diff --git "a/check-2000/trainer_state.json" "b/check-2000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/check-2000/trainer_state.json"
@@ -0,0 +1,12032 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.8,
+  "global_step": 2000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 3.5999,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 3.4907,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 3.4075,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 3.5758,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 3.4409,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 3.6086,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 3.5406,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 3.5054,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 3.4912,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 3.5096,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1e-06,
+      "loss": 3.5311,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 3.5488,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3e-06,
+      "loss": 3.6292,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 3.605,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5e-06,
+      "loss": 3.4857,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 3.604,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 3.4052,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8000000000000001e-06,
+      "loss": 3.5082,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 3.5957,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 3.5203,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 3.5292,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2e-06,
+      "loss": 3.6016,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3000000000000004e-06,
+      "loss": 3.4768,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 3.5096,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-06,
+      "loss": 3.5731,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6e-06,
+      "loss": 3.6436,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7000000000000004e-06,
+      "loss": 3.611,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 3.5204,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9e-06,
+      "loss": 3.5627,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-06,
+      "loss": 3.4304,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1000000000000004e-06,
+      "loss": 3.561,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 3.4875,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3000000000000006e-06,
+      "loss": 3.5235,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 3.5806,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5e-06,
+      "loss": 3.4574,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6000000000000003e-06,
+      "loss": 3.6234,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7e-06,
+      "loss": 3.5705,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8000000000000005e-06,
+      "loss": 3.5451,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 3.5137,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 3.556,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1e-06,
+      "loss": 3.5021,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 3.4982,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3e-06,
+      "loss": 3.518,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4e-06,
+      "loss": 3.6175,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5e-06,
+      "loss": 3.5671,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.600000000000001e-06,
+      "loss": 3.5508,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.7e-06,
+      "loss": 3.5723,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 3.514,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 3.5213,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 3.6098,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1e-06,
+      "loss": 3.5081,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2e-06,
+      "loss": 3.5695,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.300000000000001e-06,
+      "loss": 3.5433,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 3.4399,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 3.5111,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 3.4996,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.7e-06,
+      "loss": 3.541,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8e-06,
+      "loss": 3.5154,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9e-06,
+      "loss": 3.5028,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6e-06,
+      "loss": 3.5663,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.1e-06,
+      "loss": 3.5558,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.200000000000001e-06,
+      "loss": 3.5197,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 3.5376,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 3.4836,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 3.4729,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.600000000000001e-06,
+      "loss": 3.4901,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 3.4395,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 3.5207,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.9e-06,
+      "loss": 3.5453,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7e-06,
+      "loss": 3.4943,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.100000000000001e-06,
+      "loss": 3.5105,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.2000000000000005e-06,
+      "loss": 3.4329,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.3e-06,
+      "loss": 3.5058,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.4e-06,
+      "loss": 3.427,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 3.5253,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.600000000000001e-06,
+      "loss": 3.5054,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.7e-06,
+      "loss": 3.5407,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.800000000000002e-06,
+      "loss": 3.5341,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.9e-06,
+      "loss": 3.4684,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 3.5198,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.1e-06,
+      "loss": 3.5226,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.2e-06,
+      "loss": 3.4375,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.3e-06,
+      "loss": 3.5305,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 3.439,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.5e-06,
+      "loss": 3.4471,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.6e-06,
+      "loss": 3.4712,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.700000000000001e-06,
+      "loss": 3.469,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.8e-06,
+      "loss": 3.4801,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 3.4847,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9e-06,
+      "loss": 3.4125,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 3.404,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.200000000000002e-06,
+      "loss": 3.5079,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.3e-06,
+      "loss": 3.4877,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.4e-06,
+      "loss": 3.4161,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.5e-06,
+      "loss": 3.4299,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 3.5151,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.7e-06,
+      "loss": 3.4091,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 3.4965,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9e-06,
+      "loss": 3.3879,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1e-05,
+      "loss": 3.4715,
+      "step": 100
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999995716318147e-06,
+      "loss": 3.3704,
+      "step": 101
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999982865279924e-06,
+      "loss": 3.3945,
+      "step": 102
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999961446907354e-06,
+      "loss": 3.3851,
+      "step": 103
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999931461237135e-06,
+      "loss": 3.5679,
+      "step": 104
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999892908320647e-06,
+      "loss": 3.3732,
+      "step": 105
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99984578822395e-06,
+      "loss": 3.4268,
+      "step": 106
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999790101027783e-06,
+      "loss": 3.4499,
+      "step": 107
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999725846827562e-06,
+      "loss": 3.5023,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999653025733386e-06,
+      "loss": 3.4784,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999571637870035e-06,
+      "loss": 3.4703,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999481683376964e-06,
+      "loss": 3.4487,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999383162408303e-06,
+      "loss": 3.5005,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999276075132871e-06,
+      "loss": 3.3971,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999160421734155e-06,
+      "loss": 3.4392,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.999036202410324e-06,
+      "loss": 3.4317,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.998903417374228e-06,
+      "loss": 3.3962,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.998762066853388e-06,
+      "loss": 3.3806,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.998612151090004e-06,
+      "loss": 3.5255,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.998453670340954e-06,
+      "loss": 3.3639,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.998286624877786e-06,
+      "loss": 3.4282,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.998111014986735e-06,
+      "loss": 3.4476,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.997926840968699e-06,
+      "loss": 3.5013,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.997734103139255e-06,
+      "loss": 3.4478,
+      "step": 123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.997532801828659e-06,
+      "loss": 3.4249,
+      "step": 124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.997322937381829e-06,
+      "loss": 3.4951,
+      "step": 125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.997104510158365e-06,
+      "loss": 3.458,
+      "step": 126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.996877520532535e-06,
+      "loss": 3.4623,
+      "step": 127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.996641968893281e-06,
+      "loss": 3.4904,
+      "step": 128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.996397855644216e-06,
+      "loss": 3.4491,
+      "step": 129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.996145181203616e-06,
+      "loss": 3.4477,
+      "step": 130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.995883946004435e-06,
+      "loss": 3.3556,
+      "step": 131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.995614150494293e-06,
+      "loss": 3.5106,
+      "step": 132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.995335795135475e-06,
+      "loss": 3.3416,
+      "step": 133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.995048880404939e-06,
+      "loss": 3.3707,
+      "step": 134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994753406794303e-06,
+      "loss": 3.4075,
+      "step": 135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994449374809851e-06,
+      "loss": 3.3821,
+      "step": 136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994136784972537e-06,
+      "loss": 3.3885,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.993815637817974e-06,
+      "loss": 3.3104,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99348593389644e-06,
+      "loss": 3.4801,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.993147673772869e-06,
+      "loss": 3.395,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.992800858026868e-06,
+      "loss": 3.3355,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.992445487252692e-06,
+      "loss": 3.3804,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.992081562059258e-06,
+      "loss": 3.455,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.991709083070143e-06,
+      "loss": 3.4226,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.99132805092358e-06,
+      "loss": 3.3595,
+      "step": 145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.990938466272459e-06,
+      "loss": 3.4725,
+      "step": 146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.990540329784319e-06,
+      "loss": 3.3786,
+      "step": 147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.990133642141359e-06,
+      "loss": 3.2975,
+      "step": 148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.989718404040424e-06,
+      "loss": 3.3529,
+      "step": 149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.989294616193018e-06,
+      "loss": 3.3747,
+      "step": 150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.988862279325287e-06,
+      "loss": 3.3929,
+      "step": 151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.988421394178027e-06,
+      "loss": 3.4012,
+      "step": 152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.987971961506686e-06,
+      "loss": 3.4424,
+      "step": 153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.987513982081352e-06,
+      "loss": 3.3433,
+      "step": 154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98704745668676e-06,
+      "loss": 3.3748,
+      "step": 155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98657238612229e-06,
+      "loss": 3.4014,
+      "step": 156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.986088771201965e-06,
+      "loss": 3.3413,
+      "step": 157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.985596612754441e-06,
+      "loss": 3.5123,
+      "step": 158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98509591162302e-06,
+      "loss": 3.3524,
+      "step": 159
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.984586668665641e-06,
+      "loss": 3.3706,
+      "step": 160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.984068884754876e-06,
+      "loss": 3.4132,
+      "step": 161
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983542560777934e-06,
+      "loss": 3.3769,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.983007697636659e-06,
+      "loss": 3.3447,
+      "step": 163
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.982464296247523e-06,
+      "loss": 3.4153,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.981912357541628e-06,
+      "loss": 3.3385,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.981351882464706e-06,
+      "loss": 3.4655,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.980782871977119e-06,
+      "loss": 3.4103,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.98020532705385e-06,
+      "loss": 3.2835,
+      "step": 168
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.979619248684503e-06,
+      "loss": 3.414,
+      "step": 169
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.979024637873309e-06,
+      "loss": 3.3078,
+      "step": 170
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.978421495639118e-06,
+      "loss": 3.3911,
+      "step": 171
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9778098230154e-06,
+      "loss": 3.3248,
+      "step": 172
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.977189621050237e-06,
+      "loss": 3.3739,
+      "step": 173
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.976560890806328e-06,
+      "loss": 3.4135,
+      "step": 174
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.975923633360985e-06,
+      "loss": 3.3932,
+      "step": 175
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.975277849806133e-06,
+      "loss": 3.341,
+      "step": 176
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.974623541248301e-06,
+      "loss": 3.3394,
+      "step": 177
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.973960708808633e-06,
+      "loss": 3.3398,
+      "step": 178
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.973289353622872e-06,
+      "loss": 3.327,
+      "step": 179
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.972609476841368e-06,
+      "loss": 3.3923,
+      "step": 180
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.97192107962907e-06,
+      "loss": 3.388,
+      "step": 181
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.971224163165526e-06,
+      "loss": 3.3861,
+      "step": 182
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.97051872864489e-06,
+      "loss": 3.34,
+      "step": 183
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9698047772759e-06,
+      "loss": 3.3678,
+      "step": 184
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96908231028189e-06,
+      "loss": 3.4153,
+      "step": 185
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.968351328900793e-06,
+      "loss": 3.3659,
+      "step": 186
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967611834385122e-06,
+      "loss": 3.3632,
+      "step": 187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.966863828001982e-06,
+      "loss": 3.4039,
+      "step": 188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.966107311033062e-06,
+      "loss": 3.3419,
+      "step": 189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.965342284774633e-06,
+      "loss": 3.3605,
+      "step": 190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.964568750537545e-06,
+      "loss": 3.4301,
+      "step": 191
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.963786709647228e-06,
+      "loss": 3.3319,
+      "step": 192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.96299616344369e-06,
+      "loss": 3.3632,
+      "step": 193
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.962197113281508e-06,
+      "loss": 3.4386,
+      "step": 194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.961389560529835e-06,
+      "loss": 3.3504,
+      "step": 195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.960573506572391e-06,
+      "loss": 3.3849,
+      "step": 196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.959748952807457e-06,
+      "loss": 3.287,
+      "step": 197
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.95891590064789e-06,
+      "loss": 3.2633,
+      "step": 198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.958074351521097e-06,
+      "loss": 3.3263,
+      "step": 199
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.957224306869053e-06,
+      "loss": 3.475,
+      "step": 200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.956365768148284e-06,
+      "loss": 3.421,
+      "step": 201
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.955498736829876e-06,
+      "loss": 3.3657,
+      "step": 202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.954623214399457e-06,
+      "loss": 3.3685,
+      "step": 203
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.953739202357219e-06,
+      "loss": 3.3724,
+      "step": 204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.952846702217886e-06,
+      "loss": 3.3256,
+      "step": 205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.951945715510738e-06,
+      "loss": 3.3414,
+      "step": 206
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.951036243779586e-06,
+      "loss": 3.3859,
+      "step": 207
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.95011828858279e-06,
+      "loss": 3.4048,
+      "step": 208
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.949191851493236e-06,
+      "loss": 3.3455,
+      "step": 209
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.948256934098353e-06,
+      "loss": 3.3317,
+      "step": 210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.947313538000093e-06,
+      "loss": 3.4261,
+      "step": 211
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946361664814942e-06,
+      "loss": 3.4005,
+      "step": 212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.945401316173908e-06,
+      "loss": 3.1677,
+      "step": 213
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.944432493722525e-06,
+      "loss": 3.2908,
+      "step": 214
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.943455199120836e-06,
+      "loss": 3.4475,
+      "step": 215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.942469434043418e-06,
+      "loss": 3.3889,
+      "step": 216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.941475200179347e-06,
+      "loss": 3.353,
+      "step": 217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.940472499232218e-06,
+      "loss": 3.3653,
+      "step": 218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.93946133292013e-06,
+      "loss": 3.3661,
+      "step": 219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.938441702975689e-06,
+      "loss": 3.2752,
+      "step": 220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.937413611146006e-06,
+      "loss": 3.3116,
+      "step": 221
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.936377059192684e-06,
+      "loss": 3.2361,
+      "step": 222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.935332048891828e-06,
+      "loss": 3.4052,
+      "step": 223
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.934278582034037e-06,
+      "loss": 3.415,
+      "step": 224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.933216660424396e-06,
+      "loss": 3.3893,
+      "step": 225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.932146285882478e-06,
+      "loss": 3.3731,
+      "step": 226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.931067460242342e-06,
+      "loss": 3.2905,
+      "step": 227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.929980185352525e-06,
+      "loss": 3.3357,
+      "step": 228
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.928884463076045e-06,
+      "loss": 3.3722,
+      "step": 229
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.92778029529039e-06,
+      "loss": 3.3559,
+      "step": 230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.926667683887521e-06,
+      "loss": 3.3393,
+      "step": 231
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.92554663077387e-06,
+      "loss": 3.3238,
+      "step": 232
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.92441713787033e-06,
+      "loss": 3.3972,
+      "step": 233
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.923279207112256e-06,
+      "loss": 3.3088,
+      "step": 234
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.922132840449459e-06,
+      "loss": 3.3125,
+      "step": 235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.92097803984621e-06,
+      "loss": 3.2525,
+      "step": 236
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919814807281228e-06,
+      "loss": 3.3198,
+      "step": 237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.918643144747681e-06,
+      "loss": 3.284,
+      "step": 238
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.917463054253179e-06,
+      "loss": 3.3613,
+      "step": 239
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.916274537819774e-06,
+      "loss": 3.375,
+      "step": 240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.915077597483959e-06,
+      "loss": 3.3331,
+      "step": 241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.913872235296657e-06,
+      "loss": 3.3067,
+      "step": 242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.912658453323225e-06,
+      "loss": 3.3796,
+      "step": 243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.911436253643445e-06,
+      "loss": 3.3741,
+      "step": 244
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.91020563835152e-06,
+      "loss": 3.3776,
+      "step": 245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.90896660955608e-06,
+      "loss": 3.3549,
+      "step": 246
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.907719169380164e-06,
+      "loss": 3.445,
+      "step": 247
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.906463319961225e-06,
+      "loss": 3.292,
+      "step": 248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.905199063451135e-06,
+      "loss": 3.2892,
+      "step": 249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.903926402016153e-06,
+      "loss": 3.3124,
+      "step": 250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.902645337836955e-06,
+      "loss": 3.3671,
+      "step": 251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.901355873108611e-06,
+      "loss": 3.3859,
+      "step": 252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.900058010040578e-06,
+      "loss": 3.3,
+      "step": 253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.898751750856715e-06,
+      "loss": 3.2642,
+      "step": 254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.897437097795257e-06,
+      "loss": 3.3088,
+      "step": 255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.89611405310883e-06,
+      "loss": 3.4281,
+      "step": 256
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.89478261906443e-06,
+      "loss": 3.2519,
+      "step": 257
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.89344279794344e-06,
+      "loss": 3.2442,
+      "step": 258
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.892094592041602e-06,
+      "loss": 3.3424,
+      "step": 259
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.890738003669029e-06,
+      "loss": 3.3254,
+      "step": 260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889373035150202e-06,
+      "loss": 3.2116,
+      "step": 261
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887999688823955e-06,
+      "loss": 3.325,
+      "step": 262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.88661796704348e-06,
+      "loss": 3.309,
+      "step": 263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.88522787217632e-06,
+      "loss": 3.348,
+      "step": 264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.883829406604363e-06,
+      "loss": 3.4064,
+      "step": 265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.882422572723844e-06,
+      "loss": 3.2872,
+      "step": 266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.881007372945334e-06,
+      "loss": 3.2664,
