 2.3073,
+      "step": 2666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.8785586190579355e-06,
+      "loss": 2.1917,
+      "step": 2667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.874879334257514e-06,
+      "loss": 2.1715,
+      "step": 2668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.871200117249328e-06,
+      "loss": 2.3137,
+      "step": 2669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.8675209700268315e-06,
+      "loss": 2.2484,
+      "step": 2670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.863841894583441e-06,
+      "loss": 2.2259,
+      "step": 2671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.8601628929125325e-06,
+      "loss": 2.1498,
+      "step": 2672
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.856483967007447e-06,
+      "loss": 2.3463,
+      "step": 2673
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.852805118861479e-06,
+      "loss": 2.3547,
+      "step": 2674
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.849126350467884e-06,
+      "loss": 2.2718,
+      "step": 2675
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.84544766381987e-06,
+      "loss": 2.2912,
+      "step": 2676
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.8417690609106095e-06,
+      "loss": 2.2989,
+      "step": 2677
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.838090543733222e-06,
+      "loss": 2.2566,
+      "step": 2678
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.83441211428078e-06,
+      "loss": 2.2728,
+      "step": 2679
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.830733774546315e-06,
+      "loss": 2.2796,
+      "step": 2680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.827055526522805e-06,
+      "loss": 2.2792,
+      "step": 2681
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.823377372203179e-06,
+      "loss": 2.2955,
+      "step": 2682
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.819699313580314e-06,
+      "loss": 2.3667,
+      "step": 2683
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.8160213526470405e-06,
+      "loss": 2.3088,
+      "step": 2684
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.81234349139613e-06,
+      "loss": 2.328,
+      "step": 2685
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.8086657318203e-06,
+      "loss": 2.3756,
+      "step": 2686
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.804988075912221e-06,
+      "loss": 2.1752,
+      "step": 2687
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.801310525664499e-06,
+      "loss": 2.3281,
+      "step": 2688
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.797633083069683e-06,
+      "loss": 2.2007,
+      "step": 2689
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.793955750120267e-06,
+      "loss": 2.3353,
+      "step": 2690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.790278528808686e-06,
+      "loss": 2.2257,
+      "step": 2691
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.786601421127314e-06,
+      "loss": 2.317,
+      "step": 2692
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.782924429068459e-06,
+      "loss": 2.3138,
+      "step": 2693
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.779247554624374e-06,
+      "loss": 2.3511,
+      "step": 2694
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.775570799787241e-06,
+      "loss": 2.2595,
+      "step": 2695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7718941665491825e-06,
+      "loss": 2.3008,
+      "step": 2696
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.768217656902251e-06,
+      "loss": 2.2794,
+      "step": 2697
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.764541272838435e-06,
+      "loss": 2.3303,
+      "step": 2698
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.760865016349656e-06,
+      "loss": 2.199,
+      "step": 2699
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.757188889427761e-06,
+      "loss": 2.2429,
+      "step": 2700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.753512894064529e-06,
+      "loss": 2.3569,
+      "step": 2701
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.749837032251675e-06,
+      "loss": 2.2261,
+      "step": 2702
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.746161305980831e-06,
+      "loss": 2.337,
+      "step": 2703
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.74248571724356e-06,
+      "loss": 2.3427,
+      "step": 2704
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.738810268031353e-06,
+      "loss": 2.2734,
+      "step": 2705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.73513496033562e-06,
+      "loss": 2.3379,
+      "step": 2706
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.731459796147701e-06,
+      "loss": 2.2629,
+      "step": 2707
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7277847774588505e-06,
+      "loss": 2.2722,
+      "step": 2708
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.72410990626025e-06,
+      "loss": 2.2955,
+      "step": 2709
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.720435184543001e-06,
+      "loss": 2.2652,
+      "step": 2710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.716760614298118e-06,
+      "loss": 2.272,
+      "step": 2711
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.713086197516542e-06,
+      "loss": 2.2845,
+      "step": 2712
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7094119361891275e-06,
+      "loss": 2.3593,
+      "step": 2713
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.705737832306639e-06,
+      "loss": 2.3051,
+      "step": 2714
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.7020638878597615e-06,
+      "loss": 2.3638,
+      "step": 2715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.698390104839096e-06,
+      "loss": 2.2979,
+      "step": 2716
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6947164852351504e-06,
+      "loss": 2.298,
+      "step": 2717
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.691043031038346e-06,
+      "loss": 2.3204,
+      "step": 2718
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6873697442390155e-06,
+      "loss": 2.2379,
+      "step": 2719
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.683696626827401e-06,
+      "loss": 2.3764,
+      "step": 2720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.680023680793651e-06,
+      "loss": 2.2873,
+      "step": 2721
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6763509081278215e-06,
+      "loss": 2.313,
+      "step": 2722
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.672678310819878e-06,
+      "loss": 2.2694,
+      "step": 2723
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6690058908596864e-06,
+      "loss": 2.2867,
+      "step": 2724
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.665333650237016e-06,
+      "loss": 2.3355,
+      "step": 2725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.661661590941546e-06,
+      "loss": 2.2771,
+      "step": 2726
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.65798971496285e-06,
+      "loss": 2.3076,
+      "step": 2727
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.654318024290407e-06,
+      "loss": 2.3388,
+      "step": 2728
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.65064652091359e-06,
+      "loss": 2.2053,
+      "step": 2729
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.646975206821677e-06,
+      "loss": 2.4154,
+      "step": 2730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.643304084003839e-06,
+      "loss": 2.3292,
+      "step": 2731
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.639633154449146e-06,
+      "loss": 2.3153,
+      "step": 2732
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.63596242014656e-06,
+      "loss": 2.311,
+      "step": 2733
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.632291883084943e-06,
+      "loss": 2.3621,
+      "step": 2734
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.628621545253043e-06,
+      "loss": 2.2722,
+      "step": 2735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.624951408639504e-06,
+      "loss": 2.3354,
+      "step": 2736
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.621281475232863e-06,
+      "loss": 2.2645,
+      "step": 2737
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.6176117470215434e-06,
+      "loss": 2.2584,
+      "step": 2738
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.613942225993857e-06,
+      "loss": 2.2542,
+      "step": 2739
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.610272914138004e-06,
+      "loss": 2.272,
+      "step": 2740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.606603813442075e-06,
+      "loss": 2.3611,
+      "step": 2741
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.602934925894043e-06,
+      "loss": 2.2677,
+      "step": 2742
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.599266253481763e-06,
+      "loss": 2.2782,
+      "step": 2743
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.59559779819298e-06,
+      "loss": 2.2941,
+      "step": 2744
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.591929562015315e-06,
+      "loss": 2.2458,
+      "step": 2745
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.588261546936273e-06,
+      "loss": 2.2057,
+      "step": 2746
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5845937549432365e-06,
+      "loss": 2.2168,
+      "step": 2747
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5809261880234765e-06,
+      "loss": 2.3356,
+      "step": 2748
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5772588481641304e-06,
+      "loss": 2.3099,
+      "step": 2749
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.573591737352216e-06,
+      "loss": 2.2851,
+      "step": 2750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.569924857574632e-06,
+      "loss": 2.278,
+      "step": 2751
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.566258210818148e-06,
+      "loss": 2.2695,
+      "step": 2752
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.5625917990694065e-06,
+      "loss": 2.3323,
+      "step": 2753
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.558925624314923e-06,
+      "loss": 2.2622,
+      "step": 2754
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.555259688541087e-06,
+      "loss": 2.3509,
+      "step": 2755
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.551593993734156e-06,
+      "loss": 2.3517,
+      "step": 2756
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.547928541880258e-06,
+      "loss": 2.2922,
+      "step": 2757
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.544263334965389e-06,
+      "loss": 2.275,
+      "step": 2758
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.540598374975415e-06,
+      "loss": 2.2574,
+      "step": 2759
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.536933663896062e-06,
+      "loss": 2.1995,
+      "step": 2760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.533269203712925e-06,
+      "loss": 2.3375,
+      "step": 2761
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.529604996411467e-06,
+      "loss": 2.245,
+      "step": 2762
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.525941043977008e-06,
+      "loss": 2.2573,
+      "step": 2763
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.522277348394731e-06,
+      "loss": 2.328,
+      "step": 2764
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.518613911649679e-06,
+      "loss": 2.3171,
+      "step": 2765
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.51495073572676e-06,
+      "loss": 2.3462,
+      "step": 2766
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.511287822610736e-06,
+      "loss": 2.2789,
+      "step": 2767
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.507625174286226e-06,
+      "loss": 2.2281,
+      "step": 2768
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.503962792737709e-06,
+      "loss": 2.2496,
+      "step": 2769
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.500300679949518e-06,
+      "loss": 2.1289,
+      "step": 2770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4966388379058395e-06,
+      "loss": 2.2615,
+      "step": 2771
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.492977268590712e-06,
+      "loss": 2.2283,
+      "step": 2772
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.4893159739880315e-06,
+      "loss": 2.2692,
+      "step": 2773
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.48565495608154e-06,
+      "loss": 2.2278,
+      "step": 2774
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.481994216854829e-06,
+      "loss": 2.2503,
+      "step": 2775
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.478333758291347e-06,
+      "loss": 2.3164,
+      "step": 2776
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.474673582374381e-06,
+      "loss": 2.2789,
+      "step": 2777
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.47101369108707e-06,
+      "loss": 2.2525,
+      "step": 2778
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.467354086412394e-06,
+      "loss": 2.2734,
+      "step": 2779
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.463694770333187e-06,
+      "loss": 2.2363,
+      "step": 2780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.460035744832116e-06,
+      "loss": 2.3361,
+      "step": 2781
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.456377011891699e-06,
+      "loss": 2.3031,
+      "step": 2782
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.452718573494288e-06,
+      "loss": 2.3654,
+      "step": 2783
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.449060431622082e-06,
+      "loss": 2.2443,
+      "step": 2784
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.445402588257116e-06,
+      "loss": 2.2984,
+      "step": 2785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.441745045381262e-06,
+      "loss": 2.2217,
+      "step": 2786
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.438087804976234e-06,
+      "loss": 2.2531,
+      "step": 2787
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.434430869023579e-06,
+      "loss": 2.2551,
+      "step": 2788
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.430774239504678e-06,
+      "loss": 2.3265,
+      "step": 2789
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.427117918400746e-06,
+      "loss": 2.2653,
+      "step": 2790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.423461907692834e-06,
+      "loss": 2.2728,
+      "step": 2791
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.419806209361822e-06,
+      "loss": 2.2334,
+      "step": 2792
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.416150825388421e-06,
+      "loss": 2.2518,
+      "step": 2793
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.412495757753174e-06,
+      "loss": 2.3317,
+      "step": 2794
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.408841008436451e-06,
+      "loss": 2.327,
+      "step": 2795
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.405186579418448e-06,
+      "loss": 2.2295,
+      "step": 2796
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.401532472679188e-06,
+      "loss": 2.3349,
+      "step": 2797
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.397878690198524e-06,
+      "loss": 2.2926,
+      "step": 2798
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.394225233956127e-06,
+      "loss": 2.2603,
+      "step": 2799
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.390572105931492e-06,
+      "loss": 2.3217,
+      "step": 2800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.386919308103942e-06,
+      "loss": 2.298,
+      "step": 2801
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.383266842452616e-06,
+      "loss": 2.2782,
+      "step": 2802
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.379614710956474e-06,
+      "loss": 2.2524,
+      "step": 2803
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.375962915594292e-06,
+      "loss": 2.2791,
+      "step": 2804
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.372311458344672e-06,
+      "loss": 2.2602,
+      "step": 2805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.368660341186024e-06,
+      "loss": 2.324,
+      "step": 2806
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.3650095660965794e-06,
+      "loss": 2.1852,
+      "step": 2807
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.36135913505438e-06,
+      "loss": 2.238,
+      "step": 2808
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.3577090500372875e-06,
+      "loss": 2.2756,
+      "step": 2809
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.3540593130229695e-06,
+      "loss": 2.2493,
+      "step": 2810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.350409925988907e-06,
+      "loss": 2.3236,
+      "step": 2811
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.346760890912394e-06,
+      "loss": 2.304,
+      "step": 2812
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.343112209770531e-06,
+      "loss": 2.1984,
+      "step": 2813
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.339463884540229e-06,
+      "loss": 2.3425,
+      "step": 2814
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.3358159171982016e-06,
+      "loss": 2.2734,
+      "step": 2815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.332168309720975e-06,
+      "loss": 2.2588,
+      "step": 2816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.3285210640848756e-06,
+      "loss": 2.2356,
+      "step": 2817
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.324874182266035e-06,
+      "loss": 2.3399,
+      "step": 2818
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.321227666240388e-06,
+      "loss": 2.3129,
+      "step": 2819
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.3175815179836735e-06,
+      "loss": 2.3077,
+      "step": 2820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.313935739471425e-06,
+      "loss": 2.3179,
+      "step": 2821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.31029033267898e-06,
+      "loss": 2.231,
+      "step": 2822
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.306645299581478e-06,
+      "loss": 2.2036,
+      "step": 2823
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.303000642153847e-06,
+      "loss": 2.2485,
+      "step": 2824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.29935636237082e-06,
+      "loss": 2.3392,
+      "step": 2825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.29571246220692e-06,
+      "loss": 2.3039,
+      "step": 2826
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.292068943636468e-06,
+      "loss": 2.4299,
+      "step": 2827
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.2884258086335755e-06,
+      "loss": 2.4013,
+      "step": 2828
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.2847830591721456e-06,
+      "loss": 2.2017,
+      "step": 2829
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.281140697225876e-06,
+      "loss": 2.261,
+      "step": 2830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.277498724768253e-06,
+      "loss": 2.2859,
+      "step": 2831
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.27385714377255e-06,
+      "loss": 2.234,
+      "step": 2832
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.270215956211828e-06,
+      "loss": 2.2831,
+      "step": 2833
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.266575164058942e-06,
+      "loss": 2.3097,
+      "step": 2834
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.262934769286522e-06,
+      "loss": 2.3634,
+      "step": 2835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.259294773866988e-06,
+      "loss": 2.2977,
+      "step": 2836
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.2556551797725465e-06,
+      "loss": 2.2604,
+      "step": 2837
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.252015988975183e-06,
+      "loss": 2.2809,
+      "step": 2838
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.248377203446665e-06,
+      "loss": 2.2812,
+      "step": 2839
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.244738825158538e-06,
+      "loss": 2.2927,
+      "step": 2840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.241100856082134e-06,
+      "loss": 2.3006,
+      "step": 2841
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.237463298188556e-06,
+      "loss": 2.3085,
+      "step": 2842
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.233826153448685e-06,
+      "loss": 2.162,
+      "step": 2843
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.230189423833183e-06,
+      "loss": 2.2359,
+      "step": 2844
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.226553111312483e-06,
+      "loss": 2.1786,
+      "step": 2845
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.222917217856795e-06,
+      "loss": 2.2384,
+      "step": 2846
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.219281745436095e-06,
+      "loss": 2.3027,
+      "step": 2847
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.2156466960201415e-06,
+      "loss": 2.2881,
+      "step": 2848
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.2120120715784565e-06,
+      "loss": 2.2632,
+      "step": 2849
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.208377874080331e-06,
+      "loss": 2.3267,
+      "step": 2850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.2047441054948315e-06,
+      "loss": 2.249,
+      "step": 2851
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.201110767790784e-06,
+      "loss": 2.267,
+      "step": 2852
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.1974778629367875e-06,
+      "loss": 2.2071,
+      "step": 2853
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.1938453929012014e-06,
+      "loss": 2.2835,
+      "step": 2854
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.190213359652154e-06,
+      "loss": 2.2716,
+      "step": 2855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.186581765157534e-06,
+      "loss": 2.191,
+      "step": 2856
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.1829506113849935e-06,
+      "loss": 2.3093,
+      "step": 2857
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.179319900301943e-06,
+      "loss": 2.3443,
+      "step": 2858
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.17568963387556e-06,
+      "loss": 2.2809,
+      "step": 2859
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.172059814072776e-06,
+      "loss": 2.2145,
+      "step": 2860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.168430442860276e-06,
+      "loss": 2.296,
+      "step": 2861
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.164801522204514e-06,
+      "loss": 2.3101,
+      "step": 2862
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.16117305407169e-06,
+      "loss": 2.2507,
+      "step": 2863
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.157545040427763e-06,
+      "loss": 2.2303,
+      "step": 2864
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.1539174832384425e-06,
+      "loss": 2.2646,
+      "step": 2865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.150290384469194e-06,
+      "loss": 2.3299,
+      "step": 2866
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.146663746085235e-06,
+      "loss": 2.313,
+      "step": 2867
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.143037570051529e-06,
+      "loss": 2.2818,
+      "step": 2868
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.139411858332796e-06,
+      "loss": 2.278,
+      "step": 2869
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.135786612893497e-06,
+      "loss": 2.3454,
+      "step": 2870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.1321618356978465e-06,
+      "loss": 2.2744,
+      "step": 2871
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.128537528709798e-06,
+      "loss": 2.3413,
+      "step": 2872
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.12491369389306e-06,
+      "loss": 2.2909,
+      "step": 2873
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.121290333211079e-06,
+      "loss": 2.3195,
+      "step": 2874
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.1176674486270425e-06,
+      "loss": 2.2495,
+      "step": 2875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.1140450421038865e-06,
+      "loss": 2.2586,
+      "step": 2876
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.110423115604284e-06,
+      "loss": 2.347,
+      "step": 2877
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.106801671090648e-06,
+      "loss": 2.3433,
+      "step": 2878
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.10318071052513e-06,
+      "loss": 2.3221,
+      "step": 2879
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.099560235869621e-06,
+      "loss": 2.3288,
+      "step": 2880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.095940249085749e-06,
+      "loss": 2.2557,
+      "step": 2881
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.092320752134876e-06,
+      "loss": 2.3405,
+      "step": 2882
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.088701746978097e-06,
+      "loss": 2.2507,
+      "step": 2883
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.085083235576246e-06,
+      "loss": 2.2983,
+      "step": 2884
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.081465219889884e-06,
+      "loss": 2.3135,
+      "step": 2885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.077847701879307e-06,
+      "loss": 2.3218,
+      "step": 2886
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.074230683504541e-06,
+      "loss": 2.3161,
+      "step": 2887
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.070614166725337e-06,
+      "loss": 2.3461,
+      "step": 2888
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.06699815350118e-06,
+      "loss": 2.2686,
+      "step": 2889
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.063382645791278e-06,
+      "loss": 2.2716,
+      "step": 2890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.0597676455545685e-06,
+      "loss": 2.297,
+      "step": 2891
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.056153154749711e-06,
+      "loss": 2.2867,
+      "step": 2892
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.052539175335089e-06,
+      "loss": 2.2901,
+      "step": 2893
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.048925709268815e-06,
+      "loss": 2.2884,
+      "step": 2894
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.045312758508715e-06,
+      "loss": 2.1634,
+      "step": 2895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.04170032501234e-06,
+      "loss": 2.2653,
+      "step": 2896
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.038088410736957e-06,
+      "loss": 2.2737,
+      "step": 2897
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.034477017639561e-06,
+      "loss": 2.2663,
+      "step": 2898
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.030866147676855e-06,
+      "loss": 2.2871,
+      "step": 2899
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.027255802805261e-06,
+      "loss": 2.2319,
+      "step": 2900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.023645984980919e-06,
+      "loss": 2.3387,
+      "step": 2901
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.020036696159682e-06,
+      "loss": 2.2655,
+      "step": 2902
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.016427938297117e-06,
+      "loss": 2.297,
+      "step": 2903
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.012819713348499e-06,
+      "loss": 2.257,
+      "step": 2904
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.009212023268822e-06,
+      "loss": 2.2221,
+      "step": 2905
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.005604870012785e-06,
+      "loss": 2.253,
+      "step": 2906
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.001998255534793e-06,
+      "loss": 2.2904,
+      "step": 2907
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.99839218178897e-06,
+      "loss": 2.1902,
+      "step": 2908
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.994786650729137e-06,
+      "loss": 2.2488,
+      "step": 2909
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.991181664308824e-06,
+      "loss": 2.2382,
+      "step": 2910
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9875772244812665e-06,
+      "loss": 2.2758,
+      "step": 2911
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9839733331994034e-06,
+      "loss": 2.2651,
+      "step": 2912
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9803699924158775e-06,
+      "loss": 2.2314,
+      "step": 2913
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.976767204083032e-06,
+      "loss": 2.3493,
+      "step": 2914
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.97316497015291e-06,
+      "loss": 2.2888,
+      "step": 2915
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.969563292577256e-06,
+      "loss": 2.24,
+      "step": 2916
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.965962173307513e-06,
+      "loss": 2.3328,
+      "step": 2917
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9623616142948186e-06,
+      "loss": 2.2285,
+      "step": 2918
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.958761617490012e-06,
+      "loss": 2.277,
+      "step": 2919
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.955162184843625e-06,
+      "loss": 2.2368,
+      "step": 2920
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.951563318305884e-06,
+      "loss": 2.2524,
+      "step": 2921
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.947965019826702e-06,
+      "loss": 2.2875,
+      "step": 2922
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.944367291355698e-06,
+      "loss": 2.3381,
+      "step": 2923
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9407701348421726e-06,
+      "loss": 2.2704,
+      "step": 2924
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.937173552235117e-06,
+      "loss": 2.3167,
+      "step": 2925
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.933577545483215e-06,
+      "loss": 2.2724,
+      "step": 2926
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.929982116534836e-06,
+      "loss": 2.1871,
+      "step": 2927
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.926387267338036e-06,
+      "loss": 2.2214,
+      "step": 2928
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.922792999840556e-06,
+      "loss": 2.2231,
+      "step": 2929
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.91919931598983e-06,
+      "loss": 2.2651,
+      "step": 2930
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9156062177329624e-06,
+      "loss": 2.2813,
+      "step": 2931
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.912013707016748e-06,
+      "loss": 2.3318,
+      "step": 2932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.908421785787666e-06,
+      "loss": 2.3008,
+      "step": 2933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.904830455991872e-06,
+      "loss": 2.2149,
+      "step": 2934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9012397195751995e-06,
+      "loss": 2.2465,
+      "step": 2935
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8976495784831635e-06,
+      "loss": 2.3734,
+      "step": 2936
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.894060034660957e-06,
+      "loss": 2.2666,
+      "step": 2937
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.890471090053448e-06,
+      "loss": 2.2659,
+      "step": 2938
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.886882746605179e-06,
+      "loss": 2.2492,
+      "step": 2939
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.883295006260366e-06,
+      "loss": 2.2613,
+      "step": 2940
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.879707870962904e-06,
+      "loss": 2.2818,
+      "step": 2941
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8761213426563546e-06,
+      "loss": 2.3348,
+      "step": 2942
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.872535423283949e-06,
+      "loss": 2.348,
+      "step": 2943
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.868950114788597e-06,
+      "loss": 2.3024,
+      "step": 2944
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.86536541911287e-06,
+      "loss": 2.2551,
+      "step": 2945
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8617813381990086e-06,
+      "loss": 2.2803,
+      "step": 2946
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.858197873988918e-06,
+      "loss": 2.2903,
+      "step": 2947
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.854615028424178e-06,
+      "loss": 2.2231,
+      "step": 2948
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.851032803446025e-06,
+      "loss": 2.3262,
+      "step": 2949
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.847451200995361e-06,
+      "loss": 2.2673,
+      "step": 2950
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.843870223012754e-06,
+      "loss": 2.1529,
+      "step": 2951
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.840289871438427e-06,
+      "loss": 2.3324,
+      "step": 2952
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.836710148212271e-06,
+      "loss": 2.1978,
+      "step": 2953
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.83313105527383e-06,
+      "loss": 2.2051,
+      "step": 2954
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.829552594562313e-06,
+      "loss": 2.2894,
+      "step": 2955
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8259747680165835e-06,
+      "loss": 2.2055,
+      "step": 2956
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8223975775751566e-06,
+      "loss": 2.2833,
+      "step": 2957
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.818821025176212e-06,
+      "loss": 2.3427,
+      "step": 2958
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.815245112757577e-06,
+      "loss": 2.2735,
+      "step": 2959
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8116698422567335e-06,
+      "loss": 2.3142,
+      "step": 2960
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.808095215610816e-06,
+      "loss": 2.3792,
+      "step": 2961
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8045212347566117e-06,
+      "loss": 2.1906,
+      "step": 2962
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.8009479016305562e-06,
+      "loss": 2.325,
+      "step": 2963
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7973752181687336e-06,
+      "loss": 2.3615,
+      "step": 2964
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7938031863068757e-06,
+      "loss": 2.2804,
+      "step": 2965
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7902318079803657e-06,
+      "loss": 2.1959,
+      "step": 2966
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.786661085124228e-06,
+      "loss": 2.2857,
+      "step": 2967
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7830910196731303e-06,
+      "loss": 2.1612,
+      "step": 2968
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.779521613561392e-06,
+      "loss": 2.2941,
+      "step": 2969
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.775952868722968e-06,
+      "loss": 2.2709,
+      "step": 2970
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7723847870914576e-06,
+      "loss": 2.3384,
+      "step": 2971
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7688173706000986e-06,
+      "loss": 2.3621,
+      "step": 2972
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.765250621181773e-06,
+      "loss": 2.2853,
+      "step": 2973
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7616845407689972e-06,
+      "loss": 2.2685,
+      "step": 2974
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7581191312939256e-06,
+      "loss": 2.2874,
+      "step": 2975
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7545543946883527e-06,
+      "loss": 2.3454,
+      "step": 2976
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.750990332883704e-06,
+      "loss": 2.2954,
+      "step": 2977
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7474269478110416e-06,
+      "loss": 2.3393,
+      "step": 2978
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.7438642414010566e-06,
+      "loss": 2.2062,
+      "step": 2979
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.740302215584083e-06,
+      "loss": 2.2209,
+      "step": 2980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7367408722900768e-06,
+      "loss": 2.3096,
+      "step": 2981
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7331802134486227e-06,
+      "loss": 2.283,
+      "step": 2982
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7296202409889436e-06,
+      "loss": 2.2526,
+      "step": 2983
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.726060956839884e-06,
+      "loss": 2.2138,
+      "step": 2984
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7225023629299162e-06,
+      "loss": 2.2259,
+      "step": 2985
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7189444611871383e-06,
+      "loss": 2.2986,
+      "step": 2986
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7153872535392765e-06,
+      "loss": 2.2069,
+      "step": 2987
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7118307419136786e-06,
+      "loss": 2.2235,
+      "step": 2988
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7082749282373143e-06,
+      "loss": 2.1333,
+      "step": 2989
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.7047198144367746e-06,
+      "loss": 2.1857,
+      "step": 2990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.701165402438277e-06,
+      "loss": 2.2688,
+      "step": 2991
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.697611694167652e-06,
+      "loss": 2.2068,
+      "step": 2992
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6940586915503496e-06,
+      "loss": 2.2765,
+      "step": 2993
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6905063965114444e-06,
+      "loss": 2.2853,
+      "step": 2994
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6869548109756204e-06,
+      "loss": 2.3038,
+      "step": 2995
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.683403936867179e-06,
+      "loss": 2.2793,
+      "step": 2996
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.679853776110035e-06,
+      "loss": 2.2848,
+      "step": 2997
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6763043306277214e-06,
+      "loss": 2.2126,
+      "step": 2998
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6727556023433786e-06,
+      "loss": 2.3083,
+      "step": 2999
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.669207593179759e-06,
+      "loss": 2.3331,
+      "step": 3000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.66566030505923e-06,
+      "loss": 2.3414,
+      "step": 3001
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.662113739903763e-06,
+      "loss": 2.3721,
+      "step": 3002
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6585678996349394e-06,
+      "loss": 2.244,
+      "step": 3003
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6550227861739475e-06,
+      "loss": 2.2552,
+      "step": 3004
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.651478401441585e-06,
+      "loss": 2.2572,
+      "step": 3005
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6479347473582507e-06,
+      "loss": 2.3028,
+      "step": 3006
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6443918258439477e-06,
+      "loss": 2.2003,
+      "step": 3007
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6408496388182857e-06,
+      "loss": 2.2182,
+      "step": 3008
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.637308188200474e-06,
+      "loss": 2.3099,
+      "step": 3009
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.633767475909322e-06,
+      "loss": 2.2552,
+      "step": 3010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.63022750386324e-06,
+      "loss": 2.2704,
+      "step": 3011
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.626688273980239e-06,
+      "loss": 2.3008,
+      "step": 3012
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6231497881779255e-06,
+      "loss": 2.2225,
+      "step": 3013
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.619612048373503e-06,
+      "loss": 2.236,
+      "step": 3014
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.61607505648377e-06,
+      "loss": 2.3067,
+      "step": 3015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.612538814425127e-06,
+      "loss": 2.2611,
+      "step": 3016
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.609003324113557e-06,
+      "loss": 2.2637,
+      "step": 3017
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.6054685874646413e-06,
+      "loss": 2.3708,
+      "step": 3018
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.601934606393556e-06,
+      "loss": 2.2994,
+      "step": 3019
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.598401382815062e-06,
+      "loss": 2.2608,
+      "step": 3020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.5948689186435144e-06,
+      "loss": 2.227,
+      "step": 3021
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.5913372157928515e-06,
+      "loss": 2.2819,
+      "step": 3022
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.5878062761766065e-06,
+      "loss": 2.3004,
+      "step": 3023
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.584276101707892e-06,
+      "loss": 2.2591,
+      "step": 3024
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.5807466942994084e-06,
+      "loss": 2.3357,
+      "step": 3025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.5772180558634427e-06,
+      "loss": 2.228,
+      "step": 3026
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.5736901883118646e-06,
+      "loss": 2.254,
+      "step": 3027
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.5701630935561233e-06,
+      "loss": 2.2496,
+      "step": 3028
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5666367735072492e-06,
+      "loss": 2.3295,
+      "step": 3029
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5631112300758595e-06,
+      "loss": 2.2747,
+      "step": 3030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5595864651721435e-06,
+      "loss": 2.1759,
+      "step": 3031
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5560624807058707e-06,
+      "loss": 2.3018,
+      "step": 3032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5525392785863897e-06,
+      "loss": 2.3082,
+      "step": 3033
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.549016860722623e-06,
+      "loss": 2.2499,
+      "step": 3034
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5454952290230695e-06,
+      "loss": 2.3214,
+      "step": 3035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.541974385395799e-06,
+      "loss": 2.2343,
+      "step": 3036
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.53845433174846e-06,
+      "loss": 2.2036,
+      "step": 3037
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.534935069988268e-06,
+      "loss": 2.2436,
+      "step": 3038
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5314166020220123e-06,
+      "loss": 2.3523,
+      "step": 3039
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.527898929756049e-06,
+      "loss": 2.2683,
+      "step": 3040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.524382055096309e-06,
+      "loss": 2.2677,
+      "step": 3041
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.520865979948286e-06,
+      "loss": 2.3341,
+      "step": 3042
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.517350706217037e-06,
+      "loss": 2.3008,
+      "step": 3043
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.513836235807196e-06,
+      "loss": 2.2431,
+      "step": 3044
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5103225706229525e-06,
+      "loss": 2.1985,
+      "step": 3045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5068097125680623e-06,
+      "loss": 2.279,
+      "step": 3046
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.5032976635458438e-06,
+      "loss": 2.2946,
+      "step": 3047
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4997864254591784e-06,
+      "loss": 2.3324,
+      "step": 3048
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.496276000210506e-06,
+      "loss": 2.2588,
+      "step": 3049
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.492766389701826e-06,
+      "loss": 2.3315,
+      "step": 3050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4892575958347025e-06,
+      "loss": 2.3438,
+      "step": 3051
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4857496205102475e-06,
+      "loss": 2.2313,
+      "step": 3052
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4822424656291355e-06,
+      "loss": 2.3583,
+      "step": 3053
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4787361330915926e-06,
+      "loss": 2.315,
+      "step": 3054
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.475230624797406e-06,
+      "loss": 2.2729,
+      "step": 3055
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4717259426459105e-06,
+      "loss": 2.2659,
+      "step": 3056
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.468222088535994e-06,
+      "loss": 2.2957,
+      "step": 3057
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4647190643660976e-06,
+      "loss": 2.3643,
+      "step": 3058
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4612168720342127e-06,
+      "loss": 2.2308,
+      "step": 3059
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.457715513437878e-06,
+      "loss": 2.2513,
+      "step": 3060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.45421499047418e-06,
+      "loss": 2.2095,
+      "step": 3061
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.450715305039757e-06,
+      "loss": 2.3202,
+      "step": 3062
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4472164590307895e-06,
+      "loss": 2.3426,
+      "step": 3063
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4437184543430034e-06,
+      "loss": 2.3011,
+      "step": 3064
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4402212928716683e-06,
+      "loss": 2.257,
+      "step": 3065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.436724976511603e-06,
+      "loss": 2.2152,
+      "step": 3066
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4332295071571598e-06,
+      "loss": 2.3507,
+      "step": 3067
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4297348867022353e-06,
+      "loss": 2.233,
+      "step": 3068
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.42624111704027e-06,
+      "loss": 2.3239,
+      "step": 3069
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4227482000642388e-06,
+      "loss": 2.308,
+      "step": 3070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4192561376666544e-06,
+      "loss": 2.3272,
+      "step": 3071
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.415764931739569e-06,
+      "loss": 2.3093,
+      "step": 3072
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4122745841745714e-06,
+      "loss": 2.2737,
+      "step": 3073
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4087850968627823e-06,
+      "loss": 2.2417,
+      "step": 3074
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.4052964716948557e-06,
+      "loss": 2.1568,
+      "step": 3075
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.4018087105609842e-06,
+      "loss": 2.3463,
+      "step": 3076
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.398321815350889e-06,
+      "loss": 2.2864,
+      "step": 3077
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3948357879538184e-06,
+      "loss": 2.3288,
+      "step": 3078
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.391350630258553e-06,
+      "loss": 2.2368,
+      "step": 3079
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3878663441534075e-06,
+      "loss": 2.1876,
+      "step": 3080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3843829315262177e-06,
+      "loss": 2.3249,
+      "step": 3081
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.380900394264347e-06,
+      "loss": 2.3034,
+      "step": 3082
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3774187342546884e-06,
+      "loss": 2.3507,
+      "step": 3083
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3739379533836546e-06,
+      "loss": 2.2247,
+      "step": 3084
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.370458053537186e-06,
+      "loss": 2.2995,
+      "step": 3085
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3669790366007428e-06,
+      "loss": 2.2834,
+      "step": 3086
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3635009044593087e-06,
+      "loss": 2.3157,
+      "step": 3087
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3600236589973873e-06,
+      "loss": 2.1419,
+      "step": 3088
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.356547302099e-06,
+      "loss": 2.2684,
+      "step": 3089
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.353071835647692e-06,
+      "loss": 2.2936,
+      "step": 3090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3495972615265214e-06,
+      "loss": 2.2721,
+      "step": 3091
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3461235816180638e-06,
+      "loss": 2.2647,
+      "step": 3092
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3426507978044093e-06,
+      "loss": 2.2312,
+      "step": 3093
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.339178911967166e-06,
+      "loss": 2.2252,
+      "step": 3094
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3357079259874515e-06,
+      "loss": 2.1483,
+      "step": 3095
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3322378417458985e-06,
+      "loss": 2.208,
+      "step": 3096
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.328768661122648e-06,
+      "loss": 2.2808,
+      "step": 3097
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3253003859973578e-06,
+      "loss": 2.3721,
+      "step": 3098
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3218330182491876e-06,
+      "loss": 2.2788,
+      "step": 3099
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.318366559756807e-06,
+      "loss": 2.321,
+      "step": 3100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3149010123984e-06,
+      "loss": 2.2473,
+      "step": 3101
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.311436378051649e-06,
+      "loss": 2.2745,
+      "step": 3102
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3079726585937443e-06,
+      "loss": 2.1639,
+      "step": 3103
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3045098559013787e-06,
+      "loss": 2.3954,
+      "step": 3104
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.3010479718507526e-06,
+      "loss": 2.2585,
+      "step": 3105
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2975870083175665e-06,
+      "loss": 2.258,
+      "step": 3106
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2941269671770193e-06,
+      "loss": 2.3497,
+      "step": 3107
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.290667850303816e-06,
+      "loss": 2.3697,
+      "step": 3108
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2872096595721547e-06,
+      "loss": 2.2942,
+      "step": 3109
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2837523968557357e-06,
+      "loss": 2.3507,
+      "step": 3110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2802960640277525e-06,
+      "loss": 2.2259,
+      "step": 3111
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.276840662960904e-06,
+      "loss": 2.2976,
+      "step": 3112
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2733861955273727e-06,
+      "loss": 2.2459,
+      "step": 3113
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2699326635988394e-06,
+      "loss": 2.26,
+      "step": 3114
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.266480069046484e-06,
+      "loss": 2.2713,
+      "step": 3115
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2630284137409696e-06,
+      "loss": 2.3381,
+      "step": 3116
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2595776995524554e-06,
+      "loss": 2.3386,
+      "step": 3117
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2561279283505888e-06,
+      "loss": 2.264,
+      "step": 3118
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.252679102004509e-06,
+      "loss": 2.2795,
+      "step": 3119
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.24923122238284e-06,
+      "loss": 2.2107,
+      "step": 3120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.245784291353694e-06,
+      "loss": 2.2923,
+      "step": 3121
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2423383107846694e-06,
+      "loss": 2.2895,
+      "step": 3122
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2388932825428503e-06,
+      "loss": 2.2998,
+      "step": 3123
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.2354492084948042e-06,
+      "loss": 2.2567,
+      "step": 3124
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.232006090506579e-06,
+      "loss": 2.3338,
+      "step": 3125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.2285639304437124e-06,
+      "loss": 2.2312,
+      "step": 3126
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.2251227301712136e-06,
+      "loss": 2.234,
+      "step": 3127
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.221682491553578e-06,
+      "loss": 2.2414,
+      "step": 3128
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.218243216454775e-06,
+      "loss": 2.2787,
+      "step": 3129
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.214804906738259e-06,
+      "loss": 2.3446,
+      "step": 3130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.2113675642669555e-06,
+      "loss": 2.1997,
+      "step": 3131
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.2079311909032667e-06,
+      "loss": 2.2302,
+      "step": 3132
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.2044957885090723e-06,
+      "loss": 2.3819,
+      "step": 3133
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.201061358945724e-06,
+      "loss": 2.3394,
+      "step": 3134
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1976279040740456e-06,
+      "loss": 2.222,
+      "step": 3135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1941954257543333e-06,
+      "loss": 2.2413,
+      "step": 3136
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1907639258463586e-06,
+      "loss": 2.291,
+      "step": 3137
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.187333406209357e-06,
+      "loss": 2.293,
+      "step": 3138
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1839038687020317e-06,
+      "loss": 2.2579,
+      "step": 3139
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.180475315182563e-06,
+      "loss": 2.3217,
+      "step": 3140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.177047747508588e-06,
+      "loss": 2.2687,
+      "step": 3141
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1736211675372156e-06,
+      "loss": 2.3581,
+      "step": 3142
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1701955771250167e-06,
+      "loss": 2.2499,
+      "step": 3143
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1667709781280276e-06,
+      "loss": 2.2613,
+      "step": 3144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.163347372401747e-06,
+      "loss": 2.2912,
+      "step": 3145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.159924761801135e-06,
+      "loss": 2.2512,
+      "step": 3146
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1565031481806117e-06,
+      "loss": 2.3192,
+      "step": 3147
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1530825333940605e-06,
+      "loss": 2.2355,
+      "step": 3148
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1496629192948204e-06,
+      "loss": 2.3591,
+      "step": 3149
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.146244307735687e-06,
+      "loss": 2.2034,
+      "step": 3150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1428267005689182e-06,
+      "loss": 2.3203,
+      "step": 3151
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.139410099646223e-06,
+      "loss": 2.2898,
+      "step": 3152
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1359945068187665e-06,
+      "loss": 2.2923,
+      "step": 3153
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.132579923937166e-06,
+      "loss": 2.2953,
+      "step": 3154
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.129166352851496e-06,
+      "loss": 2.1744,
+      "step": 3155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.125753795411278e-06,
+      "loss": 2.2866,
+      "step": 3156
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.122342253465487e-06,
+      "loss": 2.3421,
+      "step": 3157
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1189317288625474e-06,
+      "loss": 2.2007,
+      "step": 3158
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1155222234503326e-06,
+      "loss": 2.3176,
+      "step": 3159
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.112113739076161e-06,
+      "loss": 2.2143,
+      "step": 3160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1087062775868005e-06,
+      "loss": 2.2846,
+      "step": 3161
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1052998408284664e-06,
+      "loss": 2.2179,
+      "step": 3162
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.1018944306468175e-06,
+      "loss": 2.3098,
+      "step": 3163
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.09849004888695e-06,
+      "loss": 2.2979,
+      "step": 3164
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.0950866973934148e-06,
+      "loss": 2.3187,
+      "step": 3165
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.091684378010195e-06,
+      "loss": 2.3254,
+      "step": 3166
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.088283092580719e-06,
+      "loss": 2.2339,
+      "step": 3167
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.0848828429478516e-06,
+      "loss": 2.2549,
+      "step": 3168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.081483630953902e-06,
+      "loss": 2.3309,
+      "step": 3169
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.078085458440611e-06,
+      "loss": 2.2632,
+      "step": 3170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.07468832724916e-06,
+      "loss": 2.2435,
+      "step": 3171
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.071292239220164e-06,
+      "loss": 2.336,
+      "step": 3172
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0678971961936765e-06,
+      "loss": 2.3079,
+      "step": 3173
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.06450320000918e-06,
+      "loss": 2.2061,
+      "step": 3174
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0611102525055907e-06,
+      "loss": 2.3201,
+      "step": 3175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.057718355521262e-06,
+      "loss": 2.2654,
+      "step": 3176
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.054327510893972e-06,
+      "loss": 2.3579,
+      "step": 3177
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0509377204609303e-06,
+      "loss": 2.2421,
+      "step": 3178
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0475489860587753e-06,
+      "loss": 2.215,
+      "step": 3179
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0441613095235758e-06,
+      "loss": 2.3797,
+      "step": 3180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0407746926908226e-06,
+      "loss": 2.1922,
+      "step": 3181
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0373891373954354e-06,
+      "loss": 2.2754,
+      "step": 3182
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.034004645471761e-06,
+      "loss": 2.302,
+      "step": 3183
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0306212187535653e-06,
+      "loss": 2.222,
+      "step": 3184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0272388590740388e-06,
+      "loss": 2.2856,
+      "step": 3185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0238575682657932e-06,
+      "loss": 2.3683,
+      "step": 3186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0204773481608652e-06,
+      "loss": 2.2021,
+      "step": 3187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.017098200590707e-06,
+      "loss": 2.278,
+      "step": 3188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0137201273861893e-06,
+      "loss": 2.3461,
+      "step": 3189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0103431303776053e-06,
+      "loss": 2.3076,
+      "step": 3190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0069672113946597e-06,
+      "loss": 2.2243,
+      "step": 3191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.0035923722664763e-06,
+      "loss": 2.2604,