+      "step": 267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.879583809693737e-06,
+      "loss": 3.2829,
+      "step": 268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.878151885408294e-06,
+      "loss": 3.3993,
+      "step": 269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.876711602542564e-06,
+      "loss": 3.332,
+      "step": 270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.875262963564436e-06,
+      "loss": 3.3599,
+      "step": 271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.87380597095611e-06,
+      "loss": 3.3308,
+      "step": 272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.872340627214106e-06,
+      "loss": 3.3274,
+      "step": 273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.870866934849248e-06,
+      "loss": 3.2864,
+      "step": 274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.869384896386669e-06,
+      "loss": 3.3199,
+      "step": 275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.867894514365802e-06,
+      "loss": 3.2267,
+      "step": 276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.866395791340376e-06,
+      "loss": 3.3076,
+      "step": 277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.86488872987841e-06,
+      "loss": 3.2564,
+      "step": 278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.863373332562215e-06,
+      "loss": 3.346,
+      "step": 279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.861849601988384e-06,
+      "loss": 3.3936,
+      "step": 280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.860317540767784e-06,
+      "loss": 3.3097,
+      "step": 281
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.858777151525563e-06,
+      "loss": 3.3727,
+      "step": 282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.857228436901137e-06,
+      "loss": 3.2874,
+      "step": 283
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.85567139954818e-06,
+      "loss": 3.3421,
+      "step": 284
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.854106042134642e-06,
+      "loss": 3.2561,
+      "step": 285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852532367342712e-06,
+      "loss": 3.3218,
+      "step": 286
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850950377868846e-06,
+      "loss": 3.2352,
+      "step": 287
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.849360076423736e-06,
+      "loss": 3.3338,
+      "step": 288
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.847761465732319e-06,
+      "loss": 3.344,
+      "step": 289
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.846154548533773e-06,
+      "loss": 3.3131,
+      "step": 290
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.844539327581505e-06,
+      "loss": 3.4186,
+      "step": 291
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.842915805643156e-06,
+      "loss": 3.2577,
+      "step": 292
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.841283985500583e-06,
+      "loss": 3.2355,
+      "step": 293
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.839643869949867e-06,
+      "loss": 3.303,
+      "step": 294
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.8379954618013e-06,
+      "loss": 3.2536,
+      "step": 295
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.836338763879386e-06,
+      "loss": 3.1891,
+      "step": 296
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.834673779022829e-06,
+      "loss": 3.3032,
+      "step": 297
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.833000510084537e-06,
+      "loss": 3.3232,
+      "step": 298
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.831318959931612e-06,
+      "loss": 3.237,
+      "step": 299
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.829629131445342e-06,
+      "loss": 3.2476,
+      "step": 300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.827931027521204e-06,
+      "loss": 3.2284,
+      "step": 301
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.826224651068853e-06,
+      "loss": 3.344,
+      "step": 302
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.824510005012114e-06,
+      "loss": 3.2547,
+      "step": 303
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.822787092288991e-06,
+      "loss": 3.277,
+      "step": 304
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.821055915851647e-06,
+      "loss": 3.3358,
+      "step": 305
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.819316478666406e-06,
+      "loss": 3.3453,
+      "step": 306
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.817568783713744e-06,
+      "loss": 3.2686,
+      "step": 307
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.815812833988292e-06,
+      "loss": 3.2714,
+      "step": 308
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.814048632498818e-06,
+      "loss": 3.3789,
+      "step": 309
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.812276182268236e-06,
+      "loss": 3.2791,
+      "step": 310
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810495486333592e-06,
+      "loss": 3.3049,
+      "step": 311
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808706547746057e-06,
+      "loss": 3.2565,
+      "step": 312
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.806909369570931e-06,
+      "loss": 3.2389,
+      "step": 313
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.805103954887626e-06,
+      "loss": 3.316,
+      "step": 314
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.803290306789676e-06,
+      "loss": 3.2901,
+      "step": 315
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.801468428384716e-06,
+      "loss": 3.2197,
+      "step": 316
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.799638322794482e-06,
+      "loss": 3.3154,
+      "step": 317
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.797799993154815e-06,
+      "loss": 3.2665,
+      "step": 318
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.795953442615637e-06,
+      "loss": 3.281,
+      "step": 319
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.794098674340966e-06,
+      "loss": 3.3703,
+      "step": 320
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.792235691508896e-06,
+      "loss": 3.3022,
+      "step": 321
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.790364497311597e-06,
+      "loss": 3.3318,
+      "step": 322
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.788485094955309e-06,
+      "loss": 3.3004,
+      "step": 323
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.786597487660336e-06,
+      "loss": 3.268,
+      "step": 324
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.784701678661045e-06,
+      "loss": 3.1993,
+      "step": 325
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.78279767120585e-06,
+      "loss": 3.1942,
+      "step": 326
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.780885468557216e-06,
+      "loss": 3.2478,
+      "step": 327
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.778965073991652e-06,
+      "loss": 3.2879,
+      "step": 328
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.777036490799699e-06,
+      "loss": 3.2538,
+      "step": 329
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.775099722285934e-06,
+      "loss": 3.297,
+      "step": 330
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.773154771768956e-06,
+      "loss": 3.244,
+      "step": 331
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.771201642581384e-06,
+      "loss": 3.296,
+      "step": 332
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.769240338069854e-06,
+      "loss": 3.2411,
+      "step": 333
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.767270861595006e-06,
+      "loss": 3.3241,
+      "step": 334
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.765293216531486e-06,
+      "loss": 3.3114,
+      "step": 335
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763307406267933e-06,
+      "loss": 3.2072,
+      "step": 336
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761313434206978e-06,
+      "loss": 3.3252,
+      "step": 337
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.759311303765239e-06,
+      "loss": 3.3711,
+      "step": 338
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.757301018373314e-06,
+      "loss": 3.2265,
+      "step": 339
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.755282581475769e-06,
+      "loss": 3.2717,
+      "step": 340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.75325599653114e-06,
+      "loss": 3.3098,
+      "step": 341
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.75122126701193e-06,
+      "loss": 3.2191,
+      "step": 342
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.749178396404588e-06,
+      "loss": 3.3439,
+      "step": 343
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.74712738820952e-06,
+      "loss": 3.3451,
+      "step": 344
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.745068245941071e-06,
+      "loss": 3.3172,
+      "step": 345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.743000973127523e-06,
+      "loss": 3.1871,
+      "step": 346
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.740925573311098e-06,
+      "loss": 3.3755,
+      "step": 347
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.73884205004793e-06,
+      "loss": 3.2414,
+      "step": 348
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.736750406908082e-06,
+      "loss": 3.2205,
+      "step": 349
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.73465064747553e-06,
+      "loss": 3.2859,
+      "step": 350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.732542775348151e-06,
+      "loss": 3.2808,
+      "step": 351
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.730426794137727e-06,
+      "loss": 3.3264,
+      "step": 352
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.728302707469936e-06,
+      "loss": 3.234,
+      "step": 353
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.726170518984342e-06,
+      "loss": 3.3124,
+      "step": 354
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.72403023233439e-06,
+      "loss": 3.3044,
+      "step": 355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.721881851187406e-06,
+      "loss": 3.3185,
+      "step": 356
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.71972537922458e-06,
+      "loss": 3.365,
+      "step": 357
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.717560820140968e-06,
+      "loss": 3.2736,
+      "step": 358
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.715388177645485e-06,
+      "loss": 3.2747,
+      "step": 359
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.713207455460893e-06,
+      "loss": 3.3034,
+      "step": 360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.7110186573238e-06,
+      "loss": 3.3528,
+      "step": 361
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708821786984652e-06,
+      "loss": 3.3134,
+      "step": 362
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.706616848207728e-06,
+      "loss": 3.2542,
+      "step": 363
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.704403844771128e-06,
+      "loss": 3.2582,
+      "step": 364
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.702182780466775e-06,
+      "loss": 3.2905,
+      "step": 365
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.6999536591004e-06,
+      "loss": 3.3168,
+      "step": 366
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.697716484491545e-06,
+      "loss": 3.2772,
+      "step": 367
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.695471260473546e-06,
+      "loss": 3.2707,
+      "step": 368
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.693217990893533e-06,
+      "loss": 3.3322,
+      "step": 369
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.690956679612422e-06,
+      "loss": 3.2573,
+      "step": 370
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.688687330504908e-06,
+      "loss": 3.3309,
+      "step": 371
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.68640994745946e-06,
+      "loss": 2.7976,
+      "step": 372
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.684124534378307e-06,
+      "loss": 3.2632,
+      "step": 373
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.68183109517745e-06,
+      "loss": 3.2394,
+      "step": 374
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.67952963378663e-06,
+      "loss": 3.2661,
+      "step": 375
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.677220154149338e-06,
+      "loss": 3.2171,
+      "step": 376
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.674902660222805e-06,
+      "loss": 3.3085,
+      "step": 377
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.672577155977993e-06,
+      "loss": 3.3195,
+      "step": 378
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.670243645399594e-06,
+      "loss": 3.2169,
+      "step": 379
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.667902132486009e-06,
+      "loss": 3.2748,
+      "step": 380
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.66555262124936e-06,
+      "loss": 3.2387,
+      "step": 381
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.663195115715472e-06,
+      "loss": 3.3176,
+      "step": 382
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.660829619923861e-06,
+      "loss": 3.2599,
+      "step": 383
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.658456137927745e-06,
+      "loss": 3.2751,
+      "step": 384
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.656074673794018e-06,
+      "loss": 3.2408,
+      "step": 385
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.653685231603256e-06,
+      "loss": 3.3164,
+      "step": 386
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.651287815449699e-06,
+      "loss": 3.2836,
+      "step": 387
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.648882429441258e-06,
+      "loss": 3.2736,
+      "step": 388
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.646469077699494e-06,
+      "loss": 3.2921,
+      "step": 389
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.644047764359623e-06,
+      "loss": 3.2726,
+      "step": 390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.641618493570495e-06,
+      "loss": 3.2159,
+      "step": 391
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.6391812694946e-06,
+      "loss": 3.3036,
+      "step": 392
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.63673609630806e-06,
+      "loss": 3.2619,
+      "step": 393
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.634282978200605e-06,
+      "loss": 3.3331,
+      "step": 394
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.63182191937559e-06,
+      "loss": 3.337,
+      "step": 395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.629352924049975e-06,
+      "loss": 3.2224,
+      "step": 396
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.626875996454312e-06,
+      "loss": 3.2283,
+      "step": 397
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.624391140832749e-06,
+      "loss": 3.3693,
+      "step": 398
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.62189836144302e-06,
+      "loss": 3.337,
+      "step": 399
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.619397662556434e-06,
+      "loss": 3.3051,
+      "step": 400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.616889048457872e-06,
+      "loss": 3.1995,
+      "step": 401
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.614372523445772e-06,
+      "loss": 3.3088,
+      "step": 402
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.611848091832134e-06,
+      "loss": 3.1801,
+      "step": 403
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.609315757942504e-06,
+      "loss": 3.2903,
+      "step": 404
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.606775526115963e-06,
+      "loss": 3.3434,
+      "step": 405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.604227400705134e-06,
+      "loss": 3.1599,
+      "step": 406
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.601671386076154e-06,
+      "loss": 3.2912,
+      "step": 407
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.59910748660869e-06,
+      "loss": 3.2945,
+      "step": 408
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.596535706695911e-06,
+      "loss": 3.2669,
+      "step": 409
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.593956050744493e-06,
+      "loss": 3.2938,
+      "step": 410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.591368523174604e-06,
+      "loss": 3.2197,
+      "step": 411
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.588773128419907e-06,
+      "loss": 3.1876,
+      "step": 412
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.586169870927535e-06,
+      "loss": 3.3189,
+      "step": 413
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.583558755158102e-06,
+      "loss": 3.2128,
+      "step": 414
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.58093978558568e-06,
+      "loss": 3.2416,
+      "step": 415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.578312966697807e-06,
+      "loss": 3.3712,
+      "step": 416
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.575678302995461e-06,
+      "loss": 3.2701,
+      "step": 417
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.57303579899307e-06,
+      "loss": 3.305,
+      "step": 418
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.570385459218489e-06,
+      "loss": 3.3002,
+      "step": 419
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.567727288213005e-06,
+      "loss": 3.1779,
+      "step": 420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.565061290531323e-06,
+      "loss": 3.2428,
+      "step": 421
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.562387470741555e-06,
+      "loss": 3.3119,
+      "step": 422
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.559705833425219e-06,
+      "loss": 3.2,
+      "step": 423
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.557016383177226e-06,
+      "loss": 3.3909,
+      "step": 424
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.55431912460588e-06,
+      "loss": 3.3224,
+      "step": 425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.551614062332857e-06,
+      "loss": 3.3372,
+      "step": 426
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.548901200993206e-06,
+      "loss": 3.3294,
+      "step": 427
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.546180545235344e-06,
+      "loss": 3.2543,
+      "step": 428
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.543452099721038e-06,
+      "loss": 3.2443,
+      "step": 429
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.540715869125407e-06,
+      "loss": 3.2864,
+      "step": 430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.537971858136907e-06,
+      "loss": 3.1544,
+      "step": 431
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.535220071457325e-06,
+      "loss": 3.2375,
+      "step": 432
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.532460513801774e-06,
+      "loss": 3.313,
+      "step": 433
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.52969318989868e-06,
+      "loss": 3.2307,
+      "step": 434
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.526918104489777e-06,
+      "loss": 3.153,
+      "step": 435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.524135262330098e-06,
+      "loss": 3.3485,
+      "step": 436
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.521344668187968e-06,
+      "loss": 3.2265,
+      "step": 437
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.518546326844993e-06,
+      "loss": 3.2572,
+      "step": 438
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.515740243096056e-06,
+      "loss": 3.3338,
+      "step": 439
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.512926421749305e-06,
+      "loss": 3.3278,
+      "step": 440
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.510104867626143e-06,
+      "loss": 3.3818,
+      "step": 441
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.507275585561228e-06,
+      "loss": 3.2391,
+      "step": 442
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.504438580402462e-06,
+      "loss": 3.2106,
+      "step": 443
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.501593857010968e-06,
+      "loss": 3.2261,
+      "step": 444
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.498741420261109e-06,
+      "loss": 3.1779,
+      "step": 445
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.495881275040453e-06,
+      "loss": 3.3314,
+      "step": 446
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.493013426249782e-06,
+      "loss": 3.3439,
+      "step": 447
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.490137878803078e-06,
+      "loss": 3.3397,
+      "step": 448
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.487254637627514e-06,
+      "loss": 3.2245,
+      "step": 449
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.484363707663443e-06,
+      "loss": 3.2459,
+      "step": 450
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.481465093864395e-06,
+      "loss": 3.2537,
+      "step": 451
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.478558801197065e-06,
+      "loss": 3.1819,
+      "step": 452
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.475644834641309e-06,
+      "loss": 3.2493,
+      "step": 453
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.472723199190126e-06,
+      "loss": 3.2545,
+      "step": 454
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.469793899849663e-06,
+      "loss": 3.3515,
+      "step": 455
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.46685694163919e-06,
+      "loss": 3.1441,
+      "step": 456
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.463912329591105e-06,
+      "loss": 3.2444,
+      "step": 457
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.460960068750924e-06,
+      "loss": 3.2959,
+      "step": 458
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.458000164177264e-06,
+      "loss": 3.2786,
+      "step": 459
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.45503262094184e-06,
+      "loss": 3.2556,
+      "step": 460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.452057444129457e-06,
+      "loss": 3.2427,
+      "step": 461
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.449074638838e-06,
+      "loss": 3.2356,
+      "step": 462
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.446084210178423e-06,
+      "loss": 3.1123,
+      "step": 463
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.443086163274745e-06,
+      "loss": 3.2233,
+      "step": 464
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.440080503264038e-06,
+      "loss": 3.2498,
+      "step": 465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.437067235296418e-06,
+      "loss": 3.3066,
+      "step": 466
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.434046364535036e-06,
+      "loss": 3.3344,
+      "step": 467
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.431017896156074e-06,
+      "loss": 3.2652,
+      "step": 468
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.427981835348729e-06,
+      "loss": 3.2695,
+      "step": 469
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.42493818731521e-06,
+      "loss": 3.3554,
+      "step": 470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.421886957270723e-06,
+      "loss": 3.3186,
+      "step": 471
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.418828150443469e-06,
+      "loss": 3.2721,
+      "step": 472
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.415761772074628e-06,
+      "loss": 3.3062,
+      "step": 473
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.412687827418356e-06,
+      "loss": 3.1917,
+      "step": 474
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.409606321741776e-06,
+      "loss": 3.2094,
+      "step": 475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.406517260324962e-06,
+      "loss": 3.2368,
+      "step": 476
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.403420648460937e-06,
+      "loss": 3.1868,
+      "step": 477
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.40031649145566e-06,
+      "loss": 3.1892,
+      "step": 478
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.397204794628021e-06,
+      "loss": 3.2466,
+      "step": 479
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.394085563309827e-06,
+      "loss": 3.2514,
+      "step": 480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.390958802845797e-06,
+      "loss": 3.2894,
+      "step": 481
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.387824518593546e-06,
+      "loss": 3.2293,
+      "step": 482
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.38468271592359e-06,
+      "loss": 3.279,
+      "step": 483
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.381533400219319e-06,
+      "loss": 3.2712,
+      "step": 484
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.378376576876999e-06,
+      "loss": 3.2357,
+      "step": 485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.375212251305764e-06,
+      "loss": 3.2994,
+      "step": 486
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.372040428927595e-06,
+      "loss": 3.2258,
+      "step": 487
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.368861115177327e-06,
+      "loss": 3.2778,
+      "step": 488
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.365674315502627e-06,
+      "loss": 3.281,
+      "step": 489
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.362480035363987e-06,
+      "loss": 3.2562,
+      "step": 490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.35927828023472e-06,
+      "loss": 3.2455,
+      "step": 491
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.356069055600949e-06,
+      "loss": 3.156,
+      "step": 492
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.352852366961588e-06,
+      "loss": 3.2347,
+      "step": 493
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.349628219828348e-06,
+      "loss": 3.2718,
+      "step": 494
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.34639661972572e-06,
+      "loss": 3.1883,
+      "step": 495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.343157572190957e-06,
+      "loss": 3.269,