+      "step": 3192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.000218614821593e-06,
+      "loss": 2.2537,
+      "step": 3193
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.996845940887961e-06,
+      "loss": 2.3118,
+      "step": 3194
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9934743522929476e-06,
+      "loss": 2.2268,
+      "step": 3195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9901038508633275e-06,
+      "loss": 2.154,
+      "step": 3196
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9867344384252863e-06,
+      "loss": 2.2328,
+      "step": 3197
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.983366116804427e-06,
+      "loss": 2.2734,
+      "step": 3198
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.979998887825753e-06,
+      "loss": 2.2733,
+      "step": 3199
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9766327533136776e-06,
+      "loss": 2.2905,
+      "step": 3200
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9732677150920253e-06,
+      "loss": 2.1882,
+      "step": 3201
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9699037749840232e-06,
+      "loss": 2.3076,
+      "step": 3202
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9665409348123037e-06,
+      "loss": 2.148,
+      "step": 3203
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.963179196398902e-06,
+      "loss": 2.2325,
+      "step": 3204
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.959818561565261e-06,
+      "loss": 2.2353,
+      "step": 3205
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9564590321322206e-06,
+      "loss": 2.2338,
+      "step": 3206
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9531006099200215e-06,
+      "loss": 2.1825,
+      "step": 3207
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9497432967483124e-06,
+      "loss": 2.1879,
+      "step": 3208
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9463870944361316e-06,
+      "loss": 2.1802,
+      "step": 3209
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.94303200480192e-06,
+      "loss": 2.2148,
+      "step": 3210
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.939678029663513e-06,
+      "loss": 2.1837,
+      "step": 3211
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9363251708381478e-06,
+      "loss": 2.2261,
+      "step": 3212
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9329734301424516e-06,
+      "loss": 2.2779,
+      "step": 3213
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.929622809392446e-06,
+      "loss": 2.3447,
+      "step": 3214
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.926273310403549e-06,
+      "loss": 2.2761,
+      "step": 3215
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9229249349905686e-06,
+      "loss": 2.2502,
+      "step": 3216
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9195776849677037e-06,
+      "loss": 2.3112,
+      "step": 3217
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9162315621485425e-06,
+      "loss": 2.3165,
+      "step": 3218
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9128865683460672e-06,
+      "loss": 2.2612,
+      "step": 3219
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9095427053726446e-06,
+      "loss": 2.248,
+      "step": 3220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.906199975040026e-06,
+      "loss": 2.2413,
+      "step": 3221
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9028583791593558e-06,
+      "loss": 2.1759,
+      "step": 3222
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8995179195411603e-06,
+      "loss": 2.1586,
+      "step": 3223
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8961785979953472e-06,
+      "loss": 2.3317,
+      "step": 3224
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8928404163312125e-06,
+      "loss": 2.215,
+      "step": 3225
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8895033763574343e-06,
+      "loss": 2.2762,
+      "step": 3226
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.886167479882066e-06,
+      "loss": 2.2093,
+      "step": 3227
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.882832728712551e-06,
+      "loss": 2.2164,
+      "step": 3228
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8794991246557004e-06,
+      "loss": 2.2297,
+      "step": 3229
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.876166669517718e-06,
+      "loss": 2.2805,
+      "step": 3230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8728353651041716e-06,
+      "loss": 2.3311,
+      "step": 3231
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8695052132200142e-06,
+      "loss": 2.2396,
+      "step": 3232
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8661762156695728e-06,
+      "loss": 2.2077,
+      "step": 3233
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8628483742565437e-06,
+      "loss": 2.1264,
+      "step": 3234
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8595216907840052e-06,
+      "loss": 2.2427,
+      "step": 3235
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8561961670544e-06,
+      "loss": 2.2883,
+      "step": 3236
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.852871804869547e-06,
+      "loss": 2.3039,
+      "step": 3237
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8495486060306384e-06,
+      "loss": 2.2347,
+      "step": 3238
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8462265723382255e-06,
+      "loss": 2.2062,
+      "step": 3239
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8429057055922448e-06,
+      "loss": 2.2265,
+      "step": 3240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8395860075919846e-06,
+      "loss": 2.2472,
+      "step": 3241
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8362674801361093e-06,
+      "loss": 2.2451,
+      "step": 3242
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.832950125022645e-06,
+      "loss": 2.3012,
+      "step": 3243
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8296339440489838e-06,
+      "loss": 2.3391,
+      "step": 3244
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8263189390118844e-06,
+      "loss": 2.2857,
+      "step": 3245
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8230051117074617e-06,
+      "loss": 2.2557,
+      "step": 3246
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8196924639311994e-06,
+      "loss": 2.2738,
+      "step": 3247
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8163809974779403e-06,
+      "loss": 2.2381,
+      "step": 3248
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8130707141418823e-06,
+      "loss": 2.1599,
+      "step": 3249
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8097616157165886e-06,
+      "loss": 2.2744,
+      "step": 3250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.806453703994979e-06,
+      "loss": 2.2036,
+      "step": 3251
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.8031469807693258e-06,
+      "loss": 2.2764,
+      "step": 3252
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.799841447831264e-06,
+      "loss": 2.2266,
+      "step": 3253
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7965371069717754e-06,
+      "loss": 2.3173,
+      "step": 3254
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7932339599812087e-06,
+      "loss": 2.2873,
+      "step": 3255
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7899320086492523e-06,
+      "loss": 2.2582,
+      "step": 3256
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.786631254764953e-06,
+      "loss": 2.3165,
+      "step": 3257
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7833317001167113e-06,
+      "loss": 2.3178,
+      "step": 3258
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.780033346492271e-06,
+      "loss": 2.3062,
+      "step": 3259
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.776736195678734e-06,
+      "loss": 2.2567,
+      "step": 3260
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7734402494625395e-06,
+      "loss": 2.306,
+      "step": 3261
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7701455096294837e-06,
+      "loss": 2.2486,
+      "step": 3262
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7668519779647065e-06,
+      "loss": 2.3229,
+      "step": 3263
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7635596562526867e-06,
+      "loss": 2.3265,
+      "step": 3264
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7602685462772615e-06,
+      "loss": 2.2959,
+      "step": 3265
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.7569786498215974e-06,
+      "loss": 2.2485,
+      "step": 3266
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7536899686682116e-06,
+      "loss": 2.264,
+      "step": 3267
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.750402504598958e-06,
+      "loss": 2.2658,
+      "step": 3268
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7471162593950347e-06,
+      "loss": 2.3056,
+      "step": 3269
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.74383123483698e-06,
+      "loss": 2.1846,
+      "step": 3270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7405474327046653e-06,
+      "loss": 2.2384,
+      "step": 3271
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7372648547773063e-06,
+      "loss": 2.1566,
+      "step": 3272
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.733983502833452e-06,
+      "loss": 2.3206,
+      "step": 3273
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.730703378650986e-06,
+      "loss": 2.3191,
+      "step": 3274
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7274244840071297e-06,
+      "loss": 2.3831,
+      "step": 3275
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.724146820678436e-06,
+      "loss": 2.2419,
+      "step": 3276
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.720870390440794e-06,
+      "loss": 2.2257,
+      "step": 3277
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.717595195069421e-06,
+      "loss": 2.2068,
+      "step": 3278
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7143212363388614e-06,
+      "loss": 2.2922,
+      "step": 3279
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.711048516023004e-06,
+      "loss": 2.2397,
+      "step": 3280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.7077770358950495e-06,
+      "loss": 2.3268,
+      "step": 3281
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.704506797727538e-06,
+      "loss": 2.2302,
+      "step": 3282
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.701237803292335e-06,
+      "loss": 2.272,
+      "step": 3283
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6979700543606245e-06,
+      "loss": 2.293,
+      "step": 3284
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6947035527029276e-06,
+      "loss": 2.2484,
+      "step": 3285
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6914383000890776e-06,
+      "loss": 2.2153,
+      "step": 3286
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.68817429828824e-06,
+      "loss": 2.2771,
+      "step": 3287
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6849115490689017e-06,
+      "loss": 2.3328,
+      "step": 3288
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.681650054198861e-06,
+      "loss": 2.3083,
+      "step": 3289
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6783898154452535e-06,
+      "loss": 2.1805,
+      "step": 3290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.67513083457452e-06,
+      "loss": 2.2199,
+      "step": 3291
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6718731133524267e-06,
+      "loss": 2.3451,
+      "step": 3292
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.668616653544052e-06,
+      "loss": 2.2869,
+      "step": 3293
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.665361456913797e-06,
+      "loss": 2.2929,
+      "step": 3294
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.662107525225377e-06,
+      "loss": 2.2427,
+      "step": 3295
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.658854860241816e-06,
+      "loss": 2.3484,
+      "step": 3296
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6556034637254587e-06,
+      "loss": 2.3725,
+      "step": 3297
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6523533374379616e-06,
+      "loss": 2.2803,
+      "step": 3298
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.649104483140289e-06,
+      "loss": 2.336,
+      "step": 3299
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.645856902592718e-06,
+      "loss": 2.2495,
+      "step": 3300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6426105975548376e-06,
+      "loss": 2.2808,
+      "step": 3301
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.639365569785546e-06,
+      "loss": 2.146,
+      "step": 3302
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.636121821043044e-06,
+      "loss": 2.3493,
+      "step": 3303
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6328793530848407e-06,
+      "loss": 2.3511,
+      "step": 3304
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.629638167667761e-06,
+      "loss": 2.3382,
+      "step": 3305
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.62639826654792e-06,
+      "loss": 2.1525,
+      "step": 3306
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.623159651480747e-06,
+      "loss": 2.3686,
+      "step": 3307
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.619922324220975e-06,
+      "loss": 2.3239,
+      "step": 3308
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6166862865226294e-06,
+      "loss": 2.3124,
+      "step": 3309
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.61345154013905e-06,
+      "loss": 2.2757,
+      "step": 3310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6102180868228647e-06,
+      "loss": 2.204,
+      "step": 3311
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6069859283260094e-06,
+      "loss": 2.3174,
+      "step": 3312
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.603755066399718e-06,
+      "loss": 2.3187,
+      "step": 3313
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.600525502794512e-06,
+      "loss": 2.2007,
+      "step": 3314
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.597297239260226e-06,
+      "loss": 2.2862,
+      "step": 3315
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.594070277545975e-06,
+      "loss": 2.3063,
+      "step": 3316
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5908446194001778e-06,
+      "loss": 2.2901,
+      "step": 3317
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.587620266570542e-06,
+      "loss": 2.2671,
+      "step": 3318
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.58439722080407e-06,
+      "loss": 2.3054,
+      "step": 3319
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.581175483847058e-06,
+      "loss": 2.1923,
+      "step": 3320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5779550574450883e-06,
+      "loss": 2.2986,
+      "step": 3321
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.574735943343036e-06,
+      "loss": 2.1337,
+      "step": 3322
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.571518143285068e-06,
+      "loss": 2.1976,
+      "step": 3323
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.568301659014632e-06,
+      "loss": 2.2676,
+      "step": 3324
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.565086492274469e-06,
+      "loss": 2.2628,
+      "step": 3325
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5618726448066035e-06,
+      "loss": 2.0818,
+      "step": 3326
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5586601183523485e-06,
+      "loss": 2.4115,
+      "step": 3327
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5554489146522957e-06,
+      "loss": 2.3015,
+      "step": 3328
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.55223903544632e-06,
+      "loss": 2.291,
+      "step": 3329
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.549030482473589e-06,
+      "loss": 2.3838,
+      "step": 3330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5458232574725387e-06,
+      "loss": 2.2144,
+      "step": 3331
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.542617362180893e-06,
+      "loss": 2.2386,
+      "step": 3332
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5394127983356553e-06,
+      "loss": 2.222,
+      "step": 3333
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5362095676731024e-06,
+      "loss": 2.3168,
+      "step": 3334
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.533007671928796e-06,
+      "loss": 2.2485,
+      "step": 3335
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5298071128375647e-06,
+      "loss": 2.2743,
+      "step": 3336
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.526607892133527e-06,
+      "loss": 2.2948,
+      "step": 3337
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5234100115500643e-06,
+      "loss": 2.1944,
+      "step": 3338
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.520213472819831e-06,
+      "loss": 2.3519,
+      "step": 3339
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.517018277674769e-06,
+      "loss": 2.3302,
+      "step": 3340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.513824427846075e-06,
+      "loss": 2.3291,
+      "step": 3341
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5106319250642286e-06,
+      "loss": 2.247,
+      "step": 3342
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5074407710589714e-06,
+      "loss": 2.3087,
+      "step": 3343
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5042509675593196e-06,
+      "loss": 2.3623,
+      "step": 3344
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.501062516293559e-06,
+      "loss": 2.3073,
+      "step": 3345
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.497875418989235e-06,
+      "loss": 2.3321,
+      "step": 3346
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4946896773731665e-06,
+      "loss": 2.1823,
+      "step": 3347
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.491505293171438e-06,
+      "loss": 2.3375,
+      "step": 3348
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4883222681093916e-06,
+      "loss": 2.419,
+      "step": 3349
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.48514060391164e-06,
+      "loss": 2.1959,
+      "step": 3350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4819603023020557e-06,
+      "loss": 2.3137,
+      "step": 3351
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.478781365003775e-06,
+      "loss": 2.2742,
+      "step": 3352
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.47560379373919e-06,
+      "loss": 2.2439,
+      "step": 3353
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.472427590229958e-06,
+      "loss": 2.2098,
+      "step": 3354
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4692527561969953e-06,
+      "loss": 2.2859,
+      "step": 3355
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.46607929336047e-06,
+      "loss": 2.3679,
+      "step": 3356
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4629072034398127e-06,
+      "loss": 2.2917,
+      "step": 3357
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4597364881537116e-06,
+      "loss": 2.3406,
+      "step": 3358
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4565671492201037e-06,
+      "loss": 2.2147,
+      "step": 3359
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4533991883561868e-06,
+      "loss": 2.3399,
+      "step": 3360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.4502326072784037e-06,
+      "loss": 2.3785,
+      "step": 3361
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4470674077024638e-06,
+      "loss": 2.2569,
+      "step": 3362
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4439035913433135e-06,
+      "loss": 2.2146,
+      "step": 3363
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.440741159915153e-06,
+      "loss": 2.2522,
+      "step": 3364
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4375801151314414e-06,
+      "loss": 2.3151,
+      "step": 3365
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4344204587048752e-06,
+      "loss": 2.285,
+      "step": 3366
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4312621923474063e-06,
+      "loss": 2.2585,
+      "step": 3367
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.428105317770226e-06,
+      "loss": 2.1976,
+      "step": 3368
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.424949836683778e-06,
+      "loss": 2.2056,
+      "step": 3369
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.421795750797751e-06,
+      "loss": 2.3708,
+      "step": 3370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4186430618210704e-06,
+      "loss": 2.2917,
+      "step": 3371
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.415491771461916e-06,
+      "loss": 2.2747,
+      "step": 3372
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4123418814277016e-06,
+      "loss": 2.293,
+      "step": 3373
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4091933934250816e-06,
+      "loss": 2.2883,
+      "step": 3374
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4060463091599555e-06,
+      "loss": 2.2777,
+      "step": 3375
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.4029006303374603e-06,
+      "loss": 2.3107,
+      "step": 3376
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3997563586619734e-06,
+      "loss": 2.3815,
+      "step": 3377
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.396613495837104e-06,
+      "loss": 2.3045,
+      "step": 3378
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.393472043565704e-06,
+      "loss": 2.3309,
+      "step": 3379
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3903320035498607e-06,
+      "loss": 2.2652,
+      "step": 3380
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3871933774908905e-06,
+      "loss": 2.2242,
+      "step": 3381
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.38405616708935e-06,
+      "loss": 2.2721,
+      "step": 3382
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.380920374045027e-06,
+      "loss": 2.257,
+      "step": 3383
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3777860000569388e-06,
+      "loss": 2.2496,
+      "step": 3384
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3746530468233377e-06,
+      "loss": 2.3128,
+      "step": 3385
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3715215160417e-06,
+      "loss": 2.2823,
+      "step": 3386
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3683914094087422e-06,
+      "loss": 2.2383,
+      "step": 3387
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.365262728620398e-06,
+      "loss": 2.2566,
+      "step": 3388
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.362135475371834e-06,
+      "loss": 2.3564,
+      "step": 3389
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.359009651357445e-06,
+      "loss": 2.3501,
+      "step": 3390
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.355885258270844e-06,
+      "loss": 2.3235,
+      "step": 3391
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.352762297804879e-06,
+      "loss": 2.3019,
+      "step": 3392
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.349640771651611e-06,
+      "loss": 2.3337,
+      "step": 3393
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3465206815023323e-06,
+      "loss": 2.2786,
+      "step": 3394
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.343402029047555e-06,
+      "loss": 2.2462,
+      "step": 3395
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.340284815977007e-06,
+      "loss": 2.3408,
+      "step": 3396
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3371690439796464e-06,
+      "loss": 2.3283,
+      "step": 3397
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.334054714743642e-06,
+      "loss": 2.3572,
+      "step": 3398
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3309418299563824e-06,
+      "loss": 2.3694,
+      "step": 3399
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.327830391304475e-06,
+      "loss": 2.2708,
+      "step": 3400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.324720400473745e-06,
+      "loss": 2.2994,
+      "step": 3401
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3216118591492327e-06,
+      "loss": 2.2415,
+      "step": 3402
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.318504769015188e-06,
+      "loss": 2.2716,
+      "step": 3403
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.315399131755081e-06,
+      "loss": 2.263,
+      "step": 3404
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3122949490515933e-06,
+      "loss": 2.2447,
+      "step": 3405
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.309192222586614e-06,
+      "loss": 2.2042,
+      "step": 3406
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3060909540412486e-06,
+      "loss": 2.223,
+      "step": 3407
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.3029911450958115e-06,
+      "loss": 2.2679,
+      "step": 3408
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.299892797429822e-06,
+      "loss": 2.2911,
+      "step": 3409
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.296795912722014e-06,
+      "loss": 2.3257,
+      "step": 3410
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2937004926503204e-06,
+      "loss": 2.23,
+      "step": 3411
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2906065388918934e-06,
+      "loss": 2.2814,
+      "step": 3412
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.287514053123077e-06,
+      "loss": 2.2241,
+      "step": 3413
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2844230370194285e-06,
+      "loss": 2.2269,
+      "step": 3414
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.281333492255708e-06,
+      "loss": 2.2607,
+      "step": 3415
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.278245420505873e-06,
+      "loss": 2.2296,
+      "step": 3416
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2751588234430894e-06,
+      "loss": 2.2956,
+      "step": 3417
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.272073702739719e-06,
+      "loss": 2.2259,
+      "step": 3418
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.268990060067327e-06,
+      "loss": 2.1986,
+      "step": 3419
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2659078970966786e-06,
+      "loss": 2.3638,
+      "step": 3420
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2628272154977304e-06,
+      "loss": 2.2475,
+      "step": 3421
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.259748016939649e-06,
+      "loss": 2.2498,
+      "step": 3422
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.256670303090784e-06,
+      "loss": 2.284,
+      "step": 3423
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2535940756186895e-06,
+      "loss": 2.3451,
+      "step": 3424
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2505193361901082e-06,
+      "loss": 2.3279,
+      "step": 3425
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2474460864709825e-06,
+      "loss": 2.2672,
+      "step": 3426
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2443743281264448e-06,
+      "loss": 2.3319,
+      "step": 3427
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2413040628208166e-06,
+      "loss": 2.2998,
+      "step": 3428
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2382352922176148e-06,
+      "loss": 2.2374,
+      "step": 3429
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2351680179795467e-06,
+      "loss": 2.3206,
+      "step": 3430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2321022417685034e-06,
+      "loss": 2.2874,
+      "step": 3431
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2290379652455702e-06,
+      "loss": 2.2632,
+      "step": 3432
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.225975190071019e-06,
+      "loss": 2.331,
+      "step": 3433
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2229139179043037e-06,
+      "loss": 2.1657,
+      "step": 3434
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.21985415040407e-06,
+      "loss": 2.2547,
+      "step": 3435
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2167958892281406e-06,
+      "loss": 2.2365,
+      "step": 3436
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.213739136033533e-06,
+      "loss": 2.2955,
+      "step": 3437
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2106838924764373e-06,
+      "loss": 2.2528,
+      "step": 3438
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2076301602122313e-06,
+      "loss": 2.3135,
+      "step": 3439
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2045779408954735e-06,
+      "loss": 2.3086,
+      "step": 3440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.2015272361798993e-06,
+      "loss": 2.3722,
+      "step": 3441
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1984780477184287e-06,
+      "loss": 2.1926,
+      "step": 3442
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1954303771631535e-06,
+      "loss": 2.2883,
+      "step": 3443
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.192384226165349e-06,
+      "loss": 2.3341,
+      "step": 3444
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1893395963754664e-06,
+      "loss": 2.2407,
+      "step": 3445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.186296489443127e-06,
+      "loss": 2.2247,
+      "step": 3446
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.183254907017137e-06,
+      "loss": 2.39,
+      "step": 3447
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1802148507454675e-06,
+      "loss": 2.2856,
+      "step": 3448
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.177176322275268e-06,
+      "loss": 2.2744,
+      "step": 3449
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.174139323252855e-06,
+      "loss": 2.2979,
+      "step": 3450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1711038553237224e-06,
+      "loss": 2.2596,
+      "step": 3451
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.168069920132533e-06,
+      "loss": 2.267,
+      "step": 3452
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1650375193231137e-06,
+      "loss": 2.2681,
+      "step": 3453
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1620066545384673e-06,
+      "loss": 2.2063,
+      "step": 3454
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1589773274207628e-06,
+      "loss": 2.2496,
+      "step": 3455
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1559495396113306e-06,