+      "step": 496
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.339911082774083e-06,
+      "loss": 3.2797,
+      "step": 497
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.336657157037866e-06,
+      "loss": 3.1603,
+      "step": 498
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.33339580055782e-06,
+      "loss": 3.2122,
+      "step": 499
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.330127018922195e-06,
+      "loss": 3.215,
+      "step": 500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.326850817731952e-06,
+      "loss": 3.2537,
+      "step": 501
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.323567202600777e-06,
+      "loss": 3.3396,
+      "step": 502
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.320276179155052e-06,
+      "loss": 3.1872,
+      "step": 503
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.316977753033858e-06,
+      "loss": 3.1974,
+      "step": 504
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.31367192988896e-06,
+      "loss": 3.2855,
+      "step": 505
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.310358715384793e-06,
+      "loss": 3.2451,
+      "step": 506
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.30703811519846e-06,
+      "loss": 3.2434,
+      "step": 507
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.30371013501972e-06,
+      "loss": 3.3183,
+      "step": 508
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.300374780550973e-06,
+      "loss": 3.3267,
+      "step": 509
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.297032057507264e-06,
+      "loss": 3.3623,
+      "step": 510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.293681971616252e-06,
+      "loss": 3.2339,
+      "step": 511
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.290324528618225e-06,
+      "loss": 3.3639,
+      "step": 512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.28695973426606e-06,
+      "loss": 3.211,
+      "step": 513
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.28358759432525e-06,
+      "loss": 3.1403,
+      "step": 514
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.280208114573859e-06,
+      "loss": 3.2405,
+      "step": 515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.276821300802535e-06,
+      "loss": 3.2103,
+      "step": 516
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.27342715881449e-06,
+      "loss": 3.2702,
+      "step": 517
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.270025694425497e-06,
+      "loss": 3.2341,
+      "step": 518
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.26661691346387e-06,
+      "loss": 3.2532,
+      "step": 519
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.263200821770462e-06,
+      "loss": 3.2809,
+      "step": 520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.259777425198655e-06,
+      "loss": 3.2708,
+      "step": 521
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.256346729614342e-06,
+      "loss": 3.3137,
+      "step": 522
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.252908740895932e-06,
+      "loss": 3.204,
+      "step": 523
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.24946346493432e-06,
+      "loss": 3.2348,
+      "step": 524
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.246010907632894e-06,
+      "loss": 3.2277,
+      "step": 525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.242551074907519e-06,
+      "loss": 3.2933,
+      "step": 526
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.239083972686522e-06,
+      "loss": 3.2234,
+      "step": 527
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.235609606910687e-06,
+      "loss": 3.2655,
+      "step": 528
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.232127983533247e-06,
+      "loss": 3.2639,
+      "step": 529
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.228639108519867e-06,
+      "loss": 3.2327,
+      "step": 530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.225142987848642e-06,
+      "loss": 3.2711,
+      "step": 531
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.221639627510076e-06,
+      "loss": 3.1799,
+      "step": 532
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.218129033507084e-06,
+      "loss": 3.2618,
+      "step": 533
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.214611211854973e-06,
+      "loss": 3.1684,
+      "step": 534
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.211086168581433e-06,
+      "loss": 3.2641,
+      "step": 535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.207553909726532e-06,
+      "loss": 3.2314,
+      "step": 536
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.204014441342697e-06,
+      "loss": 3.2313,
+      "step": 537
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.20046776949471e-06,
+      "loss": 3.2541,
+      "step": 538
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.196913900259699e-06,
+      "loss": 3.2482,
+      "step": 539
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.193352839727122e-06,
+      "loss": 3.264,
+      "step": 540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.189784593998757e-06,
+      "loss": 3.1881,
+      "step": 541
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.186209169188695e-06,
+      "loss": 3.2889,
+      "step": 542
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.182626571423333e-06,
+      "loss": 3.2548,
+      "step": 543
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.179036806841352e-06,
+      "loss": 3.239,
+      "step": 544
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.175439881593716e-06,
+      "loss": 3.1601,
+      "step": 545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.17183580184366e-06,
+      "loss": 3.2521,
+      "step": 546
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.168224573766673e-06,
+      "loss": 3.2785,
+      "step": 547
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.164606203550498e-06,
+      "loss": 3.3748,
+      "step": 548
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.160980697395114e-06,
+      "loss": 3.2412,
+      "step": 549
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.157348061512728e-06,
+      "loss": 3.2765,
+      "step": 550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.153708302127759e-06,
+      "loss": 3.1952,
+      "step": 551
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.150061425476839e-06,
+      "loss": 3.2652,
+      "step": 552
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.14640743780879e-06,
+      "loss": 3.2462,
+      "step": 553
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.142746345384619e-06,
+      "loss": 3.1832,
+      "step": 554
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.139078154477512e-06,
+      "loss": 3.2216,
+      "step": 555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.13540287137281e-06,
+      "loss": 3.2333,
+      "step": 556
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.131720502368013e-06,
+      "loss": 3.2228,
+      "step": 557
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.12803105377276e-06,
+      "loss": 3.278,
+      "step": 558
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.124334531908818e-06,
+      "loss": 3.1365,
+      "step": 559
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.120630943110078e-06,
+      "loss": 3.3034,
+      "step": 560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.11692029372254e-06,
+      "loss": 3.1814,
+      "step": 561
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.1132025901043e-06,
+      "loss": 3.1627,
+      "step": 562
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.10947783862554e-06,
+      "loss": 3.1275,
+      "step": 563
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.10574604566852e-06,
+      "loss": 3.212,
+      "step": 564
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.102007217627568e-06,
+      "loss": 3.2929,
+      "step": 565
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.098261360909064e-06,
+      "loss": 3.1918,
+      "step": 566
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.094508481931428e-06,
+      "loss": 3.2465,
+      "step": 567
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.090748587125118e-06,
+      "loss": 3.3306,
+      "step": 568
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.086981682932612e-06,
+      "loss": 3.3037,
+      "step": 569
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.083207775808395e-06,
+      "loss": 3.2522,
+      "step": 570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.079426872218958e-06,
+      "loss": 3.2546,
+      "step": 571
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.07563897864277e-06,
+      "loss": 3.2122,
+      "step": 572
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.071844101570291e-06,
+      "loss": 3.3023,
+      "step": 573
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.068042247503937e-06,
+      "loss": 3.194,
+      "step": 574
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.064233422958078e-06,
+      "loss": 3.2933,
+      "step": 575
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.060417634459032e-06,
+      "loss": 3.238,
+      "step": 576
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.05659488854505e-06,
+      "loss": 3.209,
+      "step": 577
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.052765191766304e-06,
+      "loss": 3.312,
+      "step": 578
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.048928550684873e-06,
+      "loss": 3.2206,
+      "step": 579
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.045084971874738e-06,
+      "loss": 3.2524,
+      "step": 580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.041234461921766e-06,
+      "loss": 3.2558,
+      "step": 581
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.0373770274237e-06,
+      "loss": 3.2616,
+      "step": 582
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.033512674990151e-06,
+      "loss": 3.1736,
+      "step": 583
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.02964141124258e-06,
+      "loss": 3.1402,
+      "step": 584
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.025763242814291e-06,
+      "loss": 3.252,
+      "step": 585
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.021878176350422e-06,
+      "loss": 3.2528,
+      "step": 586
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.01798621850793e-06,
+      "loss": 3.1614,
+      "step": 587
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.014087375955574e-06,
+      "loss": 3.2223,
+      "step": 588
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.010181655373918e-06,
+      "loss": 3.2992,
+      "step": 589
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.006269063455305e-06,
+      "loss": 3.2474,
+      "step": 590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.002349606903854e-06,
+      "loss": 3.1512,
+      "step": 591
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.998423292435455e-06,
+      "loss": 3.1838,
+      "step": 592
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.994490126777731e-06,
+      "loss": 3.1668,
+      "step": 593
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.990550116670059e-06,
+      "loss": 3.2585,
+      "step": 594
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.986603268863536e-06,
+      "loss": 3.2502,
+      "step": 595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.982649590120982e-06,
+      "loss": 3.18,
+      "step": 596
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.978689087216917e-06,
+      "loss": 3.2278,
+      "step": 597
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.97472176693755e-06,
+      "loss": 3.2473,
+      "step": 598
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.970747636080783e-06,
+      "loss": 3.3116,
+      "step": 599
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.966766701456177e-06,
+      "loss": 3.2194,
+      "step": 600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.962778969884956e-06,
+      "loss": 3.2454,
+      "step": 601
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.958784448199987e-06,
+      "loss": 3.234,
+      "step": 602
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.954783143245778e-06,
+      "loss": 3.2213,
+      "step": 603
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.950775061878453e-06,
+      "loss": 3.2561,
+      "step": 604
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.94676021096575e-06,
+      "loss": 3.2207,
+      "step": 605
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.94273859738701e-06,
+      "loss": 3.3413,
+      "step": 606
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.938710228033155e-06,
+      "loss": 3.2432,
+      "step": 607
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.934675109806688e-06,
+      "loss": 3.2921,
+      "step": 608
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.930633249621673e-06,
+      "loss": 3.2702,
+      "step": 609
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.926584654403725e-06,
+      "loss": 3.1736,
+      "step": 610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.922529331090006e-06,
+      "loss": 3.1726,
+      "step": 611
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.9184672866292e-06,
+      "loss": 3.15,
+      "step": 612
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.91439852798151e-06,
+      "loss": 3.294,
+      "step": 613
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.91032306211864e-06,
+      "loss": 3.1774,
+      "step": 614
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.906240896023794e-06,
+      "loss": 3.2373,
+      "step": 615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.902152036691649e-06,
+      "loss": 3.2387,
+      "step": 616
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.898056491128358e-06,
+      "loss": 3.2622,
+      "step": 617
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.893954266351522e-06,
+      "loss": 3.2124,
+      "step": 618
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.889845369390193e-06,
+      "loss": 3.2585,
+      "step": 619
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.885729807284855e-06,
+      "loss": 3.2835,
+      "step": 620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.881607587087412e-06,
+      "loss": 3.237,
+      "step": 621
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.877478715861173e-06,
+      "loss": 3.1625,
+      "step": 622
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.873343200680848e-06,
+      "loss": 3.2249,
+      "step": 623
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.869201048632531e-06,
+      "loss": 3.2283,
+      "step": 624
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.865052266813686e-06,
+      "loss": 3.2888,
+      "step": 625
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.860896862333135e-06,
+      "loss": 3.2022,
+      "step": 626
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.856734842311053e-06,
+      "loss": 3.1826,
+      "step": 627
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.852566213878947e-06,
+      "loss": 3.2533,
+      "step": 628
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.848390984179647e-06,
+      "loss": 3.3275,
+      "step": 629
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.844209160367298e-06,
+      "loss": 3.2859,
+      "step": 630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.84002074960734e-06,
+      "loss": 3.2522,
+      "step": 631
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.8358257590765e-06,
+      "loss": 3.2598,
+      "step": 632
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.831624195962778e-06,
+      "loss": 3.262,
+      "step": 633
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.827416067465442e-06,
+      "loss": 3.3344,
+      "step": 634
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.823201380795003e-06,
+      "loss": 3.1321,
+      "step": 635
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.818980143173212e-06,
+      "loss": 3.1976,
+      "step": 636
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.814752361833045e-06,
+      "loss": 3.2144,
+      "step": 637
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.81051804401869e-06,
+      "loss": 3.3055,
+      "step": 638
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.806277196985536e-06,
+      "loss": 3.1898,
+      "step": 639
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.802029828000157e-06,
+      "loss": 3.1974,
+      "step": 640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.797775944340303e-06,
+      "loss": 3.2199,
+      "step": 641
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.793515553294892e-06,
+      "loss": 3.2022,
+      "step": 642
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.789248662163985e-06,
+      "loss": 3.1964,
+      "step": 643
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.784975278258783e-06,
+      "loss": 3.1816,
+      "step": 644
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.780695408901613e-06,
+      "loss": 3.2692,
+      "step": 645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.776409061425919e-06,
+      "loss": 3.3504,
+      "step": 646
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.772116243176235e-06,
+      "loss": 3.197,
+      "step": 647
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.767816961508191e-06,
+      "loss": 3.2549,
+      "step": 648
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.763511223788485e-06,
+      "loss": 3.2758,
+      "step": 649
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.759199037394888e-06,
+      "loss": 3.3838,
+      "step": 650
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.754880409716208e-06,
+      "loss": 3.2936,
+      "step": 651
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.750555348152299e-06,
+      "loss": 3.2427,
+      "step": 652
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.746223860114033e-06,
+      "loss": 3.232,
+      "step": 653
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.741885953023302e-06,
+      "loss": 3.2452,
+      "step": 654
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.737541634312985e-06,
+      "loss": 3.2445,
+      "step": 655
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.733190911426957e-06,
+      "loss": 3.2265,
+      "step": 656
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.728833791820064e-06,
+      "loss": 3.2651,
+      "step": 657
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.72447028295811e-06,
+      "loss": 3.2134,
+      "step": 658
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.720100392317851e-06,
+      "loss": 3.193,
+      "step": 659
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.715724127386971e-06,
+      "loss": 3.2013,
+      "step": 660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.711341495664085e-06,
+      "loss": 3.3091,
+      "step": 661
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.706952504658712e-06,
+      "loss": 3.2111,
+      "step": 662
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.702557161891265e-06,
+      "loss": 3.3064,
+      "step": 663
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.69815547489305e-06,
+      "loss": 3.2811,
+      "step": 664
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.693747451206231e-06,
+      "loss": 3.3192,
+      "step": 665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.689333098383843e-06,
+      "loss": 3.1877,
+      "step": 666
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.684912423989755e-06,
+      "loss": 3.1826,
+      "step": 667
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.680485435598674e-06,
+      "loss": 3.2542,
+      "step": 668
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.67605214079612e-06,
+      "loss": 3.2015,
+      "step": 669
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.671612547178428e-06,
+      "loss": 3.2239,
+      "step": 670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.66716666235272e-06,
+      "loss": 3.2027,
+      "step": 671
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.662714493936895e-06,
+      "loss": 3.1454,
+      "step": 672
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.658256049559624e-06,
+      "loss": 3.2472,
+      "step": 673
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.653791336860331e-06,
+      "loss": 3.2556,
+      "step": 674
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.649320363489178e-06,
+      "loss": 3.2097,
+      "step": 675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.644843137107058e-06,
+      "loss": 3.1879,
+      "step": 676
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.640359665385575e-06,
+      "loss": 3.2022,
+      "step": 677
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.635869956007034e-06,
+      "loss": 3.2821,
+      "step": 678
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.631374016664434e-06,
+      "loss": 3.1472,
+      "step": 679
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.626871855061438e-06,
+      "loss": 3.2258,
+      "step": 680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.622363478912384e-06,
+      "loss": 3.2174,
+      "step": 681
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.617848895942246e-06,
+      "loss": 3.1946,
+      "step": 682
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.613328113886644e-06,
+      "loss": 3.1526,
+      "step": 683
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.608801140491811e-06,
+      "loss": 3.2514,
+      "step": 684
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.604267983514595e-06,
+      "loss": 3.1858,
+      "step": 685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.599728650722435e-06,
+      "loss": 3.2783,
+      "step": 686
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.595183149893353e-06,
+      "loss": 3.2067,
+      "step": 687
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.590631488815945e-06,
+      "loss": 3.247,
+      "step": 688
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.586073675289355e-06,
+      "loss": 3.2881,
+      "step": 689
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.581509717123272e-06,
+      "loss": 3.2927,
+      "step": 690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.576939622137915e-06,
+      "loss": 3.1886,
+      "step": 691
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.572363398164017e-06,
+      "loss": 3.182,
+      "step": 692
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.567781053042813e-06,
+      "loss": 3.2815,
+      "step": 693
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.563192594626027e-06,
+      "loss": 3.2567,
+      "step": 694
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.558598030775857e-06,
+      "loss": 3.2355,
+      "step": 695
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.553997369364964e-06,
+      "loss": 3.203,
+      "step": 696
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.549390618276452e-06,
+      "loss": 3.2948,
+      "step": 697
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.544777785403869e-06,
+      "loss": 3.2502,
+      "step": 698
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.540158878651173e-06,
+      "loss": 3.1626,
+      "step": 699
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.535533905932739e-06,
+      "loss": 3.3353,
+      "step": 700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.530902875173329e-06,
+      "loss": 3.2366,
+      "step": 701
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.526265794308088e-06,
+      "loss": 3.3517,
+      "step": 702
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.521622671282532e-06,
+      "loss": 3.2104,
+      "step": 703
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.51697351405252e-06,
+      "loss": 3.2289,
+      "step": 704
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.51231833058426e-06,
+      "loss": 3.2659,
+      "step": 705
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.50765712885428e-06,
+      "loss": 3.1351,
+      "step": 706
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.502989916849422e-06,
+      "loss": 3.2413,
+      "step": 707
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.498316702566828e-06,
+      "loss": 3.4203,
+      "step": 708
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.493637494013922e-06,
+      "loss": 3.1442,
+      "step": 709
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.488952299208402e-06,
+      "loss": 3.2488,
+      "step": 710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.48426112617822e-06,
+      "loss": 3.2359,
+      "step": 711
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.479563982961572e-06,
+      "loss": 3.1988,
+      "step": 712
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.474860877606887e-06,
+      "loss": 3.2822,
+      "step": 713
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.47015181817281e-06,
+      "loss": 3.2723,
+      "step": 714
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.465436812728181e-06,
+      "loss": 3.2162,
+      "step": 715
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.460715869352035e-06,
+      "loss": 3.2239,
+      "step": 716
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.455988996133583e-06,
+      "loss": 3.2199,
+      "step": 717
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.451256201172186e-06,
+      "loss": 3.2705,
+      "step": 718
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.446517492577368e-06,
+      "loss": 3.1918,
+      "step": 719
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.44177287846877e-06,
+      "loss": 3.1308,
+      "step": 720
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.437022366976165e-06,
+      "loss": 3.1468,
+      "step": 721
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.432265966239419e-06,
+      "loss": 3.1764,
+      "step": 722
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.427503684408498e-06,
+      "loss": 3.2598,
+      "step": 723
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.422735529643445e-06,
+      "loss": 3.2101,
+      "step": 724
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.417961510114357e-06,
+      "loss": 3.2766,
+      "step": 725
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.41318163400139e-06,
+      "loss": 3.2373,
+      "step": 726
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.408395909494733e-06,
+      "loss": 3.1936,
+      "step": 727
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.40360434479459e-06,
+      "loss": 3.1816,
+      "step": 728
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.398806948111181e-06,
+      "loss": 3.2694,
+      "step": 729
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.39400372766471e-06,
+      "loss": 3.3089,
+      "step": 730
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.389194691685368e-06,
+      "loss": 3.1336,
+      "step": 731
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.384379848413304e-06,
+      "loss": 3.243,
+      "step": 732
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.379559206098625e-06,
+      "loss": 3.1681,
+      "step": 733
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.374732773001367e-06,
+      "loss": 3.2069,
+      "step": 734
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.36990055739149e-06,
+      "loss": 3.1797,
+      "step": 735
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.365062567548868e-06,
+      "loss": 3.2099,
+      "step": 736
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.36021881176326e-06,
+      "loss": 3.1913,
+      "step": 737
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.355369298334316e-06,
+      "loss": 3.177,
+      "step": 738
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.350514035571539e-06,
+      "loss": 3.1906,
+      "step": 739
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.345653031794292e-06,
+      "loss": 3.2791,
+      "step": 740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.340786295331772e-06,
+      "loss": 3.1278,
+      "step": 741
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.335913834522999e-06,
+      "loss": 3.286,
+      "step": 742
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.331035657716802e-06,
+      "loss": 3.2743,
+      "step": 743
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.326151773271805e-06,
+      "loss": 3.2626,
+      "step": 744
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.32126218955641e-06,
+      "loss": 3.3073,
+      "step": 745
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.316366914948783e-06,
+      "loss": 3.2566,
+      "step": 746
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.311465957836848e-06,
+      "loss": 3.2376,
+      "step": 747
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.30655932661826e-06,
+      "loss": 3.2498,
+      "step": 748
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.301647029700396e-06,
+      "loss": 3.3053,
+      "step": 749
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.296729075500345e-06,
+      "loss": 3.2644,
+      "step": 750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.291805472444887e-06,
+      "loss": 3.2239,
+      "step": 751
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.28687622897048e-06,
+      "loss": 3.2519,
+      "step": 752
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.28194135352325e-06,
+      "loss": 3.302,
+      "step": 753
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.27700085455897e-06,
+      "loss": 3.2226,
+      "step": 754
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.272054740543053e-06,
+      "loss": 3.3108,
+      "step": 755
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.267103019950529e-06,
+      "loss": 3.2285,
+      "step": 756
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.262145701266034e-06,
+      "loss": 3.2451,
+      "step": 757
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.257182792983802e-06,
+      "loss": 3.0539,
+      "step": 758
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.252214303607639e-06,
+      "loss": 3.2336,
+      "step": 759
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.247240241650918e-06,
+      "loss": 3.4043,
+      "step": 760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.24226061563656e-06,
+      "loss": 3.2846,
+      "step": 761
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.237275434097011e-06,
+      "loss": 3.2535,
+      "step": 762
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.232284705574251e-06,
+      "loss": 3.1595,
+      "step": 763
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.227288438619754e-06,
+      "loss": 3.2562,
+      "step": 764
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.222286641794488e-06,
+      "loss": 3.3115,
+      "step": 765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.217279323668895e-06,
+      "loss": 3.2194,
+      "step": 766
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.21226649282288e-06,
+      "loss": 3.2371,
+      "step": 767
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.20724815784579e-06,
+      "loss": 3.2482,
+      "step": 768
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.202224327336406e-06,
+      "loss": 3.2071,
+      "step": 769
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.197195009902924e-06,
+      "loss": 3.2334,
+      "step": 770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.192160214162943e-06,
+      "loss": 3.2397,
+      "step": 771
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.18711994874345e-06,
+      "loss": 3.2158,
+      "step": 772
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.182074222280798e-06,
+      "loss": 3.2282,
+      "step": 773
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.177023043420706e-06,
+      "loss": 3.235,
+      "step": 774
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.171966420818227e-06,
+      "loss": 3.2136,
+      "step": 775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.166904363137752e-06,
+      "loss": 3.2061,
+      "step": 776
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.161836879052973e-06,
+      "loss": 3.2937,
+      "step": 777
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.15676397724689e-06,
+      "loss": 3.2294,
+      "step": 778
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.15168566641178e-06,
+      "loss": 3.1326,
+      "step": 779
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.146601955249187e-06,
+      "loss": 3.1861,
+      "step": 780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.141512852469919e-06,
+      "loss": 3.1417,
+      "step": 781
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.13641836679401e-06,
+      "loss": 3.2144,
+      "step": 782
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.13131850695072e-06,
+      "loss": 3.2468,
+      "step": 783
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.126213281678527e-06,
+      "loss": 3.2722,
+      "step": 784
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.12110269972509e-06,
+      "loss": 3.2945,
+      "step": 785
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.115986769847253e-06,
+      "loss": 3.2291,
+      "step": 786
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.110865500811022e-06,
+      "loss": 3.2401,
+      "step": 787
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.105738901391553e-06,
+      "loss": 3.2191,
+      "step": 788
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.100606980373133e-06,
+      "loss": 3.2247,
+      "step": 789
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.095469746549172e-06,
+      "loss": 3.2327,
+      "step": 790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.090327208722176e-06,
+      "loss": 3.2607,
+      "step": 791
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.085179375703745e-06,
+      "loss": 3.2341,
+      "step": 792
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.08002625631455e-06,
+      "loss": 3.2299,
+      "step": 793
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.074867859384322e-06,
+      "loss": 3.2125,
+      "step": 794
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.069704193751834e-06,
+      "loss": 3.2668,
+      "step": 795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.064535268264883e-06,
+      "loss": 3.1577,
+      "step": 796
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.059361091780285e-06,
+      "loss": 3.2501,
+      "step": 797
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.05418167316385e-06,
+      "loss": 3.1833,
+      "step": 798
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.04899702129037e-06,
+      "loss": 3.2276,
+      "step": 799
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.043807145043604e-06,
+      "loss": 3.1948,
+      "step": 800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.038612053316264e-06,
+      "loss": 3.2833,
+      "step": 801
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.033411755009999e-06,
+      "loss": 3.1796,
+      "step": 802
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.028206259035378e-06,
+      "loss": 3.2498,
+      "step": 803
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.022995574311876e-06,
+      "loss": 3.1624,
+      "step": 804
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.017779709767857e-06,
+      "loss": 3.1371,
+      "step": 805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.012558674340567e-06,
+      "loss": 3.2594,
+      "step": 806
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.007332476976106e-06,
+      "loss": 3.2588,
+      "step": 807
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.002101126629422e-06,
+      "loss": 3.2107,
+      "step": 808
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.996864632264287e-06,
+      "loss": 3.2491,
+      "step": 809
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.991623002853296e-06,
+      "loss": 3.1776,
+      "step": 810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.986376247377835e-06,
+      "loss": 3.1899,
+      "step": 811
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.981124374828079e-06,
+      "loss": 3.2299,
+      "step": 812
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.975867394202968e-06,
+      "loss": 3.2119,
+      "step": 813
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.970605314510194e-06,
+      "loss": 3.2754,
+      "step": 814
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.965338144766186e-06,
+      "loss": 3.2473,
+      "step": 815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.960065893996099e-06,
+      "loss": 3.2516,
+      "step": 816
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.954788571233788e-06,
+      "loss": 3.1164,
+      "step": 817
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.949506185521802e-06,
+      "loss": 3.2349,
+      "step": 818
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.944218745911367e-06,
+      "loss": 3.1655,
+      "step": 819
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.938926261462366e-06,
+      "loss": 3.2227,
+      "step": 820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.933628741243326e-06,
+      "loss": 3.1901,
+      "step": 821
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.928326194331404e-06,
+      "loss": 3.2832,
+      "step": 822
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.923018629812369e-06,
+      "loss": 3.2285,
+      "step": 823
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.917706056780588e-06,
+      "loss": 3.3164,
+      "step": 824
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.912388484339012e-06,
+      "loss": 3.1835,
+      "step": 825
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.907065921599153e-06,
+      "loss": 3.1115,
+      "step": 826
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.901738377681081e-06,
+      "loss": 3.1912,
+      "step": 827
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.896405861713393e-06,
+      "loss": 3.1811,
+      "step": 828
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.891068382833216e-06,
+      "loss": 3.2734,
+      "step": 829
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.88572595018617e-06,
+      "loss": 3.2073,
+      "step": 830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.880378572926367e-06,
+      "loss": 3.2494,
+      "step": 831
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.875026260216395e-06,
+      "loss": 3.1699,
+      "step": 832
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.869669021227293e-06,
+      "loss": 3.2252,
+      "step": 833
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.864306865138546e-06,
+      "loss": 3.3763,
+      "step": 834
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.858939801138061e-06,
+      "loss": 3.2426,
+      "step": 835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.85356783842216e-06,
+      "loss": 3.3019,
+      "step": 836
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.84819098619555e-06,
+      "loss": 3.1906,
+      "step": 837
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.842809253671321e-06,
+      "loss": 3.1998,
+      "step": 838
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.837422650070927e-06,
+      "loss": 3.1691,
+      "step": 839
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.832031184624165e-06,
+      "loss": 3.2593,
+      "step": 840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.826634866569164e-06,
+      "loss": 3.1479,
+      "step": 841
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.821233705152371e-06,
+      "loss": 3.2313,
+      "step": 842
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.815827709628526e-06,
+      "loss": 3.2401,
+      "step": 843
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.810416889260653e-06,
+      "loss": 3.1991,
+      "step": 844
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.80500125332005e-06,
+      "loss": 3.2457,
+      "step": 845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.799580811086257e-06,
+      "loss": 3.1802,
+      "step": 846
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.794155571847058e-06,
+      "loss": 3.1543,
+      "step": 847
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.788725544898452e-06,
+      "loss": 3.1861,
+      "step": 848
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.783290739544639e-06,
+      "loss": 3.3273,
+      "step": 849
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.777851165098012e-06,
+      "loss": 3.2152,
+      "step": 850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.772406830879133e-06,
+      "loss": 3.1264,
+      "step": 851
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.76695774621672e-06,
+      "loss": 3.2615,
+      "step": 852
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.761503920447636e-06,
+      "loss": 2.7846,
+      "step": 853
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.756045362916853e-06,
+      "loss": 3.2181,
+      "step": 854
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.750582082977468e-06,
+      "loss": 3.2727,
+      "step": 855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.74511408999066e-06,
+      "loss": 3.2641,
+      "step": 856
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.739641393325685e-06,
+      "loss": 3.2687,
+      "step": 857
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.734164002359864e-06,
+      "loss": 3.2,
+      "step": 858
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.72868192647855e-06,
+      "loss": 3.1803,
+      "step": 859
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.723195175075136e-06,
+      "loss": 3.2164,
+      "step": 860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.71770375755102e-06,
+      "loss": 3.258,
+      "step": 861
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.712207683315595e-06,
+      "loss": 3.2209,
+      "step": 862
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.706706961786235e-06,
+      "loss": 3.2174,
+      "step": 863
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.701201602388276e-06,
+      "loss": 3.1566,
+      "step": 864
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.695691614555002e-06,
+      "loss": 3.2083,
+      "step": 865
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.690177007727626e-06,
+      "loss": 3.2366,
+      "step": 866
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.68465779135528e-06,
+      "loss": 3.2635,
+      "step": 867
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.679133974894984e-06,
+      "loss": 3.2173,
+      "step": 868
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.673605567811652e-06,
+      "loss": 3.2062,
+      "step": 869
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.66807257957806e-06,
+      "loss": 3.2594,
+      "step": 870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.662535019674828e-06,
+      "loss": 3.2146,
+      "step": 871
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.656992897590416e-06,
+      "loss": 3.1574,
+      "step": 872
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.651446222821099e-06,
+      "loss": 3.1917,
+      "step": 873
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.645895004870953e-06,
+      "loss": 3.2257,
+      "step": 874
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.64033925325184e-06,
+      "loss": 3.2072,
+      "step": 875
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.634778977483389e-06,
+      "loss": 3.2075,
+      "step": 876
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.6292141870929784e-06,
+      "loss": 3.1872,
+      "step": 877
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.6236448916157265e-06,
+      "loss": 3.2119,
+      "step": 878
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.618071100594469e-06,
+      "loss": 3.2362,
+      "step": 879
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.612492823579744e-06,
+      "loss": 3.2198,
+      "step": 880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.60691007012978e-06,
+      "loss": 3.2129,
+      "step": 881
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.6013228498104705e-06,
+      "loss": 3.2181,
+      "step": 882
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5957311721953656e-06,
+      "loss": 3.1534,
+      "step": 883
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.590135046865652e-06,
+      "loss": 3.3588,
+      "step": 884
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.584534483410137e-06,
+      "loss": 3.2066,
+      "step": 885
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5789294914252376e-06,
+      "loss": 3.3337,
+      "step": 886
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.573320080514951e-06,
+      "loss": 3.2866,
+      "step": 887
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.5677062602908515e-06,
+      "loss": 3.2498,
+      "step": 888
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.562088040372067e-06,
+      "loss": 3.252,
+      "step": 889
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.55646543038526e-06,
+      "loss": 3.1936,
+      "step": 890
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.550838439964627e-06,
+      "loss": 3.2464,
+      "step": 891
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.545207078751858e-06,
+      "loss": 3.1959,
+      "step": 892
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.539571356396136e-06,
+      "loss": 3.221,
+      "step": 893
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.533931282554121e-06,
+      "loss": 3.2637,
+      "step": 894
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.528286866889924e-06,
+      "loss": 3.2058,
+      "step": 895
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.522638119075097e-06,
+      "loss": 3.2621,
+      "step": 896
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.516985048788617e-06,
+      "loss": 3.1907,
+      "step": 897
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.511327665716863e-06,
+      "loss": 3.2605,
+      "step": 898
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.505665979553609e-06,
+      "loss": 3.1604,
+      "step": 899
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 3.1536,
+      "step": 900
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.494329736764538e-06,
+      "loss": 3.1758,
+      "step": 901
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4886551995630615e-06,
+      "loss": 3.2479,
+      "step": 902
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.482976398118737e-06,
+      "loss": 3.2054,
+      "step": 903
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.477293342162038e-06,
+      "loss": 3.2813,
+      "step": 904
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.471606041430724e-06,
+      "loss": 3.1535,
+      "step": 905
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.465914505669829e-06,
+      "loss": 3.2613,
+      "step": 906
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4602187446316456e-06,
+      "loss": 3.075,
+      "step": 907
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.454518768075705e-06,
+      "loss": 3.2009,
+      "step": 908
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.448814585768761e-06,
+      "loss": 3.2091,
+      "step": 909
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.443106207484776e-06,
+      "loss": 3.2748,
+      "step": 910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4373936430048985e-06,
+      "loss": 3.0864,
+      "step": 911
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.431676902117453e-06,
+      "loss": 3.2262,
+      "step": 912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.425955994617919e-06,
+      "loss": 3.27,
+      "step": 913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.420230930308917e-06,
+      "loss": 3.29,
+      "step": 914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.414501719000187e-06,
+      "loss": 3.1597,
+      "step": 915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.408768370508577e-06,
+      "loss": 3.2106,
+      "step": 916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.403030894658023e-06,
+      "loss": 3.2116,
+      "step": 917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.397289301279533e-06,
+      "loss": 3.1767,
+      "step": 918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.391543600211173e-06,
+      "loss": 3.206,
+      "step": 919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.3857938012980425e-06,
+      "loss": 3.1116,
+      "step": 920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.380039914392267e-06,
+      "loss": 3.1821,
+      "step": 921
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.3742819493529725e-06,
+      "loss": 3.1724,
+      "step": 922
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.36851991604628e-06,
+      "loss": 3.2809,
+      "step": 923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.362753824345271e-06,
+      "loss": 3.1431,
+      "step": 924
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.3569836841299905e-06,
+      "loss": 3.1504,
+      "step": 925
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.351209505287412e-06,
+      "loss": 3.2581,
+      "step": 926
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.345431297711437e-06,
+      "loss": 3.2828,
+      "step": 927
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.3396490713028674e-06,
+      "loss": 3.1422,
+      "step": 928
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.333862835969389e-06,
+      "loss": 3.24,
+      "step": 929
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.328072601625558e-06,
+      "loss": 3.2324,
+      "step": 930
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.322278378192783e-06,
+      "loss": 3.2407,
+      "step": 931
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.31648017559931e-06,
+      "loss": 3.1336,
+      "step": 932
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.310678003780197e-06,
+      "loss": 3.2506,
+      "step": 933
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.304871872677313e-06,
+      "loss": 3.2562,
+      "step": 934
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2990617922393e-06,
+      "loss": 3.1182,
+      "step": 935
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.293247772421577e-06,
+      "loss": 3.1705,
+      "step": 936
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2874298231863025e-06,
+      "loss": 3.2349,
+      "step": 937
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.28160795450238e-06,
+      "loss": 3.1806,
+      "step": 938
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.27578217634542e-06,
+      "loss": 3.1935,
+      "step": 939
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.269952498697734e-06,
+      "loss": 3.2106,
+      "step": 940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.264118931548318e-06,
+      "loss": 3.2015,
+      "step": 941
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.258281484892828e-06,
+      "loss": 3.3267,
+      "step": 942
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.252440168733572e-06,
+      "loss": 3.1375,
+      "step": 943
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.246594993079483e-06,
+      "loss": 3.1335,
+      "step": 944
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.240745967946113e-06,
+      "loss": 3.187,
+      "step": 945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.2348931033556065e-06,
+      "loss": 3.2719,
+      "step": 946
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.229036409336687e-06,
+      "loss": 3.1233,
+      "step": 947
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.223175895924638e-06,
+      "loss": 3.253,
+      "step": 948
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.217311573161293e-06,
+      "loss": 3.2876,
+      "step": 949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.211443451095007e-06,
+      "loss": 3.2664,
+      "step": 950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.205571539780648e-06,
+      "loss": 3.1777,
+      "step": 951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.199695849279576e-06,
+      "loss": 3.2204,
+      "step": 952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.193816389659627e-06,
+      "loss": 3.2922,
+      "step": 953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1879331709950935e-06,