+      "loss": 2.2667,
+      "step": 3456
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.1529232927506727e-06,
+      "loss": 2.311,
+      "step": 3457
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1498985884784558e-06,
+      "loss": 2.3088,
+      "step": 3458
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1468754284335097e-06,
+      "loss": 2.2566,
+      "step": 3459
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1438538142538273e-06,
+      "loss": 2.3093,
+      "step": 3460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1408337475765597e-06,
+      "loss": 2.2411,
+      "step": 3461
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.13781523003803e-06,
+      "loss": 2.3052,
+      "step": 3462
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1347982632737124e-06,
+      "loss": 2.2121,
+      "step": 3463
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1317828489182452e-06,
+      "loss": 2.2184,
+      "step": 3464
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1287689886054257e-06,
+      "loss": 2.2671,
+      "step": 3465
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.125756683968206e-06,
+      "loss": 2.3281,
+      "step": 3466
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1227459366387e-06,
+      "loss": 2.2348,
+      "step": 3467
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1197367482481723e-06,
+      "loss": 2.3358,
+      "step": 3468
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1167291204270475e-06,
+      "loss": 2.2346,
+      "step": 3469
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1137230548049042e-06,
+      "loss": 2.3519,
+      "step": 3470
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.110718553010469e-06,
+      "loss": 2.2075,
+      "step": 3471
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.1077156166716324e-06,
+      "loss": 2.2664,
+      "step": 3472
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.104714247415425e-06,
+      "loss": 2.2281,
+      "step": 3473
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.101714446868035e-06,
+      "loss": 2.2165,
+      "step": 3474
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0987162166547974e-06,
+      "loss": 2.2853,
+      "step": 3475
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0957195584001986e-06,
+      "loss": 2.1828,
+      "step": 3476
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.092724473727875e-06,
+      "loss": 2.237,
+      "step": 3477
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.089730964260604e-06,
+      "loss": 2.2657,
+      "step": 3478
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0867390316203145e-06,
+      "loss": 2.2577,
+      "step": 3479
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0837486774280827e-06,
+      "loss": 2.3626,
+      "step": 3480
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0807599033041238e-06,
+      "loss": 2.2021,
+      "step": 3481
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0777727108678006e-06,
+      "loss": 2.2454,
+      "step": 3482
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0747871017376193e-06,
+      "loss": 2.3385,
+      "step": 3483
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.071803077531229e-06,
+      "loss": 2.2098,
+      "step": 3484
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0688206398654153e-06,
+      "loss": 2.323,
+      "step": 3485
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.065839790356105e-06,
+      "loss": 2.281,
+      "step": 3486
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.062860530618373e-06,
+      "loss": 2.2311,
+      "step": 3487
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0598828622664213e-06,
+      "loss": 2.264,
+      "step": 3488
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.056906786913597e-06,
+      "loss": 2.2671,
+      "step": 3489
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0539323061723835e-06,
+      "loss": 2.3136,
+      "step": 3490
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0509594216543956e-06,
+      "loss": 2.3097,
+      "step": 3491
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0479881349703885e-06,
+      "loss": 2.2998,
+      "step": 3492
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.045018447730245e-06,
+      "loss": 2.21,
+      "step": 3493
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0420503615429927e-06,
+      "loss": 2.3212,
+      "step": 3494
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.039083878016781e-06,
+      "loss": 2.2664,
+      "step": 3495
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.036118998758892e-06,
+      "loss": 2.3048,
+      "step": 3496
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0331557253757477e-06,
+      "loss": 2.1719,
+      "step": 3497
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.030194059472888e-06,
+      "loss": 2.2588,
+      "step": 3498
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.027234002654992e-06,
+      "loss": 2.2792,
+      "step": 3499
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.024275556525858e-06,
+      "loss": 2.2133,
+      "step": 3500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0213187226884184e-06,
+      "loss": 2.2552,
+      "step": 3501
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.0183635027447312e-06,
+      "loss": 2.1958,
+      "step": 3502
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.015409898295975e-06,
+      "loss": 2.1525,
+      "step": 3503
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.012457910942458e-06,
+      "loss": 2.2873,
+      "step": 3504
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.009507542283613e-06,
+      "loss": 2.2038,
+      "step": 3505
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.0065587939179892e-06,
+      "loss": 2.2386,
+      "step": 3506
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.0036116674432653e-06,
+      "loss": 2.2551,
+      "step": 3507
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.0006661644562376e-06,
+      "loss": 2.2675,
+      "step": 3508
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.997722286552825e-06,
+      "loss": 2.2751,
+      "step": 3509
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9947800353280627e-06,
+      "loss": 2.3515,
+      "step": 3510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.991839412376102e-06,
+      "loss": 2.3187,
+      "step": 3511
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.988900419290224e-06,
+      "loss": 2.3154,
+      "step": 3512
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9859630576628124e-06,
+      "loss": 2.2856,
+      "step": 3513
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.983027329085377e-06,
+      "loss": 2.2156,
+      "step": 3514
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9800932351485387e-06,
+      "loss": 2.3402,
+      "step": 3515
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.977160777442031e-06,
+      "loss": 2.2057,
+      "step": 3516
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9742299575547064e-06,
+      "loss": 2.3109,
+      "step": 3517
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9713007770745205e-06,
+      "loss": 2.2649,
+      "step": 3518
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9683732375885556e-06,
+      "loss": 2.273,
+      "step": 3519
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9654473406829907e-06,
+      "loss": 2.3028,
+      "step": 3520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9625230879431172e-06,
+      "loss": 2.2519,
+      "step": 3521
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9596004809533465e-06,
+      "loss": 2.2733,
+      "step": 3522
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9566795212971845e-06,
+      "loss": 2.2665,
+      "step": 3523
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9537602105572538e-06,
+      "loss": 2.2476,
+      "step": 3524
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.950842550315277e-06,
+      "loss": 2.3419,
+      "step": 3525
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.947926542152088e-06,
+      "loss": 2.1767,
+      "step": 3526
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.945012187647625e-06,
+      "loss": 2.1667,
+      "step": 3527
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.942099488380923e-06,
+      "loss": 2.2912,
+      "step": 3528
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9391884459301323e-06,
+      "loss": 2.2838,
+      "step": 3529
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.936279061872496e-06,
+      "loss": 2.3341,
+      "step": 3530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.93337133778436e-06,
+      "loss": 2.1613,
+      "step": 3531
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9304652752411735e-06,
+      "loss": 2.3191,
+      "step": 3532
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9275608758174847e-06,
+      "loss": 2.295,
+      "step": 3533
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9246581410869425e-06,
+      "loss": 2.3504,
+      "step": 3534
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9217570726222873e-06,
+      "loss": 2.2295,
+      "step": 3535
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9188576719953635e-06,
+      "loss": 2.1102,
+      "step": 3536
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9159599407771107e-06,
+      "loss": 2.3446,
+      "step": 3537
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9130638805375595e-06,
+      "loss": 2.2517,
+      "step": 3538
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.91016949284584e-06,
+      "loss": 2.2235,
+      "step": 3539
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.907276779270177e-06,
+      "loss": 2.2306,
+      "step": 3540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.904385741377881e-06,
+      "loss": 2.2575,
+      "step": 3541
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9014963807353632e-06,
+      "loss": 2.2946,
+      "step": 3542
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8986086989081164e-06,
+      "loss": 2.2827,
+      "step": 3543
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8957226974607374e-06,
+      "loss": 2.3415,
+      "step": 3544
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8928383779569004e-06,
+      "loss": 2.2517,
+      "step": 3545
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8899557419593694e-06,
+      "loss": 2.1966,
+      "step": 3546
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8870747910300064e-06,
+      "loss": 2.3337,
+      "step": 3547
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8841955267297479e-06,
+      "loss": 2.3389,
+      "step": 3548
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8813179506186258e-06,
+      "loss": 2.3151,
+      "step": 3549
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.87844206425575e-06,
+      "loss": 2.2931,
+      "step": 3550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.87556786919932e-06,
+      "loss": 2.189,
+      "step": 3551
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8726953670066195e-06,
+      "loss": 2.2973,
+      "step": 3552
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8698245592340076e-06,
+      "loss": 2.3364,
+      "step": 3553
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8669554474369366e-06,
+      "loss": 2.2556,
+      "step": 3554
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8640880331699317e-06,
+      "loss": 2.295,
+      "step": 3555
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.861222317986598e-06,
+      "loss": 2.2332,
+      "step": 3556
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8583583034396246e-06,
+      "loss": 2.2873,
+      "step": 3557
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8554959910807773e-06,
+      "loss": 2.2061,
+      "step": 3558
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8526353824609006e-06,
+      "loss": 2.2844,
+      "step": 3559
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8497764791299117e-06,
+      "loss": 2.2521,
+      "step": 3560
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8469192826368088e-06,
+      "loss": 2.2165,
+      "step": 3561
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8440637945296641e-06,
+      "loss": 2.2911,
+      "step": 3562
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8412100163556212e-06,
+      "loss": 2.2579,
+      "step": 3563
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8383579496609005e-06,
+      "loss": 2.257,
+      "step": 3564
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8355075959907958e-06,
+      "loss": 2.2614,
+      "step": 3565
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8326589568896686e-06,
+      "loss": 2.209,
+      "step": 3566
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.829812033900956e-06,
+      "loss": 2.2875,
+      "step": 3567
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8269668285671589e-06,
+      "loss": 2.2631,
+      "step": 3568
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8241233424298582e-06,
+      "loss": 2.2449,
+      "step": 3569
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8212815770296915e-06,
+      "loss": 2.3294,
+      "step": 3570
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.818441533906372e-06,
+      "loss": 2.2449,
+      "step": 3571
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8156032145986785e-06,
+      "loss": 2.3053,
+      "step": 3572
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8127666206444511e-06,
+      "loss": 2.2839,
+      "step": 3573
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.809931753580602e-06,
+      "loss": 2.2659,
+      "step": 3574
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8070986149431014e-06,
+      "loss": 2.2938,
+      "step": 3575
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8042672062669863e-06,
+      "loss": 2.2434,
+      "step": 3576
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.801437529086359e-06,
+      "loss": 2.2966,
+      "step": 3577
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7986095849343748e-06,
+      "loss": 2.2502,
+      "step": 3578
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7957833753432624e-06,
+      "loss": 2.2444,
+      "step": 3579
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7929589018443016e-06,
+      "loss": 2.2477,
+      "step": 3580
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7901361659678324e-06,
+      "loss": 2.2634,
+      "step": 3581
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7873151692432555e-06,
+      "loss": 2.223,
+      "step": 3582
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7844959131990298e-06,
+      "loss": 2.2276,
+      "step": 3583
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7816783993626713e-06,
+      "loss": 2.2355,
+      "step": 3584
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.778862629260748e-06,
+      "loss": 2.2449,
+      "step": 3585
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7760486044188874e-06,
+      "loss": 2.2345,
+      "step": 3586
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7732363263617712e-06,
+      "loss": 2.2643,
+      "step": 3587
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7704257966131305e-06,
+      "loss": 2.3727,
+      "step": 3588
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.767617016695754e-06,
+      "loss": 2.3702,
+      "step": 3589
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7648099881314812e-06,
+      "loss": 2.1723,
+      "step": 3590
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7620047124412e-06,
+      "loss": 2.2598,
+      "step": 3591
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759201191144852e-06,
+      "loss": 2.2899,
+      "step": 3592
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7563994257614224e-06,
+      "loss": 2.2446,
+      "step": 3593
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753599417808956e-06,
+      "loss": 2.3051,
+      "step": 3594
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.750801168804534e-06,
+      "loss": 2.3041,
+      "step": 3595
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7480046802642909e-06,
+      "loss": 2.2438,
+      "step": 3596
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7452099537034067e-06,
+      "loss": 2.2255,
+      "step": 3597
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7424169906361027e-06,
+      "loss": 2.2498,
+      "step": 3598
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7396257925756516e-06,
+      "loss": 2.2576,
+      "step": 3599
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.736836361034362e-06,
+      "loss": 2.3296,
+      "step": 3600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7340486975235915e-06,
+      "loss": 2.1612,
+      "step": 3601
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7312628035537388e-06,
+      "loss": 2.3889,
+      "step": 3602
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.728478680634238e-06,
+      "loss": 2.2397,
+      "step": 3603
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7256963302735752e-06,
+      "loss": 2.2608,
+      "step": 3604
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.722915753979263e-06,
+      "loss": 2.3173,
+      "step": 3605
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7201369532578633e-06,
+      "loss": 2.3304,
+      "step": 3606
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7173599296149678e-06,
+      "loss": 2.2228,
+      "step": 3607
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.714584684555211e-06,
+      "loss": 2.274,
+      "step": 3608
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.711811219582264e-06,
+      "loss": 2.2748,
+      "step": 3609
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7090395361988265e-06,
+      "loss": 2.1964,
+      "step": 3610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7062696359066411e-06,
+      "loss": 2.2574,
+      "step": 3611
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.703501520206482e-06,
+      "loss": 2.335,
+      "step": 3612
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7007351905981512e-06,
+      "loss": 2.2438,
+      "step": 3613
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6979706485804898e-06,
+      "loss": 2.3174,
+      "step": 3614
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6952078956513696e-06,
+      "loss": 2.2334,
+      "step": 3615
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.692446933307687e-06,
+      "loss": 2.1879,
+      "step": 3616
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6896877630453756e-06,
+      "loss": 2.2639,
+      "step": 3617
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.686930386359391e-06,
+      "loss": 2.2357,
+      "step": 3618
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6841748047437267e-06,
+      "loss": 2.2867,
+      "step": 3619
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6814210196913928e-06,
+      "loss": 2.2383,
+      "step": 3620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6786690326944321e-06,
+      "loss": 2.2089,
+      "step": 3621
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6759188452439152e-06,
+      "loss": 2.2498,
+      "step": 3622
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6731704588299298e-06,
+      "loss": 2.2397,
+      "step": 3623
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6704238749415958e-06,
+      "loss": 2.1948,
+      "step": 3624
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6676790950670507e-06,
+      "loss": 2.2936,
+      "step": 3625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6649361206934577e-06,
+      "loss": 2.2903,
+      "step": 3626
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6621949533070037e-06,
+      "loss": 2.2141,
+      "step": 3627
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6594555943928886e-06,
+      "loss": 2.1798,
+      "step": 3628
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6567180454353437e-06,
+      "loss": 2.2,
+      "step": 3629
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6539823079176093e-06,
+      "loss": 2.2873,
+      "step": 3630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6512483833219522e-06,
+      "loss": 2.2343,
+      "step": 3631
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6485162731296494e-06,
+      "loss": 2.2866,
+      "step": 3632
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6457859788210006e-06,
+      "loss": 2.3093,
+      "step": 3633
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6430575018753214e-06,
+      "loss": 2.3712,
+      "step": 3634
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.640330843770938e-06,
+      "loss": 2.2483,
+      "step": 3635
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6376060059851962e-06,
+      "loss": 2.3467,
+      "step": 3636
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6348829899944552e-06,
+      "loss": 2.2695,
+      "step": 3637
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.632161797274081e-06,
+      "loss": 2.2715,
+      "step": 3638
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6294424292984596e-06,
+      "loss": 2.2386,
+      "step": 3639
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6267248875409836e-06,
+      "loss": 2.2861,
+      "step": 3640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6240091734740597e-06,
+      "loss": 2.2474,
+      "step": 3641
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6212952885691008e-06,
+      "loss": 2.197,
+      "step": 3642
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6185832342965257e-06,
+      "loss": 2.2523,
+      "step": 3643
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6158730121257738e-06,
+      "loss": 2.253,
+      "step": 3644
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6131646235252773e-06,
+      "loss": 2.2655,
+      "step": 3645
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6104580699624839e-06,
+      "loss": 2.2991,
+      "step": 3646
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6077533529038458e-06,
+      "loss": 2.2619,
+      "step": 3647
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6050504738148153e-06,
+      "loss": 2.2901,
+      "step": 3648
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6023494341598566e-06,
+      "loss": 2.2812,
+      "step": 3649
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5996502354024295e-06,
+      "loss": 2.2287,
+      "step": 3650
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5969528790050009e-06,
+      "loss": 2.3788,
+      "step": 3651
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.594257366429041e-06,
+      "loss": 2.2525,
+      "step": 3652
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.591563699135013e-06,
+      "loss": 2.2852,
+      "step": 3653
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5888718785823932e-06,
+      "loss": 2.2042,
+      "step": 3654
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5861819062296441e-06,
+      "loss": 2.3303,
+      "step": 3655
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5834937835342368e-06,
+      "loss": 2.3009,
+      "step": 3656
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5808075119526323e-06,
+      "loss": 2.3781,
+      "step": 3657
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5781230929402941e-06,
+      "loss": 2.2574,
+      "step": 3658
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.575440527951682e-06,
+      "loss": 2.2244,
+      "step": 3659
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5727598184402466e-06,
+      "loss": 2.2947,
+      "step": 3660
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.570080965858437e-06,
+      "loss": 2.2828,
+      "step": 3661
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5674039716576977e-06,
+      "loss": 2.3124,
+      "step": 3662
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.56472883728846e-06,
+      "loss": 2.1889,
+      "step": 3663
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5620555642001539e-06,
+      "loss": 2.2972,
+      "step": 3664
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5593841538411974e-06,
+      "loss": 2.2059,
+      "step": 3665
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5567146076590029e-06,
+      "loss": 2.2285,
+      "step": 3666
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.554046927099968e-06,
+      "loss": 2.3036,
+      "step": 3667
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5513811136094786e-06,
+      "loss": 2.2569,
+      "step": 3668
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5487171686319187e-06,
+      "loss": 2.2766,
+      "step": 3669
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5460550936106478e-06,
+      "loss": 2.2884,
+      "step": 3670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5433948899880196e-06,
+      "loss": 2.2339,
+      "step": 3671
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5407365592053736e-06,
+      "loss": 2.29,
+      "step": 3672
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5380801027030301e-06,
+      "loss": 2.2696,
+      "step": 3673
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.535425521920299e-06,
+      "loss": 2.2438,
+      "step": 3674
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5327728182954672e-06,
+      "loss": 2.2766,
+      "step": 3675
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5301219932658157e-06,
+      "loss": 2.2903,
+      "step": 3676
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.527473048267597e-06,
+      "loss": 2.2057,
+      "step": 3677
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5248259847360465e-06,
+      "loss": 2.2405,
+      "step": 3678
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5221808041053877e-06,
+      "loss": 2.3007,
+      "step": 3679
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5195375078088148e-06,
+      "loss": 2.2307,
+      "step": 3680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5168960972785084e-06,
+      "loss": 2.4004,
+      "step": 3681
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.51425657394562e-06,
+      "loss": 2.3007,
+      "step": 3682
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5116189392402847e-06,
+      "loss": 2.3048,
+      "step": 3683
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5089831945916134e-06,
+      "loss": 2.1496,
+      "step": 3684
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5063493414276886e-06,
+      "loss": 2.316,
+      "step": 3685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.5037173811755734e-06,
+      "loss": 2.3855,
+      "step": 3686
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.501087315261303e-06,
+      "loss": 2.2482,
+      "step": 3687
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4984591451098846e-06,
+      "loss": 2.1986,
+      "step": 3688
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4958328721453002e-06,
+      "loss": 2.2201,
+      "step": 3689
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4932084977905043e-06,
+      "loss": 2.2316,
+      "step": 3690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4905860234674224e-06,
+      "loss": 2.1812,
+      "step": 3691
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4879654505969499e-06,
+      "loss": 2.2997,
+      "step": 3692
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.485346780598948e-06,
+      "loss": 2.2601,
+      "step": 3693
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4827300148922573e-06,
+      "loss": 2.1919,
+      "step": 3694
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4801151548946764e-06,
+      "loss": 2.3431,
+      "step": 3695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4775022020229758e-06,
+      "loss": 2.166,
+      "step": 3696
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4748911576928942e-06,
+      "loss": 2.3052,
+      "step": 3697
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4722820233191315e-06,
+      "loss": 2.1672,
+      "step": 3698
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4696748003153577e-06,
+      "loss": 2.3325,
+      "step": 3699
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4670694900942006e-06,
+      "loss": 2.2097,
+      "step": 3700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4644660940672628e-06,
+      "loss": 2.3071,
+      "step": 3701
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4618646136450992e-06,
+      "loss": 2.3016,
+      "step": 3702
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4592650502372275e-06,
+      "loss": 2.362,
+      "step": 3703
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4566674052521357e-06,
+      "loss": 2.2403,
+      "step": 3704
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4540716800972626e-06,
+      "loss": 2.3055,
+      "step": 3705
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4514778761790128e-06,
+      "loss": 2.2432,
+      "step": 3706
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4488859949027457e-06,
+      "loss": 2.2396,
+      "step": 3707
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4462960376727813e-06,
+      "loss": 2.2357,
+      "step": 3708
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4437080058923996e-06,
+      "loss": 2.398,
+      "step": 3709
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4411219009638294e-06,
+      "loss": 2.3804,
+      "step": 3710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.438537724288267e-06,
+      "loss": 2.2624,
+      "step": 3711
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4359554772658551e-06,
+      "loss": 2.2591,
+      "step": 3712
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4333751612956908e-06,
+      "loss": 2.2933,
+      "step": 3713
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.430796777775831e-06,
+      "loss": 2.2897,
+      "step": 3714
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4282203281032802e-06,
+      "loss": 2.2857,
+      "step": 3715
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4256458136739998e-06,
+      "loss": 2.2712,
+      "step": 3716
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.423073235882897e-06,
+      "loss": 2.2105,
+      "step": 3717
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4205025961238333e-06,
+      "loss": 2.2544,
+      "step": 3718
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4179338957896215e-06,
+      "loss": 2.1603,
+      "step": 3719
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.415367136272019e-06,
+      "loss": 2.2565,
+      "step": 3720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4128023189617352e-06,
+      "loss": 2.3259,
+      "step": 3721
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4102394452484275e-06,
+      "loss": 2.2496,
+      "step": 3722
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4076785165206962e-06,
+      "loss": 2.2448,
+      "step": 3723
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4051195341660939e-06,
+      "loss": 2.2632,
+      "step": 3724
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4025624995711102e-06,
+      "loss": 2.1935,
+      "step": 3725
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.4000074141211907e-06,
+      "loss": 2.1981,
+      "step": 3726
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3974542792007157e-06,
+      "loss": 2.2256,
+      "step": 3727
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3949030961930077e-06,
+      "loss": 2.26,
+      "step": 3728
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3923538664803432e-06,
+      "loss": 2.3222,
+      "step": 3729
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.389806591443927e-06,
+      "loss": 2.3,
+      "step": 3730
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.387261272463914e-06,
+      "loss": 2.2872,
+      "step": 3731
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3847179109193927e-06,
+      "loss": 2.3373,
+      "step": 3732
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3821765081883954e-06,
+      "loss": 2.279,
+      "step": 3733
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3796370656478936e-06,
+      "loss": 2.3023,
+      "step": 3734
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3770995846737907e-06,
+      "loss": 2.2287,
+      "step": 3735
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3745640666409372e-06,
+      "loss": 2.2396,
+      "step": 3736
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3720305129231114e-06,
+      "loss": 2.3013,
+      "step": 3737
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.369498924893029e-06,
+      "loss": 2.3534,
+      "step": 3738
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3669693039223443e-06,
+      "loss": 2.171,
+      "step": 3739
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3644416513816417e-06,
+      "loss": 2.2164,
+      "step": 3740
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3619159686404437e-06,
+      "loss": 2.2891,
+      "step": 3741
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3593922570671996e-06,
+      "loss": 2.1928,
+      "step": 3742
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.3568705180292945e-06,
+      "loss": 2.2542,
+      "step": 3743
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3543507528930471e-06,
+      "loss": 2.2702,
+      "step": 3744
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3518329630236987e-06,
+      "loss": 2.1602,
+      "step": 3745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3493171497854284e-06,
+      "loss": 2.2828,
+      "step": 3746
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3468033145413416e-06,
+      "loss": 2.25,
+      "step": 3747
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.344291458653469e-06,
+      "loss": 2.2737,
+      "step": 3748
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.341781583482774e-06,
+      "loss": 2.2319,
+      "step": 3749
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3392736903891406e-06,
+      "loss": 2.3094,
+      "step": 3750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3367677807313878e-06,
+      "loss": 2.2594,
+      "step": 3751
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3342638558672505e-06,
+      "loss": 2.2765,
+      "step": 3752
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3317619171533936e-06,
+      "loss": 2.2161,
+      "step": 3753
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.329261965945406e-06,
+      "loss": 2.3016,
+      "step": 3754
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3267640035977957e-06,
+      "loss": 2.355,
+      "step": 3755
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3242680314639995e-06,
+      "loss": 2.2969,
+      "step": 3756
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3217740508963678e-06,
+      "loss": 2.3086,
+      "step": 3757
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3192820632461783e-06,
+      "loss": 2.2025,
+      "step": 3758
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3167920698636288e-06,
+      "loss": 2.3246,
+      "step": 3759
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.31430407209783e-06,
+      "loss": 2.3247,
+      "step": 3760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.311818071296822e-06,
+      "loss": 2.3298,
+      "step": 3761
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.309334068807554e-06,
+      "loss": 2.2972,
+      "step": 3762
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3068520659758938e-06,
+      "loss": 2.203,
+      "step": 3763
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.304372064146629e-06,
+      "loss": 2.2596,
+      "step": 3764
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.3018940646634615e-06,
+      "loss": 2.2008,
+      "step": 3765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2994180688690095e-06,
+      "loss": 2.3423,
+      "step": 3766
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2969440781048015e-06,
+      "loss": 2.2762,
+      "step": 3767
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2944720937112838e-06,
+      "loss": 2.3192,
+      "step": 3768
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2920021170278157e-06,
+      "loss": 2.2927,
+      "step": 3769
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2895341493926644e-06,
+      "loss": 2.2707,
+      "step": 3770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2870681921430128e-06,
+      "loss": 2.2348,
+      "step": 3771
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.284604246614955e-06,
+      "loss": 2.3142,
+      "step": 3772
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2821423141434897e-06,
+      "loss": 2.2189,
+      "step": 3773
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.279682396062532e-06,
+      "loss": 2.1482,
+      "step": 3774
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2772244937048973e-06,
+      "loss": 2.2835,
+      "step": 3775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2747686084023192e-06,
+      "loss": 2.2547,
+      "step": 3776
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2723147414854292e-06,
+      "loss": 2.2231,
+      "step": 3777
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2698628942837698e-06,
+      "loss": 2.2139,
+      "step": 3778
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2674130681257896e-06,
+      "loss": 2.2868,
+      "step": 3779
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2649652643388383e-06,
+      "loss": 2.2466,
+      "step": 3780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2625194842491746e-06,
+      "loss": 2.2668,
+      "step": 3781
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.260075729181956e-06,
+      "loss": 2.2543,
+      "step": 3782
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.257634000461247e-06,
+      "loss": 2.257,
+      "step": 3783
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2551942994100135e-06,
+      "loss": 2.2916,
+      "step": 3784
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2527566273501175e-06,
+      "loss": 2.256,
+      "step": 3785
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2503209856023318e-06,
+      "loss": 2.272,
+      "step": 3786
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.247887375486319e-06,
+      "loss": 2.1815,
+      "step": 3787
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2454557983206479e-06,
+      "loss": 2.2586,
+      "step": 3788
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.24302625542278e-06,
+      "loss": 2.2395,
+      "step": 3789
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.2405987481090798e-06,
+      "loss": 2.3214,
+      "step": 3790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.238173277694808e-06,
+      "loss": 2.2539,
+      "step": 3791
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2357498454941175e-06,
+      "loss": 2.2669,
+      "step": 3792
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2333284528200613e-06,
+      "loss": 2.2891,
+      "step": 3793
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.230909100984587e-06,
+      "loss": 2.2311,
+      "step": 3794
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2284917912985333e-06,
+      "loss": 2.1801,
+      "step": 3795
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2260765250716356e-06,
+      "loss": 2.2071,
+      "step": 3796
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2236633036125217e-06,
+      "loss": 2.2934,
+      "step": 3797
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2212521282287093e-06,
+      "loss": 2.2701,
+      "step": 3798
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2188430002266105e-06,
+      "loss": 2.2534,
+      "step": 3799
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2164359209115235e-06,
+      "loss": 2.3389,
+      "step": 3800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2140308915876449e-06,
+      "loss": 2.2764,
+      "step": 3801
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2116279135580506e-06,
+      "loss": 2.2668,
+      "step": 3802
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2092269881247115e-06,
+      "loss": 2.348,
+      "step": 3803
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2068281165884865e-06,
+      "loss": 2.3053,
+      "step": 3804
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2044313002491165e-06,
+      "loss": 2.286,
+      "step": 3805
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.2020365404052353e-06,
+      "loss": 2.2647,
+      "step": 3806
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1996438383543564e-06,
+      "loss": 2.2552,
+      "step": 3807
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1972531953928824e-06,
+      "loss": 2.2873,
+      "step": 3808
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1948646128161007e-06,
+      "loss": 2.3202,
+      "step": 3809
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1924780919181772e-06,
+      "loss": 2.2691,
+      "step": 3810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1900936339921692e-06,
+      "loss": 2.2028,
+      "step": 3811
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1877112403300078e-06,
+      "loss": 2.1685,
+      "step": 3812
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1853309122225115e-06,
+      "loss": 2.2323,
+      "step": 3813
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1829526509593753e-06,
+      "loss": 2.3758,
+      "step": 3814
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1805764578291773e-06,
+      "loss": 2.2309,
+      "step": 3815
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1782023341193755e-06,
+      "loss": 2.2337,
+      "step": 3816
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.175830281116303e-06,
+      "loss": 2.3674,
+      "step": 3817
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1734603001051753e-06,
+      "loss": 2.2761,
+      "step": 3818
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1710923923700833e-06,
+      "loss": 2.3442,
+      "step": 3819
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1687265591939928e-06,
+      "loss": 2.3499,
+      "step": 3820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1663628018587487e-06,
+      "loss": 2.3709,
+      "step": 3821
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.164001121645069e-06,
+      "loss": 2.2829,
+      "step": 3822
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1616415198325497e-06,
+      "loss": 2.2863,
+      "step": 3823
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1592839976996557e-06,
+      "loss": 2.3105,
+      "step": 3824
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1569285565237255e-06,
+      "loss": 2.2424,
+      "step": 3825
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1545751975809772e-06,
+      "loss": 2.3301,
+      "step": 3826
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1522239221464915e-06,
+      "loss": 2.3776,
+      "step": 3827
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1498747314942254e-06,
+      "loss": 2.2652,
+      "step": 3828
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1475276268970075e-06,
+      "loss": 2.3341,
+      "step": 3829
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1451826096265306e-06,
+      "loss": 2.2892,
+      "step": 3830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1428396809533631e-06,
+      "loss": 2.2254,
+      "step": 3831
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1404988421469348e-06,
+      "loss": 2.3484,
+      "step": 3832
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1381600944755493e-06,
+      "loss": 2.2842,
+      "step": 3833
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.135823439206376e-06,
+      "loss": 2.3441,
+      "step": 3834
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1334888776054454e-06,
+      "loss": 2.2193,
+      "step": 3835
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1311564109376622e-06,
+      "loss": 2.2813,
+      "step": 3836
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1288260404667883e-06,
+      "loss": 2.228,
+      "step": 3837
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.1264977674554555e-06,
+      "loss": 2.1464,
+      "step": 3838
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1241715931651542e-06,
+      "loss": 2.2355,
+      "step": 3839
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.121847518856241e-06,
+      "loss": 2.3133,
+      "step": 3840
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1195255457879362e-06,
+      "loss": 2.2826,
+      "step": 3841
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1172056752183153e-06,
+      "loss": 2.2896,
+      "step": 3842
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1148879084043207e-06,
+      "loss": 2.1827,
+      "step": 3843
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1125722466017547e-06,
+      "loss": 2.2789,
+      "step": 3844
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1102586910652736e-06,
+      "loss": 2.2392,
+      "step": 3845
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1079472430483973e-06,
+      "loss": 2.2913,
+      "step": 3846
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1056379038035041e-06,
+      "loss": 2.3011,
+      "step": 3847
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1033306745818284e-06,
+      "loss": 2.2821,
+      "step": 3848
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.1010255566334605e-06,
+      "loss": 2.3816,
+      "step": 3849
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0987225512073447e-06,
+      "loss": 2.2325,
+      "step": 3850
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0964216595512883e-06,
+      "loss": 2.2918,
+      "step": 3851
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0941228829119456e-06,
+      "loss": 2.313,
+      "step": 3852
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0918262225348286e-06,
+      "loss": 2.2902,
+      "step": 3853
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0895316796643042e-06,
+      "loss": 2.2222,
+      "step": 3854
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0872392555435857e-06,
+      "loss": 2.2721,
+      "step": 3855
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.084948951414746e-06,
+      "loss": 2.2904,
+      "step": 3856
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0826607685187018e-06,
+      "loss": 2.3667,
+      "step": 3857
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0803747080952304e-06,
+      "loss": 2.1898,
+      "step": 3858
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0780907713829497e-06,
+      "loss": 2.281,
+      "step": 3859
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0758089596193284e-06,
+      "loss": 2.2587,
+      "step": 3860
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0735292740406911e-06,
+      "loss": 2.3328,
+      "step": 3861
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0712517158822016e-06,
+      "loss": 2.2558,
+      "step": 3862
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0689762863778775e-06,
+      "loss": 2.2889,
+      "step": 3863
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.066702986760577e-06,
+      "loss": 2.266,
+      "step": 3864
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0644318182620095e-06,
+      "loss": 2.2793,
+      "step": 3865
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.062162782112729e-06,
+      "loss": 2.2522,
+      "step": 3866
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0598958795421305e-06,
+      "loss": 2.3018,
+      "step": 3867
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.057631111778456e-06,
+      "loss": 2.2691,
+      "step": 3868
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0553684800487924e-06,
+      "loss": 2.1994,
+      "step": 3869
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0531079855790643e-06,
+      "loss": 2.2953,
+      "step": 3870
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.050849629594043e-06,
+      "loss": 2.3677,
+      "step": 3871
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0485934133173387e-06,
+      "loss": 2.2398,
+      "step": 3872
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0463393379714038e-06,
+      "loss": 2.188,
+      "step": 3873
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0440874047775296e-06,
+      "loss": 2.2847,
+      "step": 3874
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0418376149558434e-06,
+      "loss": 2.1982,
+      "step": 3875
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.039589969725321e-06,
+      "loss": 2.1762,
+      "step": 3876
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0373444703037644e-06,
+      "loss": 2.3886,
+      "step": 3877
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0351011179078213e-06,
+      "loss": 2.2531,
+      "step": 3878
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0328599137529744e-06,
+      "loss": 2.3313,
+      "step": 3879
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0306208590535383e-06,
+      "loss": 2.1931,
+      "step": 3880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0283839550226688e-06,
+      "loss": 2.3026,
+      "step": 3881
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0261492028723497e-06,
+      "loss": 2.3218,
+      "step": 3882
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0239166038134092e-06,
+      "loss": 2.3528,
+      "step": 3883
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0216861590554982e-06,
+      "loss": 2.272,
+      "step": 3884
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.019457869807104e-06,
+      "loss": 2.2546,
+      "step": 3885
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.0172317372755508e-06,
+      "loss": 2.1821,
+      "step": 3886
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0150077626669868e-06,
+      "loss": 2.3171,
+      "step": 3887
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.012785947186397e-06,
+      "loss": 2.2554,
+      "step": 3888
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0105662920375913e-06,
+      "loss": 2.2254,
+      "step": 3889
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0083487984232132e-06,
+      "loss": 2.1963,
+      "step": 3890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0061334675447353e-06,
+      "loss": 2.2652,
+      "step": 3891
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0039203006024528e-06,
+      "loss": 2.211,
+      "step": 3892
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.0017092987954969e-06,
+      "loss": 2.2004,
+      "step": 3893
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.995004633218198e-07,
+      "loss": 2.3378,
+      "step": 3894
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.972937953781985e-07,
+      "loss": 2.3051,
+      "step": 3895
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.95089296160241e-07,
+      "loss": 2.2423,
+      "step": 3896
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.928869668623776e-07,
+      "loss": 2.218,
+      "step": 3897
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.906868086778643e-07,
+      "loss": 2.2343,
+      "step": 3898
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.88488822798776e-07,
+      "loss": 2.2423,
+      "step": 3899
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.862930104160162e-07,
+      "loss": 2.2424,
+      "step": 3900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.8409937271931e-07,
+      "loss": 2.238,
+      "step": 3901
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.819079108972001e-07,
+      "loss": 2.2235,
+      "step": 3902
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.797186261370545e-07,
+      "loss": 2.2616,
+      "step": 3903
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.775315196250612e-07,
+      "loss": 2.3636,
+      "step": 3904
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.75346592546224e-07,
+      "loss": 2.2176,
+      "step": 3905
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.73163846084373e-07,
+      "loss": 2.2844,
+      "step": 3906
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.70983281422147e-07,
+      "loss": 2.2771,
+      "step": 3907
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.688048997410143e-07,
+      "loss": 2.3006,
+      "step": 3908
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.666287022212517e-07,
+      "loss": 2.3278,
+      "step": 3909
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.644546900419533e-07,
+      "loss": 2.2703,
+      "step": 3910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.622828643810356e-07,
+      "loss": 2.3495,
+      "step": 3911
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.601132264152224e-07,
+      "loss": 2.2325,
+      "step": 3912
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.579457773200585e-07,
+      "loss": 2.2497,
+      "step": 3913
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.55780518269897e-07,
+      "loss": 2.3026,
+      "step": 3914
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.536174504379087e-07,
+      "loss": 2.1845,
+      "step": 3915
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.514565749960781e-07,
+      "loss": 2.2388,
+      "step": 3916
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.492978931151947e-07,
+      "loss": 2.2431,
+      "step": 3917
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.471414059648703e-07,
+      "loss": 2.2835,
+      "step": 3918
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.449871147135181e-07,
+      "loss": 2.2115,
+      "step": 3919
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.428350205283648e-07,
+      "loss": 2.145,
+      "step": 3920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.406851245754478e-07,
+      "loss": 2.2634,
+      "step": 3921
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.38537428019613e-07,
+      "loss": 2.2721,
+      "step": 3922
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.363919320245151e-07,
+      "loss": 2.2536,
+      "step": 3923
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.342486377526134e-07,
+      "loss": 2.3322,
+      "step": 3924
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.32107546365178e-07,
+      "loss": 2.1766,
+      "step": 3925
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.299686590222845e-07,
+      "loss": 2.2752,
+      "step": 3926
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.278319768828126e-07,
+      "loss": 2.3248,
+      "step": 3927
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.256975011044483e-07,
+      "loss": 2.3216,
+      "step": 3928
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.235652328436845e-07,
+      "loss": 2.355,
+      "step": 3929
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.21435173255813e-07,
+      "loss": 2.1771,
+      "step": 3930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.193073234949346e-07,
+      "loss": 2.3469,
+      "step": 3931
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.171816847139447e-07,
+      "loss": 2.3363,
+      "step": 3932
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.150582580645528e-07,
+      "loss": 2.2061,
+      "step": 3933
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.129370446972586e-07,
+      "loss": 2.2651,
+      "step": 3934
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.108180457613675e-07,
+      "loss": 2.2603,
+      "step": 3935
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.087012624049868e-07,
+      "loss": 2.2325,
+      "step": 3936
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.06586695775018e-07,
+      "loss": 2.3757,
+      "step": 3937
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.04474347017168e-07,
+      "loss": 2.3043,
+      "step": 3938
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.02364217275935e-07,
+      "loss": 2.3159,
+      "step": 3939
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.002563076946213e-07,
+      "loss": 2.2902,
+      "step": 3940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.981506194153239e-07,
+      "loss": 2.3104,
+      "step": 3941
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.960471535789322e-07,
+      "loss": 2.242,
+      "step": 3942
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.939459113251408e-07,
+      "loss": 2.2944,
+      "step": 3943
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.918468937924312e-07,
+      "loss": 2.3114,
+      "step": 3944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.897501021180816e-07,
+      "loss": 2.2815,
+      "step": 3945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.876555374381646e-07,
+      "loss": 2.3134,
+      "step": 3946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.855632008875475e-07,
+      "loss": 2.286,
+      "step": 3947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.834730935998909e-07,
+      "loss": 2.3273,
+      "step": 3948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.813852167076431e-07,
+      "loss": 2.2498,
+      "step": 3949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.792995713420472e-07,
+      "loss": 2.2502,
+      "step": 3950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.772161586331396e-07,
+      "loss": 2.1862,
+      "step": 3951
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.751349797097403e-07,
+      "loss": 2.2678,
+      "step": 3952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.730560356994655e-07,
+      "loss": 2.2756,
+      "step": 3953
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.709793277287182e-07,
+      "loss": 2.2564,
+      "step": 3954
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.689048569226877e-07,
+      "loss": 2.2669,
+      "step": 3955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.668326244053554e-07,
+      "loss": 2.2383,
+      "step": 3956
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.647626312994839e-07,
+      "loss": 2.2935,
+      "step": 3957
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.626948787266309e-07,
+      "loss": 2.3401,
+      "step": 3958
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.606293678071321e-07,
+      "loss": 2.2266,
+      "step": 3959
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.585660996601136e-07,
+      "loss": 2.1959,
+      "step": 3960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.565050754034848e-07,
+      "loss": 2.2573,
+      "step": 3961
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.544462961539379e-07,
+      "loss": 2.2641,
+      "step": 3962
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.523897630269518e-07,
+      "loss": 2.2545,
+      "step": 3963
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.50335477136785e-07,
+      "loss": 2.2216,
+      "step": 3964
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.4828343959648e-07,
+      "loss": 2.35,
+      "step": 3965
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.462336515178643e-07,
+      "loss": 2.3283,
+      "step": 3966
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.441861140115382e-07,
+      "loss": 2.371,
+      "step": 3967
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.421408281868937e-07,
+      "loss": 2.2632,
+      "step": 3968
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.400977951520933e-07,
+      "loss": 2.2631,
+      "step": 3969
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.38057016014085e-07,
+      "loss": 2.3169,
+      "step": 3970
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.360184918785902e-07,
+      "loss": 2.3095,
+      "step": 3971
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.339822238501133e-07,
+      "loss": 2.241,
+      "step": 3972
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.319482130319362e-07,
+      "loss": 2.2237,
+      "step": 3973
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.299164605261129e-07,
+      "loss": 2.2957,
+      "step": 3974
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.278869674334788e-07,
+      "loss": 2.3186,
+      "step": 3975
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.258597348536452e-07,
+      "loss": 2.2068,
+      "step": 3976
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.238347638849931e-07,
+      "loss": 2.2399,
+      "step": 3977
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.218120556246833e-07,
+      "loss": 2.2856,
+      "step": 3978
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.197916111686511e-07,
+      "loss": 2.2721,
+      "step": 3979
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.177734316116004e-07,
+      "loss": 2.2596,
+      "step": 3980
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.15757518047014e-07,
+      "loss": 2.2741,
+      "step": 3981
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.137438715671387e-07,