+      "loss": 3.1145,
+      "step": 954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.18204620336671e-06,
+      "loss": 3.249,
+      "step": 955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.176155496861639e-06,
+      "loss": 3.1835,
+      "step": 956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.17026106157344e-06,
+      "loss": 3.189,
+      "step": 957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.164362907602072e-06,
+      "loss": 3.2445,
+      "step": 958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.158461045053856e-06,
+      "loss": 3.1725,
+      "step": 959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1525554840414765e-06,
+      "loss": 3.2114,
+      "step": 960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.14664623468395e-06,
+      "loss": 3.1741,
+      "step": 961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.140733307106615e-06,
+      "loss": 3.2338,
+      "step": 962
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.134816711441112e-06,
+      "loss": 3.1587,
+      "step": 963
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.128896457825364e-06,
+      "loss": 3.1796,
+      "step": 964
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.1229725564035665e-06,
+      "loss": 3.1385,
+      "step": 965
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.117045017326162e-06,
+      "loss": 3.2802,
+      "step": 966
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.111113850749828e-06,
+      "loss": 3.1759,
+      "step": 967
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.105179066837456e-06,
+      "loss": 3.287,
+      "step": 968
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.099240675758137e-06,
+      "loss": 3.182,
+      "step": 969
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.093298687687141e-06,
+      "loss": 3.1236,
+      "step": 970
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.087353112805905e-06,
+      "loss": 3.2497,
+      "step": 971
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.081403961302007e-06,
+      "loss": 3.3673,
+      "step": 972
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.075451243369157e-06,
+      "loss": 3.0952,
+      "step": 973
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.069494969207175e-06,
+      "loss": 2.9158,
+      "step": 974
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.063535149021974e-06,
+      "loss": 3.2447,
+      "step": 975
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.057571793025545e-06,
+      "loss": 3.2159,
+      "step": 976
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.051604911435933e-06,
+      "loss": 3.2113,
+      "step": 977
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.04563451447723e-06,
+      "loss": 3.2338,
+      "step": 978
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0396606123795465e-06,
+      "loss": 3.2243,
+      "step": 979
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.033683215379002e-06,
+      "loss": 3.2044,
+      "step": 980
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0277023337177016e-06,
+      "loss": 3.2148,
+      "step": 981
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.021717977643726e-06,
+      "loss": 3.2006,
+      "step": 982
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.015730157411103e-06,
+      "loss": 3.1667,
+      "step": 983
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.009738883279802e-06,
+      "loss": 3.2121,
+      "step": 984
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0037441655157045e-06,
+      "loss": 3.2679,
+      "step": 985
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.997746014390601e-06,
+      "loss": 3.2013,
+      "step": 986
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.991744440182156e-06,
+      "loss": 3.323,
+      "step": 987
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.985739453173903e-06,
+      "loss": 3.1152,
+      "step": 988
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.979731063655227e-06,
+      "loss": 3.3046,
+      "step": 989
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.973719281921336e-06,
+      "loss": 3.1768,
+      "step": 990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.967704118273257e-06,
+      "loss": 3.1851,
+      "step": 991
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.961685583017808e-06,
+      "loss": 3.2819,
+      "step": 992
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.955663686467585e-06,
+      "loss": 3.1482,
+      "step": 993
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.949638438940942e-06,
+      "loss": 3.1727,
+      "step": 994
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.943609850761979e-06,
+      "loss": 3.2535,
+      "step": 995
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.9375779322605154e-06,
+      "loss": 3.2449,
+      "step": 996
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.931542693772081e-06,
+      "loss": 3.2017,
+      "step": 997
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.925504145637891e-06,
+      "loss": 3.1806,
+      "step": 998
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.919462298204834e-06,
+      "loss": 3.2014,
+      "step": 999
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.913417161825449e-06,
+      "loss": 3.2783,
+      "step": 1000
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 3.1924030780792236,
+      "eval_runtime": 18.5521,
+      "eval_samples_per_second": 17.464,
+      "eval_steps_per_second": 1.132,
+      "step": 1000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.907368746857914e-06,
+      "loss": 3.1843,
+      "step": 1001
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.9013170636660255e-06,
+      "loss": 3.1117,
+      "step": 1002
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.895262122619174e-06,
+      "loss": 3.1734,
+      "step": 1003
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.889203934092337e-06,
+      "loss": 3.2769,
+      "step": 1004
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.883142508466054e-06,
+      "loss": 3.1978,
+      "step": 1005
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.877077856126416e-06,
+      "loss": 3.2059,
+      "step": 1006
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.871009987465037e-06,
+      "loss": 3.1766,
+      "step": 1007
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8649389128790455e-06,
+      "loss": 3.2221,
+      "step": 1008
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.858864642771062e-06,
+      "loss": 3.2465,
+      "step": 1009
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.852787187549182e-06,
+      "loss": 3.1709,
+      "step": 1010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.846706557626961e-06,
+      "loss": 3.2767,
+      "step": 1011
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.840622763423391e-06,
+      "loss": 3.2402,
+      "step": 1012
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.834535815362888e-06,
+      "loss": 3.2596,
+      "step": 1013
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.828445723875272e-06,
+      "loss": 3.2332,
+      "step": 1014
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.822352499395751e-06,
+      "loss": 3.0815,
+      "step": 1015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.816256152364893e-06,
+      "loss": 3.1036,
+      "step": 1016
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.810156693228627e-06,
+      "loss": 3.3245,
+      "step": 1017
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.804054132438209e-06,
+      "loss": 3.1771,
+      "step": 1018
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.797948480450212e-06,
+      "loss": 3.0671,
+      "step": 1019
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7918397477265e-06,
+      "loss": 3.1679,
+      "step": 1020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.785727944734228e-06,
+      "loss": 3.1313,
+      "step": 1021
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.779613081945795e-06,
+      "loss": 3.1828,
+      "step": 1022
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.773495169838856e-06,
+      "loss": 3.0607,
+      "step": 1023
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.767374218896286e-06,
+      "loss": 3.2193,
+      "step": 1024
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7612502396061685e-06,
+      "loss": 3.314,
+      "step": 1025
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.755123242461774e-06,
+      "loss": 3.1525,
+      "step": 1026
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.748993237961544e-06,
+      "loss": 3.2752,
+      "step": 1027
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7428602366090764e-06,
+      "loss": 3.2415,
+      "step": 1028
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.736724248913102e-06,
+      "loss": 3.1649,
+      "step": 1029
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.730585285387465e-06,
+      "loss": 3.199,
+      "step": 1030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.724443356551116e-06,
+      "loss": 3.2808,
+      "step": 1031
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.718298472928082e-06,
+      "loss": 3.2801,
+      "step": 1032
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.71215064504745e-06,
+      "loss": 3.2697,
+      "step": 1033
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.70599988344336e-06,
+      "loss": 3.3064,
+      "step": 1034
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6998461986549715e-06,
+      "loss": 3.2652,
+      "step": 1035
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.693689601226458e-06,
+      "loss": 3.2108,
+      "step": 1036
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.687530101706978e-06,
+      "loss": 3.1805,
+      "step": 1037
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.6813677106506705e-06,
+      "loss": 3.1491,
+      "step": 1038
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.67520243861662e-06,
+      "loss": 3.2793,
+      "step": 1039
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.669034296168855e-06,
+      "loss": 3.261,
+      "step": 1040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.662863293876319e-06,
+      "loss": 3.2576,
+      "step": 1041
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.656689442312855e-06,
+      "loss": 3.1796,
+      "step": 1042
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.6505127520571914e-06,
+      "loss": 3.1776,
+      "step": 1043
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.644333233692917e-06,
+      "loss": 3.1012,
+      "step": 1044
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.638150897808469e-06,
+      "loss": 3.2509,
+      "step": 1045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.63196575499711e-06,
+      "loss": 3.1603,
+      "step": 1046
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.625777815856915e-06,
+      "loss": 3.222,
+      "step": 1047
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.619587090990748e-06,
+      "loss": 3.1362,
+      "step": 1048
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.613393591006247e-06,
+      "loss": 3.1904,
+      "step": 1049
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.607197326515808e-06,
+      "loss": 3.2328,
+      "step": 1050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.600998308136559e-06,
+      "loss": 3.1748,
+      "step": 1051
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.594796546490351e-06,
+      "loss": 3.2307,
+      "step": 1052
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.58859205220373e-06,
+      "loss": 3.1514,
+      "step": 1053
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.582384835907931e-06,
+      "loss": 3.3613,
+      "step": 1054
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.57617490823885e-06,
+      "loss": 3.196,
+      "step": 1055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.569962279837025e-06,
+      "loss": 3.2338,
+      "step": 1056
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.56374696134763e-06,
+      "loss": 3.189,
+      "step": 1057
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.557528963420442e-06,
+      "loss": 3.2181,
+      "step": 1058
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.55130829670983e-06,
+      "loss": 3.2435,
+      "step": 1059
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.545084971874738e-06,
+      "loss": 3.1537,
+      "step": 1060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5388589995786635e-06,
+      "loss": 3.2199,
+      "step": 1061
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5326303904896395e-06,
+      "loss": 3.2326,
+      "step": 1062
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.526399155280218e-06,
+      "loss": 3.1918,
+      "step": 1063
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.520165304627452e-06,
+      "loss": 3.2384,
+      "step": 1064
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.513928849212874e-06,
+      "loss": 3.2774,
+      "step": 1065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.507689799722479e-06,
+      "loss": 3.3277,
+      "step": 1066
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.501448166846711e-06,
+      "loss": 3.2043,
+      "step": 1067
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.495203961280434e-06,
+      "loss": 3.105,
+      "step": 1068
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.4889571937229275e-06,
+      "loss": 3.2116,
+      "step": 1069
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.482707874877855e-06,
+      "loss": 3.2143,
+      "step": 1070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.476456015453257e-06,
+      "loss": 3.2386,
+      "step": 1071
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.47020162616152e-06,
+      "loss": 3.2,
+      "step": 1072
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.4639447177193745e-06,
+      "loss": 3.2315,
+      "step": 1073
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.457685300847858e-06,
+      "loss": 3.2553,
+      "step": 1074
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.451423386272312e-06,
+      "loss": 3.1449,
+      "step": 1075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.445158984722358e-06,
+      "loss": 3.2194,
+      "step": 1076
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.438892106931876e-06,
+      "loss": 3.1787,
+      "step": 1077
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.432622763638993e-06,
+      "loss": 3.2361,
+      "step": 1078
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.426350965586053e-06,
+      "loss": 3.1879,
+      "step": 1079
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.420076723519615e-06,
+      "loss": 3.2372,
+      "step": 1080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.413800048190417e-06,
+      "loss": 3.1623,
+      "step": 1081
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.407520950353377e-06,
+      "loss": 3.2108,
+      "step": 1082
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.401239440767555e-06,
+      "loss": 3.2098,
+      "step": 1083
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3949555301961474e-06,
+      "loss": 3.231,
+      "step": 1084
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.388669229406462e-06,
+      "loss": 3.1708,
+      "step": 1085
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.382380549169905e-06,
+      "loss": 3.3131,
+      "step": 1086
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.376089500261958e-06,
+      "loss": 3.1258,
+      "step": 1087
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.369796093462164e-06,
+      "loss": 3.299,
+      "step": 1088
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.363500339554099e-06,
+      "loss": 3.2833,
+      "step": 1089
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.3572022493253715e-06,
+      "loss": 3.2979,
+      "step": 1090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.350901833567586e-06,
+      "loss": 3.2903,
+      "step": 1091
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.344599103076329e-06,
+      "loss": 3.1143,
+      "step": 1092
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.3382940686511625e-06,
+      "loss": 3.2035,
+      "step": 1093
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.331986741095588e-06,
+      "loss": 3.1714,
+      "step": 1094
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.325677131217041e-06,
+      "loss": 3.1564,
+      "step": 1095
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.3193652498268656e-06,
+      "loss": 3.1832,
+      "step": 1096
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.313051107740298e-06,
+      "loss": 3.1328,
+      "step": 1097
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.306734715776448e-06,
+      "loss": 3.2411,
+      "step": 1098
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.300416084758284e-06,
+      "loss": 3.2438,
+      "step": 1099
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.294095225512604e-06,
+      "loss": 3.2874,
+      "step": 1100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.287772148870032e-06,
+      "loss": 3.1164,
+      "step": 1101
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.281446865664984e-06,
+      "loss": 3.1834,
+      "step": 1102
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.275119386735662e-06,
+      "loss": 3.2337,
+      "step": 1103
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.268789722924029e-06,
+      "loss": 3.1708,
+      "step": 1104
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.26245788507579e-06,
+      "loss": 3.1671,
+      "step": 1105
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.256123884040378e-06,
+      "loss": 3.3011,
+      "step": 1106
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.249787730670932e-06,
+      "loss": 3.1563,
+      "step": 1107
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.243449435824276e-06,
+      "loss": 3.1565,
+      "step": 1108
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.237109010360904e-06,
+      "loss": 3.1826,
+      "step": 1109
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.230766465144966e-06,
+      "loss": 3.1002,
+      "step": 1110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.224421811044238e-06,
+      "loss": 3.1812,
+      "step": 1111
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.218075058930113e-06,
+      "loss": 3.1917,
+      "step": 1112
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.211726219677579e-06,
+      "loss": 3.2427,
+      "step": 1113
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.205375304165194e-06,
+      "loss": 3.1377,
+      "step": 1114
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.199022323275083e-06,
+      "loss": 3.2198,
+      "step": 1115
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.192667287892905e-06,
+      "loss": 3.2424,
+      "step": 1116
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.18631020890784e-06,
+      "loss": 3.2423,
+      "step": 1117
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.179951097212566e-06,
+      "loss": 3.3114,
+      "step": 1118
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.173589963703255e-06,
+      "loss": 3.1906,
+      "step": 1119
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.1672268192795285e-06,
+      "loss": 3.169,
+      "step": 1120
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.160861674844464e-06,
+      "loss": 3.1952,
+      "step": 1121
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.154494541304562e-06,
+      "loss": 3.1839,
+      "step": 1122
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.148125429569735e-06,
+      "loss": 3.2337,
+      "step": 1123
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.141754350553279e-06,
+      "loss": 3.1544,
+      "step": 1124
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.135381315171867e-06,
+      "loss": 3.1044,
+      "step": 1125
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.1290063343455196e-06,
+      "loss": 3.2021,
+      "step": 1126
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.1226294189975925e-06,
+      "loss": 3.1994,
+      "step": 1127
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.116250580054758e-06,
+      "loss": 3.1942,
+      "step": 1128
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.10986982844698e-06,
+      "loss": 3.2991,
+      "step": 1129
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.103487175107508e-06,
+      "loss": 3.1677,
+      "step": 1130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.097102630972837e-06,
+      "loss": 3.0634,
+      "step": 1131
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.090716206982714e-06,
+      "loss": 3.168,
+      "step": 1132
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.084327914080101e-06,
+      "loss": 3.1986,
+      "step": 1133
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.077937763211166e-06,
+      "loss": 3.2522,
+      "step": 1134
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.071545765325254e-06,
+      "loss": 3.2999,
+      "step": 1135
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.065151931374884e-06,
+      "loss": 3.3113,
+      "step": 1136
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.058756272315713e-06,
+      "loss": 3.2697,
+      "step": 1137
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.052358799106528e-06,
+      "loss": 3.2196,
+      "step": 1138
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.045959522709228e-06,
+      "loss": 3.208,
+      "step": 1139
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.039558454088796e-06,
+      "loss": 3.1626,
+      "step": 1140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.033155604213291e-06,
+      "loss": 3.1877,
+      "step": 1141
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.026750984053821e-06,
+      "loss": 3.1975,
+      "step": 1142
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.020344604584526e-06,
+      "loss": 3.2098,
+      "step": 1143
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.013936476782563e-06,
+      "loss": 3.2899,
+      "step": 1144
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.0075266116280865e-06,
+      "loss": 3.1787,
+      "step": 1145
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 6.001115020104223e-06,
+      "loss": 3.1597,
+      "step": 1146
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.994701713197063e-06,
+      "loss": 3.1874,
+      "step": 1147
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.988286701895631e-06,
+      "loss": 3.2521,
+      "step": 1148
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.981869997191876e-06,
+      "loss": 3.1721,
+      "step": 1149
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.975451610080643e-06,
+      "loss": 3.2417,
+      "step": 1150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.969031551559664e-06,
+      "loss": 3.1227,
+      "step": 1151
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.962609832629538e-06,
+      "loss": 3.2339,
+      "step": 1152
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.956186464293703e-06,
+      "loss": 3.2455,
+      "step": 1153
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.949761457558424e-06,
+      "loss": 3.2122,
+      "step": 1154
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.943334823432777e-06,
+      "loss": 3.2187,
+      "step": 1155
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.936906572928625e-06,
+      "loss": 3.2767,
+      "step": 1156
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.930476717060596e-06,
+      "loss": 3.1752,
+      "step": 1157
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9240452668460775e-06,
+      "loss": 3.2189,
+      "step": 1158
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.917612233305183e-06,
+      "loss": 3.2553,
+      "step": 1159
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.911177627460739e-06,
+      "loss": 3.1458,
+      "step": 1160
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.904741460338266e-06,
+      "loss": 3.2301,
+      "step": 1161
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.898303742965964e-06,
+      "loss": 3.177,
+      "step": 1162
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.891864486374683e-06,
+      "loss": 3.1454,
+      "step": 1163
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.885423701597918e-06,
+      "loss": 3.1867,
+      "step": 1164
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.878981399671774e-06,
+      "loss": 3.165,
+      "step": 1165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.87253759163496e-06,
+      "loss": 3.1902,
+      "step": 1166
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.866092288528768e-06,
+      "loss": 3.1391,
+      "step": 1167
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.859645501397048e-06,
+      "loss": 3.2505,
+      "step": 1168
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.853197241286194e-06,
+      "loss": 3.2384,
+      "step": 1169
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.846747519245123e-06,
+      "loss": 3.1842,
+      "step": 1170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8402963463252605e-06,
+      "loss": 3.1666,
+      "step": 1171
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8338437335805124e-06,
+      "loss": 3.2628,
+      "step": 1172
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.827389692067256e-06,
+      "loss": 3.2485,
+      "step": 1173
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.820934232844315e-06,
+      "loss": 3.1873,
+      "step": 1174
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.814477366972945e-06,
+      "loss": 3.166,
+      "step": 1175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8080191055168064e-06,
+      "loss": 3.2125,
+      "step": 1176
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.801559459541956e-06,
+      "loss": 3.2129,
+      "step": 1177
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.795098440116822e-06,
+      "loss": 3.1816,
+      "step": 1178
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7886360583121825e-06,