+      "loss": 2.286,
+      "step": 3982
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.117324932630044e-07,
+      "loss": 2.2876,
+      "step": 3983
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.097233842244012e-07,
+      "loss": 2.2455,
+      "step": 3984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.077165455398955e-07,
+      "loss": 2.254,
+      "step": 3985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.057119782968243e-07,
+      "loss": 2.2828,
+      "step": 3986
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.037096835812885e-07,
+      "loss": 2.2459,
+      "step": 3987
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.01709662478164e-07,
+      "loss": 2.2509,
+      "step": 3988
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.997119160710892e-07,
+      "loss": 2.275,
+      "step": 3989
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.977164454424746e-07,
+      "loss": 2.1828,
+      "step": 3990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.957232516734964e-07,
+      "loss": 2.2806,
+      "step": 3991
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.937323358440935e-07,
+      "loss": 2.3079,
+      "step": 3992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.91743699032978e-07,
+      "loss": 2.2608,
+      "step": 3993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.897573423176197e-07,
+      "loss": 2.2676,
+      "step": 3994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.877732667742582e-07,
+      "loss": 2.2507,
+      "step": 3995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.857914734778926e-07,
+      "loss": 2.3413,
+      "step": 3996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.838119635022895e-07,
+      "loss": 2.2674,
+      "step": 3997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.818347379199781e-07,
+      "loss": 2.2087,
+      "step": 3998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.798597978022454e-07,
+      "loss": 2.2672,
+      "step": 3999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.778871442191455e-07,
+      "loss": 2.4144,
+      "step": 4000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.759167782394916e-07,
+      "loss": 2.3341,
+      "step": 4001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.73948700930856e-07,
+      "loss": 2.2492,
+      "step": 4002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.719829133595724e-07,
+      "loss": 2.3002,
+      "step": 4003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.700194165907343e-07,
+      "loss": 2.2407,
+      "step": 4004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.680582116881941e-07,
+      "loss": 2.2198,
+      "step": 4005
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.660992997145611e-07,
+      "loss": 2.3541,
+      "step": 4006
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.641426817312003e-07,
+      "loss": 2.1851,
+      "step": 4007
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.621883587982409e-07,
+      "loss": 2.2526,
+      "step": 4008
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.60236331974561e-07,
+      "loss": 2.2841,
+      "step": 4009
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.582866023177982e-07,
+      "loss": 2.3033,
+      "step": 4010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.563391708843465e-07,
+      "loss": 2.2432,
+      "step": 4011
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.543940387293514e-07,
+      "loss": 2.3183,
+      "step": 4012
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.524512069067158e-07,
+      "loss": 2.2445,
+      "step": 4013
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.505106764690933e-07,
+      "loss": 2.2351,
+      "step": 4014
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.485724484678924e-07,
+      "loss": 2.2645,
+      "step": 4015
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.466365239532758e-07,
+      "loss": 2.311,
+      "step": 4016
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.447029039741527e-07,
+      "loss": 2.3042,
+      "step": 4017
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.427715895781906e-07,
+      "loss": 2.2577,
+      "step": 4018
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.408425818118026e-07,
+      "loss": 2.2213,
+      "step": 4019
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.389158817201541e-07,
+      "loss": 2.2314,
+      "step": 4020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.369914903471592e-07,
+      "loss": 2.2503,
+      "step": 4021
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.35069408735482e-07,
+      "loss": 2.3386,
+      "step": 4022
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.331496379265363e-07,
+      "loss": 2.2215,
+      "step": 4023
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.312321789604803e-07,
+      "loss": 2.2829,
+      "step": 4024
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.293170328762222e-07,
+      "loss": 2.3049,
+      "step": 4025
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.274042007114185e-07,
+      "loss": 2.2326,
+      "step": 4026
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.254936835024673e-07,
+      "loss": 2.2744,
+      "step": 4027
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.235854822845168e-07,
+      "loss": 2.3294,
+      "step": 4028
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.216795980914582e-07,
+      "loss": 2.2018,
+      "step": 4029
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.197760319559288e-07,
+      "loss": 2.2823,
+      "step": 4030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.178747849093093e-07,
+      "loss": 2.2481,
+      "step": 4031
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.159758579817194e-07,
+      "loss": 2.2749,
+      "step": 4032
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.140792522020317e-07,
+      "loss": 2.3015,
+      "step": 4033
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.121849685978521e-07,
+      "loss": 2.2451,
+      "step": 4034
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.102930081955317e-07,
+      "loss": 2.3046,
+      "step": 4035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.084033720201649e-07,
+      "loss": 2.2328,
+      "step": 4036
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.065160610955818e-07,
+      "loss": 2.2944,
+      "step": 4037
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.046310764443576e-07,
+      "loss": 2.2286,
+      "step": 4038
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.02748419087802e-07,
+      "loss": 2.2744,
+      "step": 4039
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.008680900459713e-07,
+      "loss": 2.2723,
+      "step": 4040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.989900903376534e-07,
+      "loss": 2.1591,
+      "step": 4041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.971144209803738e-07,
+      "loss": 2.3409,
+      "step": 4042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.952410829904021e-07,
+      "loss": 2.2987,
+      "step": 4043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.93370077382739e-07,
+      "loss": 2.2002,
+      "step": 4044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.915014051711233e-07,
+      "loss": 2.2809,
+      "step": 4045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.896350673680286e-07,
+      "loss": 2.2553,
+      "step": 4046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.877710649846653e-07,
+      "loss": 2.3011,
+      "step": 4047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.859093990309779e-07,
+      "loss": 2.3636,
+      "step": 4048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.840500705156434e-07,
+      "loss": 2.3271,
+      "step": 4049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.821930804460736e-07,
+      "loss": 2.3833,
+      "step": 4050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.803384298284155e-07,
+      "loss": 2.3389,
+      "step": 4051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.784861196675441e-07,
+      "loss": 2.3012,
+      "step": 4052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.766361509670688e-07,
+      "loss": 2.3694,
+      "step": 4053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.747885247293307e-07,
+      "loss": 2.3287,
+      "step": 4054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.729432419554021e-07,
+      "loss": 2.3731,
+      "step": 4055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.711003036450825e-07,
+      "loss": 2.2935,
+      "step": 4056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.69259710796904e-07,
+      "loss": 2.2306,
+      "step": 4057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.67421464408129e-07,
+      "loss": 2.1987,
+      "step": 4058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.655855654747434e-07,
+      "loss": 2.2433,
+      "step": 4059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.637520149914667e-07,
+      "loss": 2.309,
+      "step": 4060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.619208139517441e-07,
+      "loss": 2.3217,
+      "step": 4061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.60091963347746e-07,
+      "loss": 2.195,
+      "step": 4062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.582654641703723e-07,
+      "loss": 2.334,
+      "step": 4063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.564413174092443e-07,
+      "loss": 2.3751,
+      "step": 4064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.546195240527164e-07,
+      "loss": 2.2495,
+      "step": 4065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.528000850878613e-07,
+      "loss": 2.3175,
+      "step": 4066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.509830015004753e-07,
+      "loss": 2.2176,
+      "step": 4067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.49168274275086e-07,
+      "loss": 2.2152,
+      "step": 4068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.473559043949363e-07,
+      "loss": 2.2646,
+      "step": 4069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.455458928419966e-07,
+      "loss": 2.2565,
+      "step": 4070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.437382405969573e-07,
+      "loss": 2.1857,
+      "step": 4071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.419329486392311e-07,
+      "loss": 2.2871,
+      "step": 4072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.401300179469533e-07,
+      "loss": 2.2485,
+      "step": 4073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.383294494969755e-07,
+      "loss": 2.238,
+      "step": 4074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.365312442648768e-07,
+      "loss": 2.2371,
+      "step": 4075
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.347354032249492e-07,
+      "loss": 2.2612,
+      "step": 4076
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.329419273502047e-07,
+      "loss": 2.2762,
+      "step": 4077
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.311508176123771e-07,
+      "loss": 2.2024,
+      "step": 4078
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.293620749819151e-07,
+      "loss": 2.226,
+      "step": 4079
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.275757004279875e-07,
+      "loss": 2.2784,
+      "step": 4080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.257916949184773e-07,
+      "loss": 2.2648,
+      "step": 4081
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.240100594199849e-07,
+      "loss": 2.3108,
+      "step": 4082
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.22230794897829e-07,
+      "loss": 2.3515,
+      "step": 4083
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.204539023160389e-07,
+      "loss": 2.2384,
+      "step": 4084
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.186793826373627e-07,
+      "loss": 2.2785,
+      "step": 4085
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.16907236823262e-07,
+      "loss": 2.3076,
+      "step": 4086
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.151374658339104e-07,
+      "loss": 2.2717,
+      "step": 4087
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.133700706281975e-07,
+      "loss": 2.2087,
+      "step": 4088
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.116050521637218e-07,
+      "loss": 2.3203,
+      "step": 4089
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.098424113967999e-07,
+      "loss": 2.2978,
+      "step": 4090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.08082149282454e-07,
+      "loss": 2.2376,
+      "step": 4091
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.063242667744219e-07,
+      "loss": 2.3101,
+      "step": 4092
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.045687648251508e-07,
+      "loss": 2.1551,
+      "step": 4093
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.028156443857957e-07,
+      "loss": 2.3001,
+      "step": 4094
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 6.010649064062252e-07,
+      "loss": 2.3138,
+      "step": 4095
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.993165518350136e-07,
+      "loss": 2.2479,
+      "step": 4096
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.97570581619446e-07,
+      "loss": 2.3566,
+      "step": 4097
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.958269967055164e-07,
+      "loss": 2.1507,
+      "step": 4098
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.940857980379211e-07,
+      "loss": 2.2238,
+      "step": 4099
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.923469865600728e-07,
+      "loss": 2.2714,
+      "step": 4100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.906105632140819e-07,
+      "loss": 2.2825,
+      "step": 4101
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.888765289407683e-07,
+      "loss": 2.26,
+      "step": 4102
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.871448846796584e-07,
+      "loss": 2.2912,
+      "step": 4103
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.854156313689824e-07,
+      "loss": 2.1868,
+      "step": 4104
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.836887699456773e-07,
+      "loss": 2.2924,
+      "step": 4105
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.81964301345379e-07,
+      "loss": 2.2329,
+      "step": 4106
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.802422265024316e-07,
+      "loss": 2.1625,
+      "step": 4107
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.785225463498828e-07,
+      "loss": 2.2729,
+      "step": 4108
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.768052618194775e-07,
+      "loss": 2.2287,
+      "step": 4109
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.75090373841668e-07,
+      "loss": 2.2659,
+      "step": 4110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.733778833456061e-07,
+      "loss": 2.3434,
+      "step": 4111
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.716677912591434e-07,
+      "loss": 2.2404,
+      "step": 4112
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.69960098508835e-07,
+      "loss": 2.3136,
+      "step": 4113
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.682548060199311e-07,
+      "loss": 2.2464,
+      "step": 4114
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.665519147163883e-07,
+      "loss": 2.23,
+      "step": 4115
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.648514255208553e-07,
+      "loss": 2.3744,
+      "step": 4116
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.631533393546846e-07,
+      "loss": 2.1562,
+      "step": 4117
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.614576571379237e-07,
+      "loss": 2.246,
+      "step": 4118
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.597643797893176e-07,
+      "loss": 2.3396,
+      "step": 4119
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.580735082263111e-07,
+      "loss": 2.2694,
+      "step": 4120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.563850433650409e-07,
+      "loss": 2.3485,
+      "step": 4121
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.546989861203434e-07,
+      "loss": 2.2473,
+      "step": 4122
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.530153374057512e-07,
+      "loss": 2.2454,
+      "step": 4123
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.513340981334864e-07,
+      "loss": 2.2887,
+      "step": 4124
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.496552692144741e-07,
+      "loss": 2.2804,
+      "step": 4125
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.479788515583256e-07,
+      "loss": 2.3302,
+      "step": 4126
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.463048460733494e-07,
+      "loss": 2.2359,
+      "step": 4127
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.446332536665466e-07,
+      "loss": 2.1693,
+      "step": 4128
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.429640752436105e-07,
+      "loss": 2.2475,
+      "step": 4129
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.412973117089288e-07,
+      "loss": 2.2748,
+      "step": 4130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.396329639655757e-07,
+      "loss": 2.303,
+      "step": 4131
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.37971032915321e-07,
+      "loss": 2.2784,
+      "step": 4132
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.363115194586243e-07,
+      "loss": 2.1817,
+      "step": 4133
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.346544244946334e-07,
+      "loss": 2.2554,
+      "step": 4134
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.329997489211875e-07,
+      "loss": 2.323,
+      "step": 4135
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.313474936348151e-07,
+      "loss": 2.2729,
+      "step": 4136
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.296976595307307e-07,
+      "loss": 2.3448,
+      "step": 4137
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.280502475028415e-07,
+      "loss": 2.245,
+      "step": 4138
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.264052584437368e-07,
+      "loss": 2.3035,
+      "step": 4139
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.247626932446992e-07,
+      "loss": 2.1841,
+      "step": 4140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.231225527956924e-07,
+      "loss": 2.2377,
+      "step": 4141
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.214848379853704e-07,
+      "loss": 2.2464,
+      "step": 4142
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.198495497010714e-07,
+      "loss": 2.1807,
+      "step": 4143
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.182166888288176e-07,
+      "loss": 2.2067,
+      "step": 4144
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.165862562533191e-07,
+      "loss": 2.2596,
+      "step": 4145
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.149582528579667e-07,
+      "loss": 2.2148,
+      "step": 4146
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.133326795248373e-07,
+      "loss": 2.2934,
+      "step": 4147
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.11709537134692e-07,
+      "loss": 2.2276,
+      "step": 4148
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.100888265669701e-07,
+      "loss": 2.3449,
+      "step": 4149
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.084705486998009e-07,
+      "loss": 2.4256,
+      "step": 4150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.068547044099892e-07,
+      "loss": 2.3132,
+      "step": 4151
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.05241294573024e-07,
+      "loss": 2.2524,
+      "step": 4152
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.036303200630733e-07,
+      "loss": 2.2599,
+      "step": 4153
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.020217817529882e-07,
+      "loss": 2.2162,
+      "step": 4154
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.004156805142996e-07,
+      "loss": 2.2335,
+      "step": 4155
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.988120172172151e-07,
+      "loss": 2.2367,
+      "step": 4156
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.972107927306241e-07,
+      "loss": 2.2505,
+      "step": 4157
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.956120079220944e-07,
+      "loss": 2.2873,
+      "step": 4158
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.940156636578703e-07,
+      "loss": 2.2868,
+      "step": 4159
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.924217608028748e-07,
+      "loss": 2.3431,
+      "step": 4160
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.908303002207098e-07,
+      "loss": 2.2311,
+      "step": 4161
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.892412827736503e-07,
+      "loss": 2.2245,
+      "step": 4162
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.876547093226514e-07,
+      "loss": 2.2305,
+      "step": 4163
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.860705807273386e-07,
+      "loss": 2.2319,
+      "step": 4164
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.844888978460216e-07,
+      "loss": 2.2395,
+      "step": 4165
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.829096615356755e-07,
+      "loss": 2.2329,
+      "step": 4166
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.813328726519551e-07,
+      "loss": 2.325,
+      "step": 4167
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.797585320491893e-07,
+      "loss": 2.2412,
+      "step": 4168
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.781866405803775e-07,
+      "loss": 2.2691,
+      "step": 4169
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.766171990971957e-07,
+      "loss": 2.3855,
+      "step": 4170
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.750502084499886e-07,
+      "loss": 2.1813,
+      "step": 4171
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.734856694877754e-07,
+      "loss": 2.2556,
+      "step": 4172
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7192358305824805e-07,
+      "loss": 2.3881,
+      "step": 4173
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.7036395000776556e-07,
+      "loss": 2.1774,
+      "step": 4174
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.6880677118136374e-07,
+      "loss": 2.3022,
+      "step": 4175
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.672520474227421e-07,
+      "loss": 2.2727,
+      "step": 4176
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.6569977957427514e-07,
+      "loss": 2.2524,
+      "step": 4177
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.6414996847700287e-07,
+      "loss": 2.2205,
+      "step": 4178
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.6260261497063695e-07,
+      "loss": 2.2224,
+      "step": 4179
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.6105771989355783e-07,
+      "loss": 2.2321,
+      "step": 4180
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.595152840828099e-07,
+      "loss": 2.3377,
+      "step": 4181
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.57975308374109e-07,
+      "loss": 2.1816,
+      "step": 4182
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5643779360183773e-07,
+      "loss": 2.2827,
+      "step": 4183
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5490274059904247e-07,
+      "loss": 2.275,
+      "step": 4184
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5337015019743915e-07,
+      "loss": 2.1685,
+      "step": 4185
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.5184002322740784e-07,
+      "loss": 2.2064,
+      "step": 4186
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.503123605179949e-07,
+      "loss": 2.3348,
+      "step": 4187
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.487871628969093e-07,
+      "loss": 2.2958,
+      "step": 4188
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.47264431190525e-07,
+      "loss": 2.2736,
+      "step": 4189
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.457441662238832e-07,
+      "loss": 2.2749,
+      "step": 4190
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.442263688206838e-07,
+      "loss": 2.3027,
+      "step": 4191
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4271103980329247e-07,
+      "loss": 2.2232,
+      "step": 4192
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.4119817999273894e-07,
+      "loss": 2.2436,
+      "step": 4193
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.396877902087099e-07,
+      "loss": 2.2009,
+      "step": 4194
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3817987126956006e-07,
+      "loss": 2.2716,
+      "step": 4195
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3667442399229985e-07,
+      "loss": 2.2963,
+      "step": 4196
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3517144919260345e-07,
+      "loss": 2.18,
+      "step": 4197
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3367094768480664e-07,
+      "loss": 2.2819,
+      "step": 4198
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.321729202819008e-07,
+      "loss": 2.171,
+      "step": 4199
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.3067736779554226e-07,
+      "loss": 2.1613,
+      "step": 4200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.291842910360411e-07,
+      "loss": 2.297,
+      "step": 4201
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.276936908123702e-07,
+      "loss": 2.2703,
+      "step": 4202
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.262055679321575e-07,
+      "loss": 2.3318,
+      "step": 4203
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2471992320169077e-07,
+      "loss": 2.2549,
+      "step": 4204
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2323675742591555e-07,
+      "loss": 2.2471,
+      "step": 4205
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.217560714084307e-07,
+      "loss": 2.2924,
+      "step": 4206
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.2027786595149555e-07,
+      "loss": 2.2513,
+      "step": 4207
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.18802141856024e-07,
+      "loss": 2.2823,
+      "step": 4208
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.173288999215835e-07,
+      "loss": 2.1917,
+      "step": 4209
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1585814094639965e-07,
+      "loss": 2.1978,
+      "step": 4210
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.14389865727351e-07,
+      "loss": 2.2245,
+      "step": 4211
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1292407505997277e-07,
+      "loss": 2.287,
+      "step": 4212
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.114607697384504e-07,
+      "loss": 2.2579,
+      "step": 4213
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0999995055562235e-07,
+      "loss": 2.2745,
+      "step": 4214
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0854161830298667e-07,
+      "loss": 2.2723,
+      "step": 4215
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0708577377068626e-07,
+      "loss": 2.3537,
+      "step": 4216
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.056324177475202e-07,
+      "loss": 2.2624,
+      "step": 4217
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.041815510209396e-07,
+      "loss": 2.2663,
+      "step": 4218
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0273317437704293e-07,
+      "loss": 2.294,
+      "step": 4219
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0128728860058477e-07,
+      "loss": 2.2537,
+      "step": 4220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.9984389447496494e-07,
+      "loss": 2.3131,
+      "step": 4221
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.984029927822386e-07,
+      "loss": 2.2967,
+      "step": 4222
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.969645843031061e-07,
+      "loss": 2.2648,
+      "step": 4223
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.955286698169175e-07,
+      "loss": 2.2807,
+      "step": 4224
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.940952501016754e-07,
+      "loss": 2.3028,
+      "step": 4225
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.926643259340251e-07,
+      "loss": 2.358,
+      "step": 4226
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.91235898089265e-07,
+      "loss": 2.266,
+      "step": 4227
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.8980996734133634e-07,
+      "loss": 2.2868,
+      "step": 4228
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.883865344628307e-07,
+      "loss": 2.2389,
+      "step": 4229
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.869656002249861e-07,
+      "loss": 2.3638,
+      "step": 4230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.8554716539768334e-07,
+      "loss": 2.3092,
+      "step": 4231
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.841312307494532e-07,
+      "loss": 2.2897,
+      "step": 4232
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.8271779704746993e-07,
+      "loss": 2.3134,
+      "step": 4233
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.813068650575524e-07,
+      "loss": 2.2087,
+      "step": 4234
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.798984355441648e-07,
+      "loss": 2.2403,
+      "step": 4235
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.7849250927041505e-07,
+      "loss": 2.2653,
+      "step": 4236
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.770890869980559e-07,
+      "loss": 2.2276,
+      "step": 4237
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.756881694874809e-07,
+      "loss": 2.2157,
+      "step": 4238
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.7428975749772833e-07,
+      "loss": 2.2663,
+      "step": 4239
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.728938517864794e-07,
+      "loss": 2.2242,
+      "step": 4240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.715004531100552e-07,
+      "loss": 2.271,
+      "step": 4241
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.7010956222342074e-07,
+      "loss": 2.254,
+      "step": 4242
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.687211798801821e-07,
+      "loss": 2.2123,
+      "step": 4243
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6733530683258354e-07,
+      "loss": 2.3535,
+      "step": 4244
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.659519438315134e-07,
+      "loss": 2.3124,
+      "step": 4245
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.645710916264961e-07,
+      "loss": 2.2631,
+      "step": 4246
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.631927509657013e-07,
+      "loss": 2.2128,
+      "step": 4247
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.618169225959317e-07,
+      "loss": 2.263,
+      "step": 4248
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.6044360726263106e-07,
+      "loss": 2.3131,
+      "step": 4249
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5907280570988525e-07,
+      "loss": 2.3283,
+      "step": 4250
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.577045186804118e-07,
+      "loss": 2.2444,
+      "step": 4251
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.563387469155705e-07,
+      "loss": 2.2157,
+      "step": 4252
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5497549115535603e-07,
+      "loss": 2.2896,
+      "step": 4253
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5361475213840093e-07,
+      "loss": 2.3304,
+      "step": 4254
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5225653060197426e-07,
+      "loss": 2.2909,
+      "step": 4255
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.509008272819786e-07,
+      "loss": 2.1893,
+      "step": 4256
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.495476429129574e-07,
+      "loss": 2.255,
+      "step": 4257
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4819697822808475e-07,
+      "loss": 2.2474,
+      "step": 4258
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.468488339591691e-07,
+      "loss": 2.2508,
+      "step": 4259
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.455032108366568e-07,
+      "loss": 2.3038,
+      "step": 4260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.441601095896258e-07,
+      "loss": 2.3292,
+      "step": 4261
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4281953094578877e-07,
+      "loss": 2.2982,
+      "step": 4262
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.414814756314905e-07,
+      "loss": 2.3014,
+      "step": 4263
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4014594437170976e-07,
+      "loss": 2.3091,
+      "step": 4264
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.388129378900579e-07,
+      "loss": 2.2459,
+      "step": 4265
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.374824569087759e-07,
+      "loss": 2.2688,
+      "step": 4266
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.361545021487389e-07,
+      "loss": 2.2166,
+      "step": 4267
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3482907432945353e-07,
+      "loss": 2.2685,
+      "step": 4268
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.335061741690543e-07,
+      "loss": 2.2694,
+      "step": 4269
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.3218580238431e-07,
+      "loss": 2.2448,
+      "step": 4270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.3086795969061556e-07,
+      "loss": 2.2069,
+      "step": 4271
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.295526468020005e-07,
+      "loss": 2.1893,
+      "step": 4272
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.2823986443111856e-07,
+      "loss": 2.258,
+      "step": 4273
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.269296132892552e-07,
+      "loss": 2.2364,
+      "step": 4274
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.2562189408632596e-07,
+      "loss": 2.2326,
+      "step": 4275
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.2431670753086954e-07,
+      "loss": 2.3045,
+      "step": 4276
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.230140543300575e-07,
+      "loss": 2.3016,
+      "step": 4277
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.217139351896853e-07,
+      "loss": 2.3304,
+      "step": 4278
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.2041635081417723e-07,
+      "loss": 2.2617,
+      "step": 4279
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.191213019065853e-07,
+      "loss": 2.3475,
+      "step": 4280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.178287891685827e-07,
+      "loss": 2.3402,
+      "step": 4281
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.1653881330047656e-07,
+      "loss": 2.2865,
+      "step": 4282
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.1525137500119207e-07,
+      "loss": 2.3197,
+      "step": 4283
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.1396647496828245e-07,
+      "loss": 2.2389,
+      "step": 4284
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.1268411389792587e-07,
+      "loss": 2.247,
+      "step": 4285
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.114042924849248e-07,
+      "loss": 2.279,
+      "step": 4286
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.1012701142270676e-07,
+      "loss": 2.2483,
+      "step": 4287
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0885227140331987e-07,
+      "loss": 2.2335,
+      "step": 4288
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.075800731174372e-07,
+      "loss": 2.2056,
+      "step": 4289
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.063104172543563e-07,
+      "loss": 2.2452,
+      "step": 4290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.050433045019935e-07,
+      "loss": 2.2419,
+      "step": 4291
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.037787355468902e-07,
+      "loss": 2.2395,
+      "step": 4292
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0251671107421e-07,
+      "loss": 2.2,
+      "step": 4293
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0125723176773425e-07,
+      "loss": 2.3265,
+      "step": 4294
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0000029830986933e-07,
+      "loss": 2.294,
+      "step": 4295
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9874591138163767e-07,
+      "loss": 2.2193,
+      "step": 4296
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.97494071662689e-07,
+      "loss": 2.318,
+      "step": 4297
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.962447798312851e-07,
+      "loss": 2.3342,
+      "step": 4298
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9499803656431236e-07,
+      "loss": 2.3056,
+      "step": 4299
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9375384253727544e-07,
+      "loss": 2.2571,
+      "step": 4300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.925121984242951e-07,
+      "loss": 2.2453,
+      "step": 4301
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.912731048981149e-07,
+      "loss": 2.3358,
+      "step": 4302
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.900365626300922e-07,
+      "loss": 2.2906,
+      "step": 4303
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8880257229020434e-07,
+      "loss": 2.2631,
+      "step": 4304
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8757113454704666e-07,
+      "loss": 2.3534,
+      "step": 4305
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8634225006782867e-07,
+      "loss": 2.2602,
+      "step": 4306
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.851159195183806e-07,
+      "loss": 2.2599,
+      "step": 4307
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8389214356314475e-07,
+      "loss": 2.2522,
+      "step": 4308
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8267092286518194e-07,
+      "loss": 2.3001,
+      "step": 4309
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.814522580861667e-07,
+      "loss": 2.282,
+      "step": 4310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8023614988639037e-07,
+      "loss": 2.2029,
+      "step": 4311
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7902259892475915e-07,
+      "loss": 2.2974,
+      "step": 4312
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7781160585879153e-07,
+      "loss": 2.2596,
+      "step": 4313
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7660317134462265e-07,
+      "loss": 2.296,
+      "step": 4314
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7539729603699996e-07,
+      "loss": 2.2773,
+      "step": 4315
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7419398058928403e-07,
+      "loss": 2.2859,
+      "step": 4316
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7299322565344956e-07,
+      "loss": 2.2395,
+      "step": 4317
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.717950318800838e-07,
+      "loss": 2.3331,
+      "step": 4318
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7059939991838535e-07,
+      "loss": 2.2628,
+      "step": 4319
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.694063304161659e-07,
+      "loss": 2.253,
+      "step": 4320
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6821582401984624e-07,
+      "loss": 2.2816,
+      "step": 4321
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6702788137446365e-07,
+      "loss": 2.1857,
+      "step": 4322
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6584250312366054e-07,
+      "loss": 2.2538,
+      "step": 4323
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.646596899096937e-07,
+      "loss": 2.253,
+      "step": 4324
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6347944237342937e-07,
+      "loss": 2.2643,
+      "step": 4325
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6230176115434104e-07,
+      "loss": 2.3277,
+      "step": 4326
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6112664689051657e-07,
+      "loss": 2.2621,
+      "step": 4327
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.599541002186479e-07,
+      "loss": 2.2885,
+      "step": 4328
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.587841217740394e-07,
+      "loss": 2.2127,
+      "step": 4329
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.576167121906037e-07,
+      "loss": 2.2478,
+      "step": 4330
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.5645187210085797e-07,
+      "loss": 2.4211,
+      "step": 4331
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.5528960213593247e-07,
+      "loss": 2.1661,
+      "step": 4332
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.541299029255612e-07,
+      "loss": 2.3563,
+      "step": 4333
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.5297277509808703e-07,
+      "loss": 2.2267,
+      "step": 4334
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.51818219280458e-07,
+      "loss": 2.3083,
+      "step": 4335
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.5066623609823037e-07,
+      "loss": 2.2372,
+      "step": 4336
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4951682617556595e-07,
+      "loss": 2.2543,
+      "step": 4337
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.48369990135231e-07,
+      "loss": 2.2771,
+      "step": 4338
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4722572859859907e-07,
+      "loss": 2.2651,
+      "step": 4339
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4608404218564865e-07,
+      "loss": 2.2347,
+      "step": 4340
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.44944931514961e-07,
+      "loss": 2.3734,
+      "step": 4341
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.438083972037242e-07,
+      "loss": 2.2427,
+      "step": 4342
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4267443986772887e-07,
+      "loss": 2.2823,
+      "step": 4343
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4154306012137084e-07,
+      "loss": 2.2594,
+      "step": 4344
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.404142585776481e-07,
+      "loss": 2.2785,
+      "step": 4345
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.392880358481603e-07,
+      "loss": 2.2884,
+      "step": 4346
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.381643925431143e-07,
+      "loss": 2.2435,
+      "step": 4347
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3704332927131534e-07,
+      "loss": 2.2315,
+      "step": 4348
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.35924846640172e-07,
+      "loss": 2.3256,
+      "step": 4349
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3480894525569564e-07,
+      "loss": 2.3053,
+      "step": 4350
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.336956257224976e-07,
+      "loss": 2.3835,
+      "step": 4351
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3258488864379147e-07,
+      "loss": 2.2461,
+      "step": 4352
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3147673462139032e-07,
+      "loss": 2.289,
+      "step": 4353
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3037116425570827e-07,
+      "loss": 2.2667,
+      "step": 4354
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2926817814576174e-07,
+      "loss": 2.27,
+      "step": 4355
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2816777688916158e-07,
+      "loss": 2.2422,
+      "step": 4356
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2706996108212477e-07,
+      "loss": 2.282,
+      "step": 4357
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2597473131946267e-07,
+      "loss": 2.2318,
+      "step": 4358
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.248820881945879e-07,
+      "loss": 2.185,
+      "step": 4359
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2379203229950964e-07,
+      "loss": 2.3089,
+      "step": 4360
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2270456422483655e-07,
+      "loss": 2.2566,
+      "step": 4361
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2161968455977622e-07,
+      "loss": 2.2516,
+      "step": 4362
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2053739389213124e-07,
+      "loss": 2.3766,
+      "step": 4363
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1945769280830252e-07,
+      "loss": 2.2765,
+      "step": 4364
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.183805818932899e-07,
+      "loss": 2.2694,
+      "step": 4365
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1730606173068658e-07,
+      "loss": 2.2791,
+      "step": 4366
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1623413290268347e-07,
+      "loss": 2.2526,
+      "step": 4367
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1516479599006823e-07,
+      "loss": 2.4178,
+      "step": 4368
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1409805157222353e-07,
+      "loss": 2.2519,
+      "step": 4369
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.13033900227127e-07,
+      "loss": 2.1854,
+      "step": 4370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.119723425313508e-07,
+      "loss": 2.3523,
+      "step": 4371
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.109133790600648e-07,
+      "loss": 2.3095,
+      "step": 4372
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.09857010387029e-07,
+      "loss": 2.2418,
+      "step": 4373
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.088032370846005e-07,
+      "loss": 2.1928,
+      "step": 4374
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0775205972373036e-07,
+      "loss": 2.2367,
+      "step": 4375
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0670347887396024e-07,
+      "loss": 2.3487,
+      "step": 4376
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.056574951034279e-07,
+      "loss": 2.2037,
+      "step": 4377
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0461410897886225e-07,
+      "loss": 2.2707,
+      "step": 4378
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0357332106558602e-07,
+      "loss": 2.2558,
+      "step": 4379
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0253513192751374e-07,
+      "loss": 2.3239,
+      "step": 4380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0149954212714984e-07,
+      "loss": 2.2328,
+      "step": 4381
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.0046655222559498e-07,
+      "loss": 2.3344,
+      "step": 4382
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.994361627825364e-07,
+      "loss": 2.2921,
+      "step": 4383
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9840837435625594e-07,
+      "loss": 2.2818,
+      "step": 4384
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9738318750362306e-07,
+      "loss": 2.3433,
+      "step": 4385
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9636060278010017e-07,
+      "loss": 2.2973,
+      "step": 4386
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9534062073973902e-07,
+      "loss": 2.2721,
+      "step": 4387
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9432324193518037e-07,
+      "loss": 2.3432,
+      "step": 4388
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9330846691765492e-07,
+      "loss": 2.2676,
+      "step": 4389
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9229629623698452e-07,
+      "loss": 2.1689,
+      "step": 4390
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9128673044157607e-07,
+      "loss": 2.3435,
+      "step": 4391
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9027977007842868e-07,
+      "loss": 2.2219,
+      "step": 4392
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8927541569312812e-07,
+      "loss": 2.2579,
+      "step": 4393
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8827366782984913e-07,
+      "loss": 2.2113,
+      "step": 4394
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8727452703135308e-07,
+      "loss": 2.3008,
+      "step": 4395
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8627799383898748e-07,
+      "loss": 2.3248,
+      "step": 4396
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8528406879269212e-07,
+      "loss": 2.2485,
+      "step": 4397
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8429275243098844e-07,
+      "loss": 2.2421,
+      "step": 4398
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8330404529098622e-07,
+      "loss": 2.2651,
+      "step": 4399
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8231794790838254e-07,
+      "loss": 2.3504,
+      "step": 4400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8133446081745887e-07,
+      "loss": 2.2287,
+      "step": 4401
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8035358455108398e-07,
+      "loss": 2.2728,
+      "step": 4402
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7937531964070943e-07,
+      "loss": 2.2979,
+      "step": 4403
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.783996666163762e-07,
+      "loss": 2.2654,
+      "step": 4404
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7742662600670645e-07,
+      "loss": 2.1768,
+      "step": 4405
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7645619833890625e-07,
+      "loss": 2.2529,
+      "step": 4406
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.754883841387711e-07,
+      "loss": 2.2835,
+      "step": 4407
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7452318393067436e-07,
+      "loss": 2.3128,
+      "step": 4408
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.735605982375771e-07,
+      "loss": 2.2178,
+      "step": 4409
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7260062758102214e-07,
+      "loss": 2.2628,
+      "step": 4410
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.716432724811362e-07,
+      "loss": 2.3265,
+      "step": 4411
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.706885334566283e-07,
+      "loss": 2.1489,
+      "step": 4412
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.697364110247901e-07,
+      "loss": 2.2957,
+      "step": 4413
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6878690570149515e-07,
+      "loss": 2.2687,
+      "step": 4414
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6784001800120132e-07,
+      "loss": 2.2939,
+      "step": 4415
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6689574843694433e-07,
+      "loss": 2.2269,
+      "step": 4416
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6595409752034442e-07,
+      "loss": 2.2798,
+      "step": 4417
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6501506576160176e-07,
+      "loss": 2.2286,
+      "step": 4418
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6407865366949882e-07,
+      "loss": 2.3238,
+      "step": 4419
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6314486175139588e-07,
+      "loss": 2.3369,
+      "step": 4420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6221369051323655e-07,
+      "loss": 2.2872,
+      "step": 4421
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.612851404595428e-07,
+      "loss": 2.2314,
+      "step": 4422
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6035921209341666e-07,
+      "loss": 2.2027,
+      "step": 4423
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5943590591653957e-07,
+      "loss": 2.2607,
+      "step": 4424
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5851522242917362e-07,
+      "loss": 2.2035,
+      "step": 4425
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5759716213015753e-07,
+      "loss": 2.2712,
+      "step": 4426
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5668172551691174e-07,
+      "loss": 2.239,
+      "step": 4427
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5576891308543063e-07,
+      "loss": 2.2459,
+      "step": 4428
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5485872533029246e-07,
+      "loss": 2.2695,
+      "step": 4429
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5395116274464882e-07,
+      "loss": 2.2145,
+      "step": 4430
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.530462258202303e-07,
+      "loss": 2.265,
+      "step": 4431
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5214391504734694e-07,
+      "loss": 2.3352,
+      "step": 4432
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5124423091488205e-07,
+      "loss": 2.3515,
+      "step": 4433
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5034717391029962e-07,
+      "loss": 2.274,
+      "step": 4434
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4945274451963754e-07,
+      "loss": 2.3298,
+      "step": 4435
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4856094322751036e-07,
+      "loss": 2.2861,
+      "step": 4436
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.47671770517111e-07,
+      "loss": 2.2693,
+      "step": 4437
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4678522687020414e-07,
+      "loss": 2.3338,
+      "step": 4438
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4590131276713437e-07,
+      "loss": 2.2967,
+      "step": 4439
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4502002868681863e-07,
+      "loss": 2.3415,
+      "step": 4440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4414137510674942e-07,
+      "loss": 2.2266,
+      "step": 4441
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4326535250299478e-07,
+      "loss": 2.3084,
+      "step": 4442
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4239196135019618e-07,
+      "loss": 2.3188,
+      "step": 4443
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4152120212157118e-07,
+      "loss": 2.1961,
+      "step": 4444
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4065307528890848e-07,
+      "loss": 2.2157,
+      "step": 4445
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3978758132257299e-07,
+      "loss": 2.2614,
+      "step": 4446
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3892472069150287e-07,
+      "loss": 2.2366,
+      "step": 4447
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3806449386320696e-07,
+      "loss": 2.2899,
+      "step": 4448
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3720690130377024e-07,
+      "loss": 2.1187,
+      "step": 4449
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.363519434778493e-07,
+      "loss": 2.2406,
+      "step": 4450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3549962084867197e-07,
+      "loss": 2.2459,
+      "step": 4451
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3464993387803994e-07,
+      "loss": 2.2679,
+      "step": 4452
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.338028830263255e-07,
+      "loss": 2.2088,
+      "step": 4453
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3295846875247487e-07,
+      "loss": 2.3369,
+      "step": 4454
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3211669151400319e-07,
+      "loss": 2.2696,
+      "step": 4455
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3127755176699787e-07,
+      "loss": 2.2744,
+      "step": 4456
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3044104996611796e-07,
+      "loss": 2.2662,
+      "step": 4457
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2960718656459147e-07,
+      "loss": 2.3398,
+      "step": 4458
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2877596201421926e-07,
+      "loss": 2.3018,
+      "step": 4459
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2794737676536993e-07,
+      "loss": 2.2359,
+      "step": 4460
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2712143126698385e-07,
+      "loss": 2.2437,
+      "step": 4461
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2629812596657075e-07,
+      "loss": 2.2854,
+      "step": 4462
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.254774613102089e-07,
+      "loss": 2.2685,
+      "step": 4463
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2465943774254752e-07,
+      "loss": 2.2464,
+      "step": 4464
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.238440557068038e-07,
+      "loss": 2.2019,
+      "step": 4465
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2303131564476322e-07,
+      "loss": 2.2534,
+      "step": 4466
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2222121799678076e-07,
+      "loss": 2.3213,
+      "step": 4467
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2141376320177923e-07,
+      "loss": 2.3115,
+      "step": 4468
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2060895169724973e-07,
+      "loss": 2.2802,
+      "step": 4469
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1980678391925126e-07,
+      "loss": 2.3327,
+      "step": 4470
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1900726030241005e-07,
+      "loss": 2.2309,
+      "step": 4471
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1821038127992013e-07,
+      "loss": 2.298,
+      "step": 4472
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1741614728354167e-07,
+      "loss": 2.2531,
+      "step": 4473
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1662455874360323e-07,
+      "loss": 2.2989,
+      "step": 4474
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1583561608899951e-07,
+      "loss": 2.2507,
+      "step": 4475
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1504931974718969e-07,
+      "loss": 2.2182,
+      "step": 4476
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1426567014420297e-07,
+      "loss": 2.3477,
+      "step": 4477
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1348466770462974e-07,
+      "loss": 2.3209,
+      "step": 4478
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1270631285163203e-07,
+      "loss": 2.3345,
+      "step": 4479
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1193060600693085e-07,
+      "loss": 2.2282,
+      "step": 4480
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1115754759081776e-07,
+      "loss": 2.2591,
+      "step": 4481
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1038713802214718e-07,
+      "loss": 2.3242,
+      "step": 4482
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0961937771833687e-07,
+      "loss": 2.1932,
+      "step": 4483
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0885426709537295e-07,
+      "loss": 2.2464,
+      "step": 4484
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0809180656780161e-07,
+      "loss": 2.2155,
+      "step": 4485
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0733199654873683e-07,
+      "loss": 2.3049,
+      "step": 4486
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.065748374498543e-07,
+      "loss": 2.2334,
+      "step": 4487
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0582032968139366e-07,
+      "loss": 2.309,
+      "step": 4488
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0506847365215955e-07,
+      "loss": 2.293,
+      "step": 4489
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0431926976951778e-07,
+      "loss": 2.2061,
+      "step": 4490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0357271843939864e-07,
+      "loss": 2.3163,
+      "step": 4491
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0282882006629413e-07,
+      "loss": 2.2375,
+      "step": 4492
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0208757505326016e-07,
+      "loss": 2.2512,
+      "step": 4493
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0134898380191438e-07,
+      "loss": 2.2768,
+      "step": 4494
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0061304671243554e-07,
+      "loss": 2.2717,
+      "step": 4495
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.987976418356581e-08,
+      "loss": 2.2835,
+      "step": 4496
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.914913661260905e-08,
+      "loss": 2.2893,
+      "step": 4497
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.84211643954297e-08,
+      "loss": 2.2969,
+      "step": 4498
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.769584792645337e-08,
+      "loss": 2.238,
+      "step": 4499
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.697318759866791e-08,
+      "loss": 2.2092,
+      "step": 4500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.62531838036218e-08,
+      "loss": 2.203,
+      "step": 4501
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.553583693142299e-08,
+      "loss": 2.3248,
+      "step": 4502
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.482114737074055e-08,
+      "loss": 2.2491,
+      "step": 4503
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.410911550880474e-08,
+      "loss": 2.2495,
+      "step": 4504
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.339974173140476e-08,
+      "loss": 2.194,
+      "step": 4505
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.269302642288924e-08,
+      "loss": 2.2193,
+      "step": 4506
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.198896996616802e-08,
+      "loss": 2.2258,
+      "step": 4507
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.128757274270816e-08,
+      "loss": 2.2478,
+      "step": 4508