+      "loss": 3.1814,
+      "step": 1179
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.782172325201155e-06,
+      "loss": 3.2904,
+      "step": 1180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.775707251859168e-06,
+      "loss": 3.1919,
+      "step": 1181
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7692408493639514e-06,
+      "loss": 3.1892,
+      "step": 1182
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7627731287955054e-06,
+      "loss": 3.1724,
+      "step": 1183
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7563041012360975e-06,
+      "loss": 3.3055,
+      "step": 1184
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.749833777770225e-06,
+      "loss": 3.3335,
+      "step": 1185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.743362169484617e-06,
+      "loss": 3.2546,
+      "step": 1186
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7368892874681905e-06,
+      "loss": 3.2237,
+      "step": 1187
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.730415142812059e-06,
+      "loss": 3.218,
+      "step": 1188
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.72393974660949e-06,
+      "loss": 3.1693,
+      "step": 1189
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.717463109955896e-06,
+      "loss": 3.1623,
+      "step": 1190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.710985243948823e-06,
+      "loss": 3.1659,
+      "step": 1191
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.704506159687914e-06,
+      "loss": 3.2461,
+      "step": 1192
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.698025868274903e-06,
+      "loss": 3.1881,
+      "step": 1193
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6915443808135964e-06,
+      "loss": 3.2432,
+      "step": 1194
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6850617084098416e-06,
+      "loss": 3.168,
+      "step": 1195
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.678577862171523e-06,
+      "loss": 3.1354,
+      "step": 1196
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.672092853208535e-06,
+      "loss": 3.1912,
+      "step": 1197
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.665606692632762e-06,
+      "loss": 3.2247,
+      "step": 1198
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.659119391558065e-06,
+      "loss": 3.1545,
+      "step": 1199
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.65263096110026e-06,
+      "loss": 3.264,
+      "step": 1200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.646141412377089e-06,
+      "loss": 3.2575,
+      "step": 1201
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.639650756508222e-06,
+      "loss": 3.2718,
+      "step": 1202
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.63315900461522e-06,
+      "loss": 3.2411,
+      "step": 1203
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.626666167821522e-06,
+      "loss": 3.1743,
+      "step": 1204
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.620172257252427e-06,
+      "loss": 3.297,
+      "step": 1205
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.613677284035075e-06,
+      "loss": 3.1519,
+      "step": 1206
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.607181259298424e-06,
+      "loss": 3.3125,
+      "step": 1207
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.600684194173236e-06,
+      "loss": 3.2104,
+      "step": 1208
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.594186099792055e-06,
+      "loss": 3.0979,
+      "step": 1209
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.587686987289189e-06,
+      "loss": 3.1148,
+      "step": 1210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.58118686780069e-06,
+      "loss": 3.2809,
+      "step": 1211
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5746857524643335e-06,
+      "loss": 3.2083,
+      "step": 1212
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.5681836524196065e-06,
+      "loss": 3.1961,
+      "step": 1213
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.561680578807677e-06,
+      "loss": 3.1381,
+      "step": 1214
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.555176542771389e-06,
+      "loss": 3.2045,
+      "step": 1215
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.548671555455226e-06,
+      "loss": 3.1693,
+      "step": 1216
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.542165628005311e-06,
+      "loss": 3.1524,
+      "step": 1217
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.535658771569369e-06,
+      "loss": 3.1945,
+      "step": 1218
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.529150997296724e-06,
+      "loss": 3.1942,
+      "step": 1219
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.522642316338268e-06,
+      "loss": 3.1646,
+      "step": 1220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.51613273984645e-06,
+      "loss": 3.1388,
+      "step": 1221
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.50962227897525e-06,
+      "loss": 3.1582,
+      "step": 1222
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.503110944880169e-06,
+      "loss": 3.275,
+      "step": 1223
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.496598748718196e-06,
+      "loss": 3.1973,
+      "step": 1224
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.490085701647805e-06,
+      "loss": 3.1897,
+      "step": 1225
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.483571814828921e-06,
+      "loss": 3.2434,
+      "step": 1226
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.477057099422913e-06,
+      "loss": 3.3099,
+      "step": 1227
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.470541566592573e-06,
+      "loss": 3.1529,
+      "step": 1228
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.464025227502082e-06,
+      "loss": 3.1991,
+      "step": 1229
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.457508093317013e-06,
+      "loss": 3.1772,
+      "step": 1230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.450990175204296e-06,
+      "loss": 3.0706,
+      "step": 1231
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4444714843322085e-06,
+      "loss": 3.1609,
+      "step": 1232
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.437952031870347e-06,
+      "loss": 3.1478,
+      "step": 1233
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.431431828989618e-06,
+      "loss": 3.0769,
+      "step": 1234
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4249108868622095e-06,
+      "loss": 3.1381,
+      "step": 1235
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.41838921666158e-06,
+      "loss": 3.1338,
+      "step": 1236
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.411866829562429e-06,
+      "loss": 3.2123,
+      "step": 1237
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.405343736740695e-06,
+      "loss": 3.1142,
+      "step": 1238
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.398819949373516e-06,
+      "loss": 3.242,
+      "step": 1239
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.392295478639226e-06,
+      "loss": 3.2293,
+      "step": 1240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.385770335717327e-06,
+      "loss": 3.2592,
+      "step": 1241
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.37924453178847e-06,
+      "loss": 3.2421,
+      "step": 1242
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.372718078034449e-06,
+      "loss": 3.1198,
+      "step": 1243
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.366190985638159e-06,
+      "loss": 3.2697,
+      "step": 1244
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3596632657835975e-06,
+      "loss": 3.1407,
+      "step": 1245
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.353134929655834e-06,
+      "loss": 3.1622,
+      "step": 1246
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.346605988440995e-06,
+      "loss": 3.1838,
+      "step": 1247
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3400764533262415e-06,
+      "loss": 3.1142,
+      "step": 1248
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.333546335499756e-06,
+      "loss": 3.2389,
+      "step": 1249
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.327015646150716e-06,
+      "loss": 3.1972,
+      "step": 1250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.320484396469281e-06,
+      "loss": 3.1767,
+      "step": 1251
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3139525976465675e-06,
+      "loss": 3.2256,
+      "step": 1252
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3074202608746365e-06,
+      "loss": 3.2116,
+      "step": 1253
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.300887397346468e-06,
+      "loss": 3.1298,
+      "step": 1254
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.294354018255945e-06,
+      "loss": 3.2281,
+      "step": 1255
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.287820134797837e-06,
+      "loss": 3.197,
+      "step": 1256
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2812857581677735e-06,
+      "loss": 3.2727,
+      "step": 1257
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.27475089956223e-06,
+      "loss": 3.25,
+      "step": 1258
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.268215570178509e-06,
+      "loss": 3.1623,
+      "step": 1259
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2616797812147205e-06,
+      "loss": 3.2397,
+      "step": 1260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.255143543869759e-06,
+      "loss": 3.1748,
+      "step": 1261
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.24860686934329e-06,
+      "loss": 3.1384,
+      "step": 1262
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.242069768835728e-06,
+      "loss": 3.2042,
+      "step": 1263
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.235532253548213e-06,
+      "loss": 3.1026,
+      "step": 1264
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.228994334682605e-06,
+      "loss": 3.1507,
+      "step": 1265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.2224560234414446e-06,
+      "loss": 3.2038,
+      "step": 1266
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.215917331027952e-06,
+      "loss": 3.1757,
+      "step": 1267
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.209378268645998e-06,
+      "loss": 3.1422,
+      "step": 1268
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.20283884750009e-06,
+      "loss": 3.1846,
+      "step": 1269
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1962990787953436e-06,
+      "loss": 3.2494,
+      "step": 1270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.189758973737478e-06,
+      "loss": 3.186,
+      "step": 1271
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.183218543532782e-06,
+      "loss": 3.1453,
+      "step": 1272
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.176677799388107e-06,
+      "loss": 3.2778,
+      "step": 1273
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.170136752510837e-06,
+      "loss": 3.1194,
+      "step": 1274
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1635954141088815e-06,
+      "loss": 3.1643,
+      "step": 1275
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.157053795390642e-06,
+      "loss": 3.2604,
+      "step": 1276
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.150511907565006e-06,
+      "loss": 3.2187,
+      "step": 1277
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.143969761841317e-06,
+      "loss": 3.2622,
+      "step": 1278
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1374273694293676e-06,
+      "loss": 3.1734,
+      "step": 1279
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.130884741539367e-06,
+      "loss": 3.1383,
+      "step": 1280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.124341889381929e-06,
+      "loss": 3.2382,
+      "step": 1281
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.117798824168052e-06,
+      "loss": 3.2297,
+      "step": 1282
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1112555571091005e-06,
+      "loss": 3.208,
+      "step": 1283
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1047120994167855e-06,
+      "loss": 3.2188,
+      "step": 1284
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.098168462303141e-06,
+      "loss": 3.0841,
+      "step": 1285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.091624656980515e-06,
+      "loss": 3.197,
+      "step": 1286
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.085080694661535e-06,
+      "loss": 3.2536,
+      "step": 1287
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.078536586559104e-06,
+      "loss": 3.2338,
+      "step": 1288
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.071992343886373e-06,
+      "loss": 3.2886,
+      "step": 1289
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.065447977856723e-06,
+      "loss": 3.1157,
+      "step": 1290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.058903499683746e-06,
+      "loss": 3.1361,
+      "step": 1291
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.05235892058123e-06,
+      "loss": 3.1509,
+      "step": 1292
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.04581425176313e-06,
+      "loss": 3.1873,
+      "step": 1293
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.039269504443557e-06,
+      "loss": 3.0944,
+      "step": 1294
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.0327246898367595e-06,
+      "loss": 3.2042,
+      "step": 1295
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.0261798191570975e-06,
+      "loss": 3.2401,
+      "step": 1296
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.019634903619031e-06,
+      "loss": 3.2601,
+      "step": 1297
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.013089954437091e-06,
+      "loss": 3.2099,
+      "step": 1298
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.006544982825872e-06,
+      "loss": 3.1052,
+      "step": 1299
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5e-06,
+      "loss": 3.1165,
+      "step": 1300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.993455017174132e-06,
+      "loss": 3.2485,
+      "step": 1301
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9869100455629105e-06,
+      "loss": 3.2244,
+      "step": 1302
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.980365096380971e-06,
+      "loss": 3.1045,
+      "step": 1303
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9738201808429025e-06,
+      "loss": 3.1881,
+      "step": 1304
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.967275310163241e-06,
+      "loss": 3.1331,
+      "step": 1305
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9607304955564456e-06,
+      "loss": 3.2112,
+      "step": 1306
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.954185748236871e-06,
+      "loss": 3.2407,
+      "step": 1307
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9476410794187726e-06,
+      "loss": 3.1725,
+      "step": 1308
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.941096500316254e-06,
+      "loss": 3.2135,
+      "step": 1309
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.934552022143279e-06,
+      "loss": 3.3007,
+      "step": 1310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.928007656113628e-06,
+      "loss": 3.164,
+      "step": 1311
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.921463413440898e-06,
+      "loss": 3.198,
+      "step": 1312
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.914919305338468e-06,
+      "loss": 3.1353,
+      "step": 1313
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.908375343019487e-06,
+      "loss": 3.2526,
+      "step": 1314
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.90183153769686e-06,
+      "loss": 3.3072,
+      "step": 1315
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.895287900583216e-06,
+      "loss": 3.1138,
+      "step": 1316
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.888744442890901e-06,
+      "loss": 3.3092,
+      "step": 1317
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.88220117583195e-06,
+      "loss": 3.2259,
+      "step": 1318
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.875658110618074e-06,
+      "loss": 3.2591,
+      "step": 1319
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.869115258460636e-06,
+      "loss": 3.0782,
+      "step": 1320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.862572630570633e-06,
+      "loss": 3.1724,
+      "step": 1321
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8560302381586835e-06,
+      "loss": 3.1725,
+      "step": 1322
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.849488092434995e-06,
+      "loss": 3.269,
+      "step": 1323
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.842946204609359e-06,
+      "loss": 3.2549,
+      "step": 1324
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.83640458589112e-06,
+      "loss": 3.1116,
+      "step": 1325
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.829863247489163e-06,
+      "loss": 3.2153,
+      "step": 1326
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.823322200611895e-06,
+      "loss": 3.1436,
+      "step": 1327
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.816781456467218e-06,
+      "loss": 3.1156,
+      "step": 1328
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.810241026262524e-06,
+      "loss": 3.2352,
+      "step": 1329
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.803700921204659e-06,
+      "loss": 3.1148,
+      "step": 1330
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.797161152499911e-06,
+      "loss": 3.2119,
+      "step": 1331
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7906217313540035e-06,
+      "loss": 3.0578,
+      "step": 1332
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.784082668972048e-06,
+      "loss": 3.2253,
+      "step": 1333
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.777543976558557e-06,
+      "loss": 3.1783,
+      "step": 1334
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.771005665317398e-06,
+      "loss": 3.1936,
+      "step": 1335
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7644677464517874e-06,
+      "loss": 3.2025,
+      "step": 1336
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.757930231164275e-06,
+      "loss": 3.0976,
+      "step": 1337
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.751393130656712e-06,
+      "loss": 3.1915,
+      "step": 1338
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.744856456130243e-06,
+      "loss": 3.1763,
+      "step": 1339
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.738320218785281e-06,
+      "loss": 3.1125,
+      "step": 1340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.731784429821493e-06,
+      "loss": 3.1606,
+      "step": 1341
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.725249100437773e-06,
+      "loss": 3.1873,
+      "step": 1342
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.718714241832228e-06,
+      "loss": 3.2094,
+      "step": 1343
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.712179865202164e-06,
+      "loss": 3.0934,
+      "step": 1344
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.705645981744055e-06,
+      "loss": 3.1817,
+      "step": 1345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6991126026535336e-06,
+      "loss": 3.2807,
+      "step": 1346
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.692579739125365e-06,
+      "loss": 3.1362,
+      "step": 1347
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.686047402353433e-06,
+      "loss": 3.2797,
+      "step": 1348
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.679515603530721e-06,
+      "loss": 3.1226,
+      "step": 1349
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.672984353849285e-06,
+      "loss": 3.189,
+      "step": 1350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6664536645002456e-06,
+      "loss": 3.1425,
+      "step": 1351
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.659923546673761e-06,
+      "loss": 3.1081,
+      "step": 1352
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.653394011559007e-06,
+      "loss": 3.1693,
+      "step": 1353
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.646865070344168e-06,
+      "loss": 3.2424,
+      "step": 1354
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.640336734216403e-06,
+      "loss": 3.0934,
+      "step": 1355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6338090143618435e-06,
+      "loss": 3.2794,
+      "step": 1356
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.627281921965552e-06,
+      "loss": 3.2483,
+      "step": 1357
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.620755468211531e-06,
+      "loss": 3.206,
+      "step": 1358
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.614229664282677e-06,
+      "loss": 3.1397,
+      "step": 1359
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6077045213607765e-06,
+      "loss": 3.1163,
+      "step": 1360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.601180050626486e-06,
+      "loss": 3.2424,
+      "step": 1361
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.594656263259307e-06,
+      "loss": 3.1533,
+      "step": 1362
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.588133170437572e-06,
+      "loss": 3.1193,
+      "step": 1363
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.581610783338424e-06,
+      "loss": 3.2341,
+      "step": 1364
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.575089113137792e-06,
+      "loss": 3.2386,
+      "step": 1365
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.568568171010384e-06,
+      "loss": 3.248,
+      "step": 1366
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.562047968129654e-06,
+      "loss": 3.2058,
+      "step": 1367
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.555528515667793e-06,
+      "loss": 3.2441,
+      "step": 1368
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5490098247957045e-06,
+      "loss": 3.219,
+      "step": 1369
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.542491906682988e-06,
+      "loss": 3.1146,
+      "step": 1370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5359747724979205e-06,
+      "loss": 3.1859,
+      "step": 1371
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.529458433407429e-06,
+      "loss": 3.065,
+      "step": 1372
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5229429005770875e-06,
+      "loss": 3.2551,
+      "step": 1373
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.516428185171079e-06,
+      "loss": 3.2076,
+      "step": 1374
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.509914298352197e-06,
+      "loss": 3.1431,
+      "step": 1375
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5034012512818065e-06,
+      "loss": 3.2388,
+      "step": 1376
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.496889055119832e-06,
+      "loss": 3.2553,
+      "step": 1377
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.490377721024751e-06,
+      "loss": 3.2008,
+      "step": 1378
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.483867260153551e-06,
+      "loss": 3.137,
+      "step": 1379
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.477357683661734e-06,
+      "loss": 3.2674,
+      "step": 1380
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.470849002703279e-06,
+      "loss": 3.1706,
+      "step": 1381
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.464341228430632e-06,
+      "loss": 3.1881,
+      "step": 1382
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.457834371994691e-06,
+      "loss": 3.2533,
+      "step": 1383
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.451328444544774e-06,
+      "loss": 3.243,
+      "step": 1384
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4448234572286126e-06,
+      "loss": 3.1779,
+      "step": 1385
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.438319421192323e-06,
+      "loss": 3.2345,
+      "step": 1386
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.431816347580395e-06,
+      "loss": 3.1413,
+      "step": 1387
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.425314247535668e-06,
+      "loss": 3.1654,
+      "step": 1388
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.418813132199312e-06,
+      "loss": 3.2238,
+      "step": 1389
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.4123130127108125e-06,
+      "loss": 3.2088,
+      "step": 1390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.405813900207945e-06,
+      "loss": 3.011,
+      "step": 1391
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.399315805826765e-06,
+      "loss": 3.167,
+      "step": 1392
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.392818740701579e-06,
+      "loss": 3.1208,
+      "step": 1393
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.386322715964926e-06,
+      "loss": 3.1386,
+      "step": 1394
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.379827742747575e-06,
+      "loss": 3.2404,
+      "step": 1395
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.373333832178478e-06,
+      "loss": 3.1251,
+      "step": 1396
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3668409953847825e-06,
+      "loss": 3.1766,
+      "step": 1397
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3603492434917784e-06,
+      "loss": 3.1535,
+      "step": 1398
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.353858587622913e-06,
+      "loss": 3.2429,
+      "step": 1399
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.347369038899744e-06,
+      "loss": 3.269,
+      "step": 1400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.340880608441935e-06,
+      "loss": 3.2071,
+      "step": 1401
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3343933073672395e-06,
+      "loss": 3.2544,
+      "step": 1402
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.327907146791467e-06,
+      "loss": 3.2159,
+      "step": 1403
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.321422137828479e-06,
+      "loss": 3.0539,
+      "step": 1404
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.314938291590161e-06,
+      "loss": 3.3009,
+      "step": 1405