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.058883513253846e-08,
+      "loss": 2.1854,
+      "step": 4509
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.989275751424442e-08,
+      "loss": 2.2321,
+      "step": 4510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.919934026497101e-08,
+      "loss": 2.2838,
+      "step": 4511
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.85085837604227e-08,
+      "loss": 2.2289,
+      "step": 4512
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.782048837486068e-08,
+      "loss": 2.2435,
+      "step": 4513
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.713505448110671e-08,
+      "loss": 2.2524,
+      "step": 4514
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.645228245053761e-08,
+      "loss": 2.2676,
+      "step": 4515
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.577217265309079e-08,
+      "loss": 2.3278,
+      "step": 4516
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.509472545725872e-08,
+      "loss": 2.2209,
+      "step": 4517
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.441994123009223e-08,
+      "loss": 2.3708,
+      "step": 4518
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.374782033720052e-08,
+      "loss": 2.2312,
+      "step": 4519
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.307836314274786e-08,
+      "loss": 2.3103,
+      "step": 4520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.241157000945631e-08,
+      "loss": 2.2237,
+      "step": 4521
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.174744129860468e-08,
+      "loss": 2.2513,
+      "step": 4522
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.108597737002676e-08,
+      "loss": 2.3186,
+      "step": 4523
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.042717858211424e-08,
+      "loss": 2.2749,
+      "step": 4524
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.977104529181378e-08,
+      "loss": 2.3433,
+      "step": 4525
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.911757785462882e-08,
+      "loss": 2.1882,
+      "step": 4526
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.846677662461721e-08,
+      "loss": 2.2207,
+      "step": 4527
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.781864195439248e-08,
+      "loss": 2.3405,
+      "step": 4528
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.717317419512426e-08,
+      "loss": 2.2657,
+      "step": 4529
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.65303736965356e-08,
+      "loss": 2.2603,
+      "step": 4530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.589024080690621e-08,
+      "loss": 2.225,
+      "step": 4531
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.525277587306978e-08,
+      "loss": 2.2315,
+      "step": 4532
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.461797924041336e-08,
+      "loss": 2.267,
+      "step": 4533
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.398585125288016e-08,
+      "loss": 2.2298,
+      "step": 4534
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.335639225296565e-08,
+      "loss": 2.2459,
+      "step": 4535
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.272960258172035e-08,
+      "loss": 2.3784,
+      "step": 4536
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.210548257874927e-08,
+      "loss": 2.3238,
+      "step": 4537
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.148403258220749e-08,
+      "loss": 2.2921,
+      "step": 4538
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.086525292880786e-08,
+      "loss": 2.2268,
+      "step": 4539
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.024914395381388e-08,
+      "loss": 2.3339,
+      "step": 4540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.963570599104297e-08,
+      "loss": 2.3941,
+      "step": 4541
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.902493937286369e-08,
+      "loss": 2.2959,
+      "step": 4542
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.84168444301997e-08,
+      "loss": 2.1687,
+      "step": 4543
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.781142149252518e-08,
+      "loss": 2.273,
+      "step": 4544
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.72086708878672e-08,
+      "loss": 2.287,
+      "step": 4545
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.660859294280564e-08,
+      "loss": 2.2287,
+      "step": 4546
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.601118798247153e-08,
+      "loss": 2.1679,
+      "step": 4547
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.54164563305465e-08,
+      "loss": 2.2722,
+      "step": 4548
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.482439830926668e-08,
+      "loss": 2.2622,
+      "step": 4549
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.423501423941658e-08,
+      "loss": 2.2133,
+      "step": 4550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.364830444033465e-08,
+      "loss": 2.3283,
+      "step": 4551
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.306426922990827e-08,
+      "loss": 2.3029,
+      "step": 4552
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.248290892457543e-08,
+      "loss": 2.3315,
+      "step": 4553
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.190422383932693e-08,
+      "loss": 2.365,
+      "step": 4554
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.132821428770308e-08,
+      "loss": 2.2079,
+      "step": 4555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.075488058179369e-08,
+      "loss": 2.2242,
+      "step": 4556
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.01842230322397e-08,
+      "loss": 2.3192,
+      "step": 4557
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.961624194823157e-08,
+      "loss": 2.3267,
+      "step": 4558
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.905093763751091e-08,
+      "loss": 2.2696,
+      "step": 4559
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.84883104063666e-08,
+      "loss": 2.3102,
+      "step": 4560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.7928360559639793e-08,
+      "loss": 2.2961,
+      "step": 4561
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.737108840071892e-08,
+      "loss": 2.1843,
+      "step": 4562
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.68164942315419e-08,
+      "loss": 2.2385,
+      "step": 4563
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.626457835259724e-08,
+      "loss": 2.2786,
+      "step": 4564
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.5715341062919645e-08,
+      "loss": 2.2282,
+      "step": 4565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.516878266009551e-08,
+      "loss": 2.2096,
+      "step": 4566
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.462490344025739e-08,
+      "loss": 2.309,
+      "step": 4567
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.4083703698087356e-08,
+      "loss": 2.27,
+      "step": 4568
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.3545183726815834e-08,
+      "loss": 2.2453,
+      "step": 4569
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.3009343818219985e-08,
+      "loss": 2.2829,
+      "step": 4570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.247618426262702e-08,
+      "loss": 2.3569,
+      "step": 4571
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.194570534890975e-08,
+      "loss": 2.2026,
+      "step": 4572
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.141790736448993e-08,
+      "loss": 2.1594,
+      "step": 4573
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.089279059533658e-08,
+      "loss": 2.2783,
+      "step": 4574
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.0370355325964904e-08,
+      "loss": 2.2696,
+      "step": 4575
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.985060183943957e-08,
+      "loss": 2.2251,
+      "step": 4576
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9333530417369767e-08,
+      "loss": 2.2207,
+      "step": 4577
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.8819141339911394e-08,
+      "loss": 2.2847,
+      "step": 4578
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.830743488577039e-08,
+      "loss": 2.2964,
+      "step": 4579
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.779841133219554e-08,
+      "loss": 2.2083,
+      "step": 4580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.729207095498345e-08,
+      "loss": 2.2671,
+      "step": 4581
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.678841402847689e-08,
+      "loss": 2.3094,
+      "step": 4582
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.628744082556536e-08,
+      "loss": 2.2123,
+      "step": 4583
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.578915161768283e-08,
+      "loss": 2.1684,
+      "step": 4584
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.52935466748089e-08,
+      "loss": 2.2914,
+      "step": 4585
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.48006262654721e-08,
+      "loss": 2.3634,
+      "step": 4586
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.431039065674214e-08,
+      "loss": 2.3128,
+      "step": 4587
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.382284011423599e-08,
+      "loss": 2.2407,
+      "step": 4588
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.333797490211733e-08,
+      "loss": 2.3025,
+      "step": 4589
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.285579528309158e-08,
+      "loss": 2.3021,
+      "step": 4590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.237630151841199e-08,
+      "loss": 2.2483,
+      "step": 4591
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.189949386787462e-08,
+      "loss": 2.2927,
+      "step": 4592
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.142537258982116e-08,
+      "loss": 2.187,
+      "step": 4593
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.095393794113778e-08,
+      "loss": 2.2214,
+      "step": 4594
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0485190177254054e-08,
+      "loss": 2.2979,
+      "step": 4595
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.001912955214626e-08,
+      "loss": 2.3201,
+      "step": 4596
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.955575631833131e-08,
+      "loss": 2.2454,
+      "step": 4597
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9095070726871685e-08,
+      "loss": 2.219,
+      "step": 4598
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.863707302737441e-08,
+      "loss": 2.2664,
+      "step": 4599
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8181763467989295e-08,
+      "loss": 2.2791,
+      "step": 4600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.772914229540958e-08,
+      "loss": 2.1319,
+      "step": 4601
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7279209754872424e-08,
+      "loss": 2.2135,
+      "step": 4602
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.683196609015782e-08,
+      "loss": 2.324,
+      "step": 4603
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.638741154358916e-08,
+      "loss": 2.2217,
+      "step": 4604
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.594554635603209e-08,
+      "loss": 2.2394,
+      "step": 4605
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5506370766896225e-08,
+      "loss": 2.285,
+      "step": 4606
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5069885014134e-08,
+      "loss": 2.3778,
+      "step": 4607
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.463608933423845e-08,
+      "loss": 2.201,
+      "step": 4608
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4204983962248253e-08,
+      "loss": 2.1864,
+      "step": 4609
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.377656913174099e-08,
+      "loss": 2.2372,
+      "step": 4610
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3350845074839324e-08,
+      "loss": 2.3364,
+      "step": 4611
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.292781202220652e-08,
+      "loss": 2.2275,
+      "step": 4612
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2507470203048116e-08,
+      "loss": 2.2459,
+      "step": 4613
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2089819845111946e-08,
+      "loss": 2.2368,
+      "step": 4614
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.167486117468643e-08,
+      "loss": 2.2511,
+      "step": 4615
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.126259441660284e-08,
+      "loss": 2.2283,
+      "step": 4616
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.085301979423305e-08,
+      "loss": 2.304,
+      "step": 4617
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0446137529491215e-08,
+      "loss": 2.2614,
+      "step": 4618
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.00419478428321e-08,
+      "loss": 2.2177,
+      "step": 4619
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9640450953251078e-08,
+      "loss": 2.3704,
+      "step": 4620
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9241647078285807e-08,
+      "loss": 2.2899,
+      "step": 4621
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8845536434013998e-08,
+      "loss": 2.2663,
+      "step": 4622
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8452119235053976e-08,
+      "loss": 2.3422,
+      "step": 4623
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8061395694565785e-08,
+      "loss": 2.2822,
+      "step": 4624
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.767336602424786e-08,
+      "loss": 2.2768,
+      "step": 4625
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7288030434342027e-08,
+      "loss": 2.2913,
+      "step": 4626
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.690538913362739e-08,
+      "loss": 2.2092,
+      "step": 4627
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6525442329425334e-08,
+      "loss": 2.2417,
+      "step": 4628
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.614819022759729e-08,
+      "loss": 2.1818,
+      "step": 4629
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.57736330325431e-08,
+      "loss": 2.3223,
+      "step": 4630
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5401770947203753e-08,
+      "loss": 2.3552,
+      "step": 4631
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5032604173060305e-08,
+      "loss": 2.2239,
+      "step": 4632
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4666132910131647e-08,
+      "loss": 2.2278,
+      "step": 4633
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4302357356977836e-08,
+      "loss": 2.1822,
+      "step": 4634
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3941277710697874e-08,
+      "loss": 2.3048,
+      "step": 4635
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.358289416693027e-08,
+      "loss": 2.1353,
+      "step": 4636
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.322720691985192e-08,
+      "loss": 2.3212,
+      "step": 4637
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.287421616218033e-08,
+      "loss": 2.2722,
+      "step": 4638
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2523922085170846e-08,
+      "loss": 2.2733,
+      "step": 4639
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.217632487861776e-08,
+      "loss": 2.2838,
+      "step": 4640
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1831424730854866e-08,
+      "loss": 2.2046,
+      "step": 4641
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1489221828753237e-08,
+      "loss": 2.2249,
+      "step": 4642
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.114971635772456e-08,
+      "loss": 2.2532,
+      "step": 4643
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0812908501717245e-08,
+      "loss": 2.3105,
+      "step": 4644
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.047879844321865e-08,
+      "loss": 2.3069,
+      "step": 4645
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.014738636325564e-08,
+      "loss": 2.3152,
+      "step": 4646
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.981867244139124e-08,
+      "loss": 2.293,
+      "step": 4647
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9492656855727986e-08,
+      "loss": 2.3154,
+      "step": 4648
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9169339782905138e-08,
+      "loss": 2.2961,
+      "step": 4649
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.884872139810201e-08,
+      "loss": 2.3162,
+      "step": 4650
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8530801875033532e-08,
+      "loss": 2.2733,
+      "step": 4651
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.821558138595303e-08,
+      "loss": 2.2459,
+      "step": 4652
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7903060101652213e-08,
+      "loss": 2.2647,
+      "step": 4653
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7593238191459526e-08,
+      "loss": 2.1998,
+      "step": 4654
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7286115823241245e-08,
+      "loss": 2.2534,
+      "step": 4655
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6981693163400925e-08,
+      "loss": 2.2961,
+      "step": 4656
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.667997037687885e-08,
+      "loss": 2.2323,
+      "step": 4657
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6380947627153143e-08,
+      "loss": 2.2587,
+      "step": 4658
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6084625076238647e-08,
+      "loss": 2.2776,
+      "step": 4659
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.579100288468749e-08,
+      "loss": 2.1516,
+      "step": 4660
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5500081211588525e-08,
+      "loss": 2.1814,
+      "step": 4661
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.521186021456733e-08,
+      "loss": 2.2845,
+      "step": 4662
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4926340049786214e-08,
+      "loss": 2.2903,
+      "step": 4663
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4643520871944761e-08,
+      "loss": 2.286,
+      "step": 4664
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4363402834277617e-08,
+      "loss": 2.2669,
+      "step": 4665
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4085986088557824e-08,
+      "loss": 2.227,
+      "step": 4666
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3811270785094032e-08,
+      "loss": 2.1613,
+      "step": 4667
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3539257072729962e-08,
+      "loss": 2.3392,
+      "step": 4668
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3269945098847714e-08,
+      "loss": 2.2814,
+      "step": 4669
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3003335009363905e-08,
+      "loss": 2.241,
+      "step": 4670
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2739426948732426e-08,
+      "loss": 2.2063,
+      "step": 4671
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2478221059941675e-08,
+      "loss": 2.2505,
+      "step": 4672
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.221971748451789e-08,
+      "loss": 2.2194,
+      "step": 4673
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.196391636252181e-08,
+      "loss": 2.3513,
+      "step": 4674
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1710817832549792e-08,
+      "loss": 2.2928,
+      "step": 4675
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1460422031735474e-08,
+      "loss": 2.2054,
+      "step": 4676
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1212729095745334e-08,
+      "loss": 2.2937,
+      "step": 4677
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0967739158784795e-08,
+      "loss": 2.2515,
+      "step": 4678
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0725452353592126e-08,
+      "loss": 2.2067,
+      "step": 4679
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0485868811441757e-08,
+      "loss": 2.1554,
+      "step": 4680
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0248988662144299e-08,
+      "loss": 2.2269,
+      "step": 4681
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0014812034044308e-08,
+      "loss": 2.2495,
+      "step": 4682
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.783339054022512e-09,
+      "loss": 2.3818,
+      "step": 4683
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.5545698474947e-09,
+      "loss": 2.2832,
+      "step": 4684
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.32850453841061e-09,
+      "loss": 2.2521,
+      "step": 4685
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.105143249256154e-09,
+      "loss": 2.1949,
+      "step": 4686
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.884486101052303e-09,
+      "loss": 2.2379,
+      "step": 4687
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.666533213353422e-09,
+      "loss": 2.27,
+      "step": 4688
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.451284704250607e-09,
+      "loss": 2.2618,
+      "step": 4689
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.238740690367786e-09,
+      "loss": 2.2434,
+      "step": 4690
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.02890128686562e-09,
+      "loss": 2.3784,
+      "step": 4691
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.821766607437054e-09,
+      "loss": 2.3387,
+      "step": 4692
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.617336764310645e-09,
+      "loss": 2.2132,
+      "step": 4693
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.415611868251127e-09,
+      "loss": 2.191,
+      "step": 4694
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.216592028553848e-09,
+      "loss": 2.2481,
+      "step": 4695
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.020277353052551e-09,
+      "loss": 2.2679,
+      "step": 4696
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.8266679481127086e-09,
+      "loss": 2.3234,
+      "step": 4697
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.635763918634852e-09,
+      "loss": 2.2163,
+      "step": 4698
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.4475653680534655e-09,
+      "loss": 2.2738,
+      "step": 4699
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.262072398336982e-09,
+      "loss": 2.2987,
+      "step": 4700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.0792851099894526e-09,
+      "loss": 2.2857,
+      "step": 4701
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.899203602046655e-09,
+      "loss": 2.3361,
+      "step": 4702
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.7218279720794305e-09,
+      "loss": 2.2913,
+      "step": 4703
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.547158316193124e-09,
+      "loss": 2.283,
+      "step": 4704
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.375194729025923e-09,
+      "loss": 2.2411,
+      "step": 4705
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.205937303750519e-09,
+      "loss": 2.245,
+      "step": 4706
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.039386132072444e-09,
+      "loss": 2.3144,
+      "step": 4707
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.875541304232845e-09,
+      "loss": 2.318,
+      "step": 4708
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.714402909004601e-09,
+      "loss": 2.2843,
+      "step": 4709
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.555971033693984e-09,
+      "loss": 2.2762,
+      "step": 4710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4002457641434404e-09,
+      "loss": 2.3206,
+      "step": 4711
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.247227184726033e-09,
+      "loss": 2.3031,
+      "step": 4712
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.096915378349886e-09,
+      "loss": 2.299,
+      "step": 4713
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.949310426456521e-09,
+      "loss": 2.265,
+      "step": 4714
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8044124090197425e-09,
+      "loss": 2.2584,
+      "step": 4715
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.662221404547861e-09,
+      "loss": 2.1805,
+      "step": 4716
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.522737490082029e-09,
+      "loss": 2.2459,
+      "step": 4717
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.385960741196792e-09,
+      "loss": 2.2309,
+      "step": 4718
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.251891231999538e-09,
+      "loss": 2.1804,
+      "step": 4719
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.120529035131603e-09,
+      "loss": 2.2762,
+      "step": 4720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.991874221766611e-09,
+      "loss": 2.2499,
+      "step": 4721
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8659268616115785e-09,
+      "loss": 2.3134,
+      "step": 4722
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7426870229069204e-09,
+      "loss": 2.3141,
+      "step": 4723
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6221547724253337e-09,
+      "loss": 2.2724,
+      "step": 4724
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5043301754740235e-09,
+      "loss": 2.3145,
+      "step": 4725
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3892132958908133e-09,
+      "loss": 2.2162,
+      "step": 4726
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2768041960485875e-09,
+      "loss": 2.2522,
+      "step": 4727
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1671029368525165e-09,
+      "loss": 2.2658,
+      "step": 4728
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0601095777395e-09,
+      "loss": 2.2748,
+      "step": 4729
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9558241766803877e-09,
+      "loss": 2.3583,
+      "step": 4730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8542467901788708e-09,
+      "loss": 2.2912,
+      "step": 4731
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.75537747327037e-09,
+      "loss": 2.2709,
+      "step": 4732
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6592162795248112e-09,
+      "loss": 2.1873,
+      "step": 4733
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565763261043296e-09,
+      "loss": 2.2015,
+      "step": 4734
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4750184684597657e-09,
+      "loss": 2.279,
+      "step": 4735
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3869819509415573e-09,
+      "loss": 2.3364,
+      "step": 4736
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3016537561877373e-09,
+      "loss": 2.2193,
+      "step": 4737
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2190339304307686e-09,
+      "loss": 2.3622,
+      "step": 4738
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1391225184348431e-09,
+      "loss": 2.2465,
+      "step": 4739
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0619195634975489e-09,
+      "loss": 2.3058,
+      "step": 4740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.874251074482033e-10,
+      "loss": 2.2435,
+      "step": 4741
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.156391906495199e-10,
+      "loss": 2.2465,
+      "step": 4742
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.465618519959417e-10,
+      "loss": 2.1963,
+      "step": 4743
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.801931289147524e-10,
+      "loss": 2.2842,
+      "step": 4744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.165330573649654e-10,
+      "loss": 2.2641,
+      "step": 4745
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.555816718389896e-10,
+      "loss": 2.308,
+      "step": 4746
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.973390053615191e-10,
+      "loss": 2.2773,
+      "step": 4747
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.418050894884231e-10,
+      "loss": 2.2744,
+      "step": 4748
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.889799543095208e-10,
+      "loss": 2.242,
+      "step": 4749
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.38863628445807e-10,
+      "loss": 2.402,
+      "step": 4750
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.914561390516713e-10,
+      "loss": 2.2651,
+      "step": 4751
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.467575118126787e-10,
+      "loss": 2.2978,
+      "step": 4752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.047677709472341e-10,
+      "loss": 2.3125,
+      "step": 4753
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6548693920602774e-10,
+      "loss": 2.2332,
+      "step": 4754
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2891503787203507e-10,
+      "loss": 2.2559,
+      "step": 4755
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9505208676051657e-10,
+      "loss": 2.2677,
+      "step": 4756
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6389810421846286e-10,
+      "loss": 2.2585,
+      "step": 4757
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3545310712681503e-10,
+      "loss": 2.4068,
+      "step": 4758
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0971711089602377e-10,
+      "loss": 2.1959,
+      "step": 4759
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.669012947104538e-11,
+      "loss": 2.2326,
+      "step": 4760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.63721753280111e-11,
+      "loss": 2.2954,
+      "step": 4761
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.876325947589244e-11,
+      "loss": 2.2713,
+      "step": 4762
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.386339145483586e-11,
+      "loss": 2.2129,
+      "step": 4763
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1672579337828072e-11,
+      "loss": 2.2854,
+      "step": 4764
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2190829730696074e-11,
+      "loss": 2.2564,
+      "step": 4765
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.4181477698866944e-12,
+      "loss": 2.3053,
+      "step": 4766
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3545371257972506e-12,
+      "loss": 2.1705,
+      "step": 4767
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 2.3676,
+      "step": 4768
+    },
+    {
+      "epoch": 1.0,
+      "step": 4768,
+      "total_flos": 1.5791930895771894e+18,
+      "train_loss": 2.2993583149157915,
+      "train_runtime": 12628.2761,
+      "train_samples_per_second": 48.334,
+      "train_steps_per_second": 0.378
+    }
+  ],
+  "max_steps": 4768,
+  "num_train_epochs": 1,
+  "total_flos": 1.5791930895771894e+18,
+  "trial_name": null,
+  "trial_params": null
+}