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.308455619186406e-06,
+      "loss": 3.1085,
+      "step": 1406
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.301974131725098e-06,
+      "loss": 3.2018,
+      "step": 1407
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.295493840312087e-06,
+      "loss": 3.183,
+      "step": 1408
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.289014756051178e-06,
+      "loss": 3.2034,
+      "step": 1409
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.282536890044105e-06,
+      "loss": 3.1388,
+      "step": 1410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.276060253390511e-06,
+      "loss": 3.1818,
+      "step": 1411
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.269584857187942e-06,
+      "loss": 3.1753,
+      "step": 1412
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.263110712531809e-06,
+      "loss": 3.2303,
+      "step": 1413
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.256637830515385e-06,
+      "loss": 3.2332,
+      "step": 1414
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.250166222229775e-06,
+      "loss": 3.2396,
+      "step": 1415
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.243695898763904e-06,
+      "loss": 3.1901,
+      "step": 1416
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.237226871204496e-06,
+      "loss": 3.1854,
+      "step": 1417
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.230759150636049e-06,
+      "loss": 3.1413,
+      "step": 1418
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2242927481408335e-06,
+      "loss": 3.2136,
+      "step": 1419
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.217827674798845e-06,
+      "loss": 3.1769,
+      "step": 1420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.211363941687819e-06,
+      "loss": 3.2533,
+      "step": 1421
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2049015598831805e-06,
+      "loss": 3.0673,
+      "step": 1422
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.198440540458045e-06,
+      "loss": 3.2192,
+      "step": 1423
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.191980894483195e-06,
+      "loss": 3.1742,
+      "step": 1424
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.185522633027057e-06,
+      "loss": 3.1711,
+      "step": 1425
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.179065767155686e-06,
+      "loss": 3.2077,
+      "step": 1426
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.172610307932747e-06,
+      "loss": 3.0881,
+      "step": 1427
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.166156266419489e-06,
+      "loss": 3.1765,
+      "step": 1428
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.159703653674741e-06,
+      "loss": 3.1444,
+      "step": 1429
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1532524807548776e-06,
+      "loss": 3.1505,
+      "step": 1430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.146802758713807e-06,
+      "loss": 3.1738,
+      "step": 1431
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.140354498602952e-06,
+      "loss": 3.2707,
+      "step": 1432
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1339077114712324e-06,
+      "loss": 3.2022,
+      "step": 1433
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1274624083650414e-06,
+      "loss": 3.3805,
+      "step": 1434
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1210186003282275e-06,
+      "loss": 3.1362,
+      "step": 1435
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.114576298402085e-06,
+      "loss": 3.1734,
+      "step": 1436
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.108135513625317e-06,
+      "loss": 3.202,
+      "step": 1437
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.1016962570340375e-06,
+      "loss": 3.1654,
+      "step": 1438
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.095258539661736e-06,
+      "loss": 3.1604,
+      "step": 1439
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.088822372539263e-06,
+      "loss": 3.1897,
+      "step": 1440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.082387766694819e-06,
+      "loss": 3.2439,
+      "step": 1441
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0759547331539224e-06,
+      "loss": 3.1981,
+      "step": 1442
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0695232829394046e-06,
+      "loss": 3.1855,
+      "step": 1443
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.063093427071376e-06,
+      "loss": 3.306,
+      "step": 1444
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.056665176567225e-06,
+      "loss": 3.1201,
+      "step": 1445
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.050238542441578e-06,
+      "loss": 3.2779,
+      "step": 1446
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.043813535706299e-06,
+      "loss": 3.2007,
+      "step": 1447
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.037390167370464e-06,
+      "loss": 3.1994,
+      "step": 1448
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.030968448440337e-06,
+      "loss": 3.1358,
+      "step": 1449
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.02454838991936e-06,
+      "loss": 3.1865,
+      "step": 1450
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.018130002808127e-06,
+      "loss": 3.2632,
+      "step": 1451
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0117132981043695e-06,
+      "loss": 3.182,
+      "step": 1452
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.005298286802938e-06,
+      "loss": 3.168,
+      "step": 1453
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.998884979895777e-06,
+      "loss": 3.1236,
+      "step": 1454
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.992473388371914e-06,
+      "loss": 3.2108,
+      "step": 1455
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.986063523217439e-06,
+      "loss": 3.1784,
+      "step": 1456
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.979655395415476e-06,
+      "loss": 3.166,
+      "step": 1457
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.973249015946182e-06,
+      "loss": 3.2462,
+      "step": 1458
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.966844395786709e-06,
+      "loss": 3.1882,
+      "step": 1459
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.960441545911205e-06,
+      "loss": 3.1534,
+      "step": 1460
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.954040477290771e-06,
+      "loss": 3.1787,
+      "step": 1461
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.947641200893473e-06,
+      "loss": 3.2078,
+      "step": 1462
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9412437276842895e-06,
+      "loss": 3.0838,
+      "step": 1463
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.934848068625117e-06,
+      "loss": 3.1682,
+      "step": 1464
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.928454234674748e-06,
+      "loss": 3.2444,
+      "step": 1465
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.922062236788836e-06,
+      "loss": 3.2046,
+      "step": 1466
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.9156720859199006e-06,
+      "loss": 3.2541,
+      "step": 1467
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.909283793017289e-06,
+      "loss": 3.2443,
+      "step": 1468
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.902897369027165e-06,
+      "loss": 3.3363,
+      "step": 1469
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8965128248924956e-06,
+      "loss": 3.2431,
+      "step": 1470
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.890130171553021e-06,
+      "loss": 3.1894,
+      "step": 1471
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.883749419945244e-06,
+      "loss": 3.1946,
+      "step": 1472
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.877370581002408e-06,
+      "loss": 3.1512,
+      "step": 1473
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.870993665654482e-06,
+      "loss": 3.2515,
+      "step": 1474
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.864618684828135e-06,
+      "loss": 3.161,
+      "step": 1475
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8582456494467214e-06,
+      "loss": 3.1688,
+      "step": 1476
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.851874570430266e-06,
+      "loss": 3.0961,
+      "step": 1477
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.845505458695438e-06,
+      "loss": 3.2013,
+      "step": 1478
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8391383251555375e-06,
+      "loss": 3.1205,
+      "step": 1479
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.832773180720475e-06,
+      "loss": 3.1814,
+      "step": 1480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.826410036296747e-06,
+      "loss": 3.0417,
+      "step": 1481
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.820048902787435e-06,
+      "loss": 3.1488,
+      "step": 1482
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.813689791092161e-06,
+      "loss": 3.1845,
+      "step": 1483
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8073327121070968e-06,
+      "loss": 3.2157,
+      "step": 1484
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.800977676724919e-06,
+      "loss": 3.1818,
+      "step": 1485
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7946246958348077e-06,
+      "loss": 3.1864,
+      "step": 1486
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7882737803224243e-06,
+      "loss": 3.3374,
+      "step": 1487
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7819249410698877e-06,
+      "loss": 3.1716,
+      "step": 1488
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.775578188955763e-06,
+      "loss": 3.3401,
+      "step": 1489
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.769233534855035e-06,
+      "loss": 3.1746,
+      "step": 1490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7628909896390976e-06,
+      "loss": 3.2549,
+      "step": 1491
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.756550564175727e-06,
+      "loss": 3.164,
+      "step": 1492
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.75021226932907e-06,
+      "loss": 3.1975,
+      "step": 1493
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.743876115959623e-06,
+      "loss": 3.2384,
+      "step": 1494
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7375421149242102e-06,
+      "loss": 3.1715,
+      "step": 1495
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7312102770759724e-06,
+      "loss": 3.2351,
+      "step": 1496
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7248806132643395e-06,
+      "loss": 3.1733,
+      "step": 1497
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7185531343350167e-06,
+      "loss": 3.1078,
+      "step": 1498
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7122278511299697e-06,
+      "loss": 3.1881,
+      "step": 1499
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.705904774487396e-06,
+      "loss": 3.1291,
+      "step": 1500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6995839152417173e-06,
+      "loss": 3.2019,
+      "step": 1501
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6932652842235537e-06,
+      "loss": 3.1588,
+      "step": 1502
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.686948892259703e-06,
+      "loss": 3.142,
+      "step": 1503
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.680634750173137e-06,
+      "loss": 3.1251,
+      "step": 1504
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6743228687829596e-06,
+      "loss": 3.1897,
+      "step": 1505
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.668013258904414e-06,
+      "loss": 3.1117,
+      "step": 1506
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.661705931348838e-06,
+      "loss": 3.2502,
+      "step": 1507
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.655400896923672e-06,
+      "loss": 3.247,
+      "step": 1508
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.649098166432417e-06,
+      "loss": 3.1607,
+      "step": 1509
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6427977506746293e-06,
+      "loss": 3.1336,
+      "step": 1510
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6364996604459017e-06,
+      "loss": 3.2196,
+      "step": 1511
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.630203906537838e-06,
+      "loss": 3.1554,
+      "step": 1512
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.623910499738043e-06,
+      "loss": 3.0993,
+      "step": 1513
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6176194508300973e-06,
+      "loss": 3.1703,
+      "step": 1514
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6113307705935398e-06,
+      "loss": 3.2074,
+      "step": 1515
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.6050444698038547e-06,
+      "loss": 3.1514,
+      "step": 1516
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5987605592324453e-06,
+      "loss": 3.1549,
+      "step": 1517
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5924790496466233e-06,
+      "loss": 3.213,
+      "step": 1518
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5861999518095827e-06,
+      "loss": 3.1459,
+      "step": 1519
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.579923276480387e-06,
+      "loss": 3.1408,
+      "step": 1520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5736490344139494e-06,
+      "loss": 3.1466,
+      "step": 1521
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5673772363610083e-06,
+      "loss": 3.2178,
+      "step": 1522
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5611078930681254e-06,
+      "loss": 3.2527,
+      "step": 1523
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5548410152776414e-06,
+      "loss": 3.2531,
+      "step": 1524
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5485766137276894e-06,
+      "loss": 3.217,
+      "step": 1525
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.542314699152145e-06,
+      "loss": 3.2459,
+      "step": 1526
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.536055282280627e-06,
+      "loss": 3.1783,
+      "step": 1527
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5297983738384813e-06,
+      "loss": 3.1679,
+      "step": 1528
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.523543984546744e-06,
+      "loss": 3.115,
+      "step": 1529
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.517292125122146e-06,
+      "loss": 3.1214,
+      "step": 1530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.511042806277075e-06,
+      "loss": 3.3255,
+      "step": 1531
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5047960387195673e-06,
+      "loss": 3.0573,
+      "step": 1532
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.498551833153292e-06,
+      "loss": 3.157,
+      "step": 1533
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.492310200277522e-06,
+      "loss": 3.1135,
+      "step": 1534
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.486071150787128e-06,
+      "loss": 3.1718,
+      "step": 1535
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4798346953725487e-06,
+      "loss": 3.2063,
+      "step": 1536
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.473600844719783e-06,
+      "loss": 3.0915,
+      "step": 1537
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4673696095103626e-06,
+      "loss": 3.2111,
+      "step": 1538
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4611410004213373e-06,
+      "loss": 3.1159,
+      "step": 1539
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4549150281252635e-06,
+      "loss": 3.1435,
+      "step": 1540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4486917032901712e-06,
+      "loss": 3.1931,
+      "step": 1541
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.44247103657956e-06,
+      "loss": 3.249,
+      "step": 1542
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.436253038652373e-06,
+      "loss": 3.2688,
+      "step": 1543
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4300377201629753e-06,
+      "loss": 3.3145,
+      "step": 1544
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4238250917611533e-06,
+      "loss": 3.2782,
+      "step": 1545
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4176151640920696e-06,
+      "loss": 3.2051,
+      "step": 1546
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4114079477962713e-06,
+      "loss": 3.165,
+      "step": 1547
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.40520345350965e-06,
+      "loss": 3.1607,
+      "step": 1548
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3990016918634415e-06,
+      "loss": 3.2165,
+      "step": 1549
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3928026734841935e-06,
+      "loss": 3.1658,
+      "step": 1550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3866064089937533e-06,
+      "loss": 3.2044,
+      "step": 1551
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3804129090092542e-06,
+      "loss": 3.1528,
+      "step": 1552
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3742221841430866e-06,
+      "loss": 3.2104,
+      "step": 1553
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.368034245002892e-06,
+      "loss": 3.1933,
+      "step": 1554
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3618491021915334e-06,
+      "loss": 3.1484,
+      "step": 1555
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.355666766307084e-06,
+      "loss": 3.1865,
+      "step": 1556
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.34948724794281e-06,
+      "loss": 3.2011,
+      "step": 1557
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3433105576871448e-06,
+      "loss": 3.2022,
+      "step": 1558
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3371367061236826e-06,
+      "loss": 3.185,
+      "step": 1559
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.330965703831146e-06,
+      "loss": 3.2317,
+      "step": 1560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3247975613833805e-06,
+      "loss": 3.3217,
+      "step": 1561
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3186322893493316e-06,
+      "loss": 3.3601,
+      "step": 1562
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.312469898293022e-06,
+      "loss": 3.1844,
+      "step": 1563
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3063103987735433e-06,
+      "loss": 3.1795,
+      "step": 1564
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.3001538013450285e-06,
+      "loss": 3.1724,
+      "step": 1565
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.294000116556641e-06,
+      "loss": 3.1773,
+      "step": 1566
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.287849354952552e-06,
+      "loss": 3.1975,
+      "step": 1567
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.28170152707192e-06,
+      "loss": 3.197,
+      "step": 1568
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2755566434488855e-06,
+      "loss": 3.1742,
+      "step": 1569
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.269414714612534e-06,
+      "loss": 3.2061,
+      "step": 1570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2632757510869005e-06,
+      "loss": 3.1339,
+      "step": 1571
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2571397633909252e-06,
+      "loss": 3.2435,
+      "step": 1572
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2510067620384566e-06,
+      "loss": 3.1469,
+      "step": 1573
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2448767575382284e-06,
+      "loss": 3.2635,
+      "step": 1574
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2387497603938327e-06,
+      "loss": 3.1707,
+      "step": 1575
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2326257811037154e-06,
+      "loss": 3.272,
+      "step": 1576
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2265048301611458e-06,
+      "loss": 3.1644,
+      "step": 1577
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.220386918054206e-06,
+      "loss": 3.0515,
+      "step": 1578
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2142720552657746e-06,
+      "loss": 3.3126,
+      "step": 1579
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2081602522734987e-06,
+      "loss": 3.1743,
+      "step": 1580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2020515195497894e-06,
+      "loss": 3.2188,
+      "step": 1581
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.195945867561791e-06,
+      "loss": 3.2716,
+      "step": 1582
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.189843306771374e-06,
+      "loss": 3.1909,
+      "step": 1583
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.183743847635109e-06,
+      "loss": 3.15,
+      "step": 1584
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.177647500604252e-06,
+      "loss": 3.2147,
+      "step": 1585
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1715542761247286e-06,
+      "loss": 3.3327,
+      "step": 1586
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1654641846371125e-06,
+      "loss": 3.2304,
+      "step": 1587
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1593772365766107e-06,
+      "loss": 3.163,
+      "step": 1588
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1532934423730423e-06,
+      "loss": 3.1834,
+      "step": 1589
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.147212812450819e-06,
+      "loss": 3.1955,
+      "step": 1590
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1411353572289404e-06,
+      "loss": 3.1665,
+      "step": 1591
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1350610871209553e-06,
+      "loss": 3.2601,
+      "step": 1592
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1289900125349646e-06,
+      "loss": 3.1716,
+      "step": 1593
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.122922143873584e-06,
+      "loss": 3.2863,
+      "step": 1594
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1168574915339465e-06,
+      "loss": 3.1872,
+      "step": 1595
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.110796065907665e-06,
+      "loss": 3.0662,
+      "step": 1596
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.104737877380828e-06,
+      "loss": 3.1208,
+      "step": 1597
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0986829363339766e-06,
+      "loss": 3.151,
+      "step": 1598
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0926312531420856e-06,
+      "loss": 3.161,
+      "step": 1599
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0865828381745515e-06,
+      "loss": 3.0703,
+      "step": 1600
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.080537701795168e-06,
+      "loss": 3.1951,
+      "step": 1601
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.07449585436211e-06,
+      "loss": 3.2302,
+      "step": 1602
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.068457306227921e-06,
+      "loss": 3.228,
+      "step": 1603
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0624220677394854e-06,
+      "loss": 3.1921,
+      "step": 1604
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.056390149238022e-06,
+      "loss": 3.314,
+      "step": 1605
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0503615610590605e-06,
+      "loss": 3.1841,
+      "step": 1606
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0443363135324167e-06,
+      "loss": 3.2433,
+      "step": 1607
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0383144169821944e-06,
+      "loss": 3.1355,
+      "step": 1608
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0322958817267428e-06,
+      "loss": 3.2017,
+      "step": 1609
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0262807180786647e-06,
+      "loss": 3.2044,
+      "step": 1610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0202689363447734e-06,
+      "loss": 3.1917,
+      "step": 1611
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0142605468260976e-06,
+      "loss": 3.1664,
+      "step": 1612
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0082555598178466e-06,
+      "loss": 3.0882,
+      "step": 1613
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.0022539856094007e-06,
+      "loss": 3.2212,
+      "step": 1614
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9962558344842963e-06,
+      "loss": 3.1681,
+      "step": 1615
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9902611167202e-06,
+      "loss": 3.1565,
+      "step": 1616
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9842698425888984e-06,
+      "loss": 3.1806,
+      "step": 1617
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9782820223562758e-06,
+      "loss": 3.1815,
+      "step": 1618
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.972297666282299e-06,
+      "loss": 3.2708,
+      "step": 1619
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.966316784621e-06,
+      "loss": 3.1248,
+      "step": 1620
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9603393876204543e-06,
+      "loss": 3.2314,
+      "step": 1621
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.954365485522771e-06,
+      "loss": 3.1952,
+      "step": 1622
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.948395088564067e-06,
+      "loss": 3.1952,
+      "step": 1623
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9424282069744564e-06,
+      "loss": 3.1409,
+      "step": 1624
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.936464850978027e-06,
+      "loss": 3.165,
+      "step": 1625
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.930505030792826e-06,
+      "loss": 3.2007,
+      "step": 1626
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9245487566308447e-06,
+      "loss": 3.1195,
+      "step": 1627
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.918596038697995e-06,
+      "loss": 3.2431,
+      "step": 1628
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.912646887194098e-06,
+      "loss": 3.2114,
+      "step": 1629
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.906701312312861e-06,
+      "loss": 3.1505,
+      "step": 1630
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.900759324241864e-06,
+      "loss": 3.2167,
+      "step": 1631
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8948209331625454e-06,
+      "loss": 3.1668,
+      "step": 1632
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8888861492501733e-06,
+      "loss": 3.1916,
+      "step": 1633
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8829549826738403e-06,
+      "loss": 3.1945,
+      "step": 1634
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8770274435964356e-06,
+      "loss": 3.2051,
+      "step": 1635
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.871103542174637e-06,
+      "loss": 3.211,
+      "step": 1636
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8651832885588893e-06,
+      "loss": 3.1546,
+      "step": 1637
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.859266692893386e-06,
+      "loss": 3.097,
+      "step": 1638
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8533537653160512e-06,
+      "loss": 3.1706,
+      "step": 1639
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8474445159585235e-06,
+      "loss": 3.1711,
+      "step": 1640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8415389549461446e-06,
+      "loss": 3.2142,
+      "step": 1641
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8356370923979326e-06,
+      "loss": 3.3211,
+      "step": 1642
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.829738938426561e-06,
+      "loss": 3.2405,
+      "step": 1643
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8238445031383634e-06,
+      "loss": 3.2107,
+      "step": 1644
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.817953796633289e-06,
+      "loss": 3.2155,
+      "step": 1645
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8120668290049085e-06,
+      "loss": 3.1446,
+      "step": 1646
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8061836103403755e-06,
+      "loss": 3.2511,
+      "step": 1647
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.800304150720424e-06,
+      "loss": 3.1332,
+      "step": 1648
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.794428460219353e-06,
+      "loss": 3.1001,
+      "step": 1649
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7885565489049948e-06,
+      "loss": 3.2587,
+      "step": 1650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.782688426838709e-06,
+      "loss": 3.2562,
+      "step": 1651
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.776824104075364e-06,
+      "loss": 3.2219,
+      "step": 1652
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.770963590663315e-06,
+      "loss": 3.1518,
+      "step": 1653
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.765106896644395e-06,
+      "loss": 3.1964,
+      "step": 1654
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.759254032053888e-06,
+      "loss": 3.1127,
+      "step": 1655
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.753405006920518e-06,
+      "loss": 3.1996,
+      "step": 1656
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7475598312664285e-06,
+      "loss": 3.1912,
+      "step": 1657
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.741718515107172e-06,
+      "loss": 3.1023,
+      "step": 1658
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7358810684516827e-06,
+      "loss": 3.1383,
+      "step": 1659
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7300475013022666e-06,
+      "loss": 3.1537,
+      "step": 1660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.724217823654582e-06,
+      "loss": 3.2243,
+      "step": 1661
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.71839204549762e-06,
+      "loss": 3.2642,
+      "step": 1662
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7125701768136974e-06,
+      "loss": 3.1532,
+      "step": 1663
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7067522275784275e-06,
+      "loss": 3.1398,
+      "step": 1664
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.700938207760701e-06,
+      "loss": 3.1508,
+      "step": 1665
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6951281273226894e-06,
+      "loss": 3.2111,
+      "step": 1666
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.689321996219802e-06,
+      "loss": 3.1152,
+      "step": 1667
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.683519824400693e-06,
+      "loss": 3.0597,
+      "step": 1668
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.677721621807217e-06,
+      "loss": 3.1634,
+      "step": 1669
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.671927398374443e-06,
+      "loss": 3.1266,
+      "step": 1670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.666137164030612e-06,
+      "loss": 3.1923,
+      "step": 1671
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6603509286971342e-06,
+      "loss": 3.1525,
+      "step": 1672
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6545687022885648e-06,
+      "loss": 3.2361,
+      "step": 1673
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6487904947125885e-06,
+      "loss": 3.1546,
+      "step": 1674
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6430163158700116e-06,
+      "loss": 3.2083,
+      "step": 1675
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.637246175654731e-06,
+      "loss": 3.1549,
+      "step": 1676
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6314800839537227e-06,
+      "loss": 3.2631,
+      "step": 1677
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6257180506470283e-06,
+      "loss": 3.1352,
+      "step": 1678
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6199600856077333e-06,
+      "loss": 3.2108,
+      "step": 1679
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.614206198701958e-06,
+      "loss": 3.1995,
+      "step": 1680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.60845639978883e-06,
+      "loss": 3.2085,
+      "step": 1681
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6027106987204677e-06,
+      "loss": 3.2011,
+      "step": 1682
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.596969105341979e-06,
+      "loss": 3.2064,
+      "step": 1683
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5912316294914232e-06,
+      "loss": 3.1667,
+      "step": 1684
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5854982809998154e-06,
+      "loss": 3.2581,
+      "step": 1685
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.579769069691084e-06,
+      "loss": 3.1819,
+      "step": 1686
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5740440053820814e-06,
+      "loss": 3.2361,
+      "step": 1687
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.568323097882548e-06,
+      "loss": 3.1283,
+      "step": 1688
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5626063569951036e-06,
+      "loss": 3.1887,
+      "step": 1689
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5568937925152272e-06,
+      "loss": 3.2228,
+      "step": 1690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5511854142312396e-06,
+      "loss": 3.1849,
+      "step": 1691
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.545481231924296e-06,
+      "loss": 3.1437,
+      "step": 1692
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5397812553683552e-06,
+      "loss": 3.1804,
+      "step": 1693
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5340854943301727e-06,
+      "loss": 3.0913,
+      "step": 1694
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5283939585692787e-06,
+      "loss": 3.2439,
+      "step": 1695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5227066578379624e-06,
+      "loss": 3.1261,
+      "step": 1696
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5170236018812626e-06,
+      "loss": 3.2982,
+      "step": 1697
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5113448004369397e-06,
+      "loss": 3.2327,
+      "step": 1698
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.505670263235464e-06,
+      "loss": 3.1953,
+      "step": 1699
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5000000000000015e-06,
+      "loss": 3.1643,
+      "step": 1700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4943340204463908e-06,
+      "loss": 3.1524,
+      "step": 1701
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4886723342831375e-06,
+      "loss": 3.1101,
+      "step": 1702
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.483014951211385e-06,
+      "loss": 3.1889,
+      "step": 1703
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4773618809249045e-06,
+      "loss": 3.1762,
+      "step": 1704
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.471713133110078e-06,
+      "loss": 3.2733,
+      "step": 1705
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4660687174458793e-06,
+      "loss": 3.2642,
+      "step": 1706
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.460428643603866e-06,
+      "loss": 3.1482,
+      "step": 1707
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4547929212481436e-06,
+      "loss": 3.2018,
+      "step": 1708
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4491615600353747e-06,
+      "loss": 3.1725,
+      "step": 1709
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4435345696147404e-06,
+      "loss": 3.0817,
+      "step": 1710
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4379119596279367e-06,
+      "loss": 3.1222,
+      "step": 1711
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.432293739709151e-06,
+      "loss": 3.0979,
+      "step": 1712
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4266799194850493e-06,
+      "loss": 3.1666,
+      "step": 1713
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4210705085747633e-06,
+      "loss": 3.1416,
+      "step": 1714
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4154655165898626e-06,
+      "loss": 3.2403,
+      "step": 1715
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.40986495313435e-06,
+      "loss": 3.1607,
+      "step": 1716
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4042688278046374e-06,
+      "loss": 3.2225,
+      "step": 1717
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.39867715018953e-06,
+      "loss": 3.201,
+      "step": 1718
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.393089929870221e-06,
+      "loss": 3.1844,
+      "step": 1719
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.387507176420256e-06,
+      "loss": 3.1997,
+      "step": 1720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.381928899405533e-06,
+      "loss": 3.1238,
+      "step": 1721
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3763551083842756e-06,
+      "loss": 3.276,
+      "step": 1722
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.370785812907022e-06,
+      "loss": 3.1979,
+      "step": 1723
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3652210225166122e-06,
+      "loss": 3.1566,
+      "step": 1724
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3596607467481602e-06,
+      "loss": 3.177,
+      "step": 1725
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.354104995129048e-06,
+      "loss": 3.2432,
+      "step": 1726
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.348553777178903e-06,
+      "loss": 3.1923,
+      "step": 1727
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3430071024095853e-06,
+      "loss": 3.2303,
+      "step": 1728
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.337464980325176e-06,
+      "loss": 3.1615,
+      "step": 1729
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3319274204219427e-06,
+      "loss": 3.1008,
+      "step": 1730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.326394432188349e-06,
+      "loss": 3.0808,
+      "step": 1731
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.320866025105016e-06,
+      "loss": 3.155,
+      "step": 1732
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3153422086447237e-06,
+      "loss": 3.1762,
+      "step": 1733
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.309822992272376e-06,
+      "loss": 3.1683,
+      "step": 1734
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.304308385444999e-06,
+      "loss": 3.1889,
+      "step": 1735
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.298798397611725e-06,
+      "loss": 3.2471,
+      "step": 1736
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.293293038213767e-06,
+      "loss": 3.2024,
+      "step": 1737
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2877923166844073e-06,
+      "loss": 3.1177,
+      "step": 1738
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2822962424489824e-06,
+      "loss": 3.2237,
+      "step": 1739
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2768048249248648e-06,
+      "loss": 3.1759,
+      "step": 1740
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.271318073521451e-06,
+      "loss": 3.2133,
+      "step": 1741
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.265835997640139e-06,
+      "loss": 3.1584,
+      "step": 1742
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.260358606674316e-06,
+      "loss": 3.1929,
+      "step": 1743
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.254885910009341e-06,
+      "loss": 3.1176,
+      "step": 1744
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2494179170225333e-06,
+      "loss": 3.1594,
+      "step": 1745
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.24395463708315e-06,
+      "loss": 3.2493,
+      "step": 1746
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2384960795523677e-06,
+      "loss": 3.2202,
+      "step": 1747
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.23304225378328e-06,
+      "loss": 3.1539,
+      "step": 1748
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2275931691208667e-06,
+      "loss": 3.2327,
+      "step": 1749
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2221488349019903e-06,
+      "loss": 3.1949,
+      "step": 1750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2167092604553637e-06,
+      "loss": 3.1777,
+      "step": 1751
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2112744551015496e-06,
+      "loss": 3.2256,
+      "step": 1752
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2058444281529423e-06,
+      "loss": 3.2266,
+      "step": 1753
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2004191889137417e-06,
+      "loss": 3.1998,
+      "step": 1754
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1949987466799524e-06,
+      "loss": 3.247,
+      "step": 1755
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1895831107393485e-06,
+      "loss": 3.2281,
+      "step": 1756
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.184172290371476e-06,
+      "loss": 3.2416,
+      "step": 1757
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1787662948476302e-06,
+      "loss": 3.1674,
+      "step": 1758
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1733651334308364e-06,
+      "loss": 3.2171,
+      "step": 1759
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1679688153758373e-06,
+      "loss": 3.2577,
+      "step": 1760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.162577349929074e-06,
+      "loss": 3.1747,
+      "step": 1761
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.15719074632868e-06,
+      "loss": 3.2818,
+      "step": 1762
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1518090138044525e-06,
+      "loss": 3.2297,
+      "step": 1763
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.146432161577842e-06,
+      "loss": 3.2626,
+      "step": 1764
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1410601988619394e-06,
+      "loss": 3.2281,
+      "step": 1765
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1356931348614546e-06,
+      "loss": 3.3066,
+      "step": 1766
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1303309787727084e-06,
+      "loss": 3.2202,
+      "step": 1767
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.124973739783609e-06,
+      "loss": 3.2564,
+      "step": 1768
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.119621427073635e-06,
+      "loss": 3.201,
+      "step": 1769
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1142740498138327e-06,
+      "loss": 3.1245,
+      "step": 1770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.108931617166784e-06,
+      "loss": 3.2043,
+      "step": 1771
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.103594138286607e-06,
+      "loss": 3.1533,
+      "step": 1772
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0982616223189196e-06,
+      "loss": 3.2064,
+      "step": 1773
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0929340784008474e-06,
+      "loss": 3.3361,
+      "step": 1774
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.08761151566099e-06,
+      "loss": 3.1962,
+      "step": 1775
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0822939432194134e-06,
+      "loss": 3.286,
+      "step": 1776
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0769813701876336e-06,
+      "loss": 3.2475,
+      "step": 1777
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.071673805668597e-06,
+      "loss": 3.2627,
+      "step": 1778
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.066371258756675e-06,
+      "loss": 3.2712,
+      "step": 1779
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.061073738537635e-06,
+      "loss": 3.1604,
+      "step": 1780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0557812540886334e-06,
+      "loss": 3.2667,
+      "step": 1781
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0504938144781987e-06,
+      "loss": 3.1408,
+      "step": 1782
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0452114287662127e-06,
+      "loss": 3.2013,
+      "step": 1783
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0399341060039023e-06,
+      "loss": 3.1878,
+      "step": 1784
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.034661855233815e-06,
+      "loss": 3.0591,
+      "step": 1785
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.029394685489808e-06,
+      "loss": 3.1647,
+      "step": 1786
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.024132605797034e-06,
+      "loss": 3.1988,
+      "step": 1787
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0188756251719204e-06,
+      "loss": 3.2197,
+      "step": 1788
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0136237526221646e-06,
+      "loss": 3.1126,
+      "step": 1789
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.008376997146705e-06,
+      "loss": 3.1941,
+      "step": 1790
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.003135367735714e-06,
+      "loss": 3.3215,
+      "step": 1791
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9978988733705807e-06,
+      "loss": 3.2469,
+      "step": 1792
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9926675230238936e-06,
+      "loss": 3.1142,
+      "step": 1793
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9874413256594343e-06,
+      "loss": 3.2038,
+      "step": 1794
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.982220290232143e-06,
+      "loss": 3.145,
+      "step": 1795
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.977004425688126e-06,
+      "loss": 3.2699,
+      "step": 1796
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9717937409646236e-06,
+      "loss": 3.2384,
+      "step": 1797
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9665882449900024e-06,
+      "loss": 3.1781,
+      "step": 1798
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9613879466837378e-06,
+      "loss": 3.214,
+      "step": 1799
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.956192854956397e-06,
+      "loss": 3.1535,
+      "step": 1800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.951002978709631e-06,
+      "loss": 3.1668,
+      "step": 1801
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9458183268361514e-06,
+      "loss": 3.2537,
+      "step": 1802
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9406389082197163e-06,
+      "loss": 3.1481,
+      "step": 1803
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9354647317351187e-06,
+      "loss": 3.1773,
+      "step": 1804
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9302958062481673e-06,
+      "loss": 3.1906,
+      "step": 1805
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9251321406156787e-06,
+      "loss": 3.1865,
+      "step": 1806
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9199737436854517e-06,
+      "loss": 3.3423,
+      "step": 1807
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9148206242962575e-06,
+      "loss": 3.1429,
+      "step": 1808
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9096727912778266e-06,
+      "loss": 3.2534,
+      "step": 1809
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9045302534508298e-06,
+      "loss": 3.0582,
+      "step": 1810
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8993930196268672e-06,
+      "loss": 3.2316,
+      "step": 1811
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8942610986084487e-06,
+      "loss": 3.2386,
+      "step": 1812
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8891344991889798e-06,
+      "loss": 3.222,
+      "step": 1813
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8840132301527497e-06,
+      "loss": 3.2744,
+      "step": 1814
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8788973002749112e-06,
+      "loss": 3.2476,
+      "step": 1815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.873786718321476e-06,
+      "loss": 3.2371,
+      "step": 1816
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8686814930492796e-06,
+      "loss": 3.1384,
+      "step": 1817
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8635816332059925e-06,
+      "loss": 3.1602,
+      "step": 1818
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8584871475300814e-06,
+      "loss": 3.2116,
+      "step": 1819
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8533980447508138e-06,
+      "loss": 3.1168,
+      "step": 1820
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8483143335882237e-06,
+      "loss": 3.1685,
+      "step": 1821
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8432360227531116e-06,
+      "loss": 3.1863,
+      "step": 1822
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.838163120947028e-06,
+      "loss": 3.2759,
+      "step": 1823
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8330956368622498e-06,
+      "loss": 3.2202,
+      "step": 1824
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8280335791817733e-06,
+      "loss": 3.1328,
+      "step": 1825
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8229769565792964e-06,
+      "loss": 3.1558,
+      "step": 1826
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8179257777192021e-06,
+      "loss": 3.1889,
+      "step": 1827
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8128800512565514e-06,
+      "loss": 3.2354,
+      "step": 1828
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8078397858370573e-06,
+      "loss": 3.1775,
+      "step": 1829
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8028049900970768e-06,
+      "loss": 3.2231,
+      "step": 1830
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.797775672663596e-06,
+      "loss": 3.2102,
+      "step": 1831
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7927518421542106e-06,
+      "loss": 3.1858,
+      "step": 1832
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7877335071771224e-06,
+      "loss": 3.2397,
+      "step": 1833
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7827206763311055e-06,
+      "loss": 3.1861,
+      "step": 1834
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.777713358205514e-06,
+      "loss": 3.163,
+      "step": 1835
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7727115613802465e-06,
+      "loss": 3.1666,
+      "step": 1836
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7677152944257514e-06,
+      "loss": 3.2776,
+      "step": 1837
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7627245659029913e-06,
+      "loss": 3.1421,
+      "step": 1838
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7577393843634426e-06,
+      "loss": 3.2383,
+      "step": 1839
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7527597583490825e-06,
+      "loss": 3.0975,
+      "step": 1840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7477856963923611e-06,
+      "loss": 3.2473,
+      "step": 1841
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7428172070161992e-06,
+      "loss": 3.1978,
+      "step": 1842
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7378542987339675e-06,
+      "loss": 3.1371,
+      "step": 1843
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7328969800494727e-06,
+      "loss": 3.1863,
+      "step": 1844
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7279452594569484e-06,
+      "loss": 3.2177,
+      "step": 1845
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.722999145441031e-06,
+      "loss": 3.1195,
+      "step": 1846
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7180586464767523e-06,
+      "loss": 3.0631,
+      "step": 1847
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7131237710295207e-06,
+      "loss": 3.2313,
+      "step": 1848
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7081945275551142e-06,
+      "loss": 3.2232,
+      "step": 1849
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7032709244996559e-06,
+      "loss": 3.2007,
+      "step": 1850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6983529702996049e-06,
+      "loss": 3.2028,
+      "step": 1851
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6934406733817417e-06,
+      "loss": 3.1802,
+      "step": 1852
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6885340421631513e-06,
+      "loss": 3.2131,
+      "step": 1853
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6836330850512162e-06,
+      "loss": 3.165,
+      "step": 1854
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6787378104435931e-06,
+      "loss": 3.1023,
+      "step": 1855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6738482267281963e-06,
+      "loss": 3.1929,
+      "step": 1856
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6689643422831992e-06,
+      "loss": 3.1881,
+      "step": 1857
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6640861654770007e-06,
+      "loss": 3.1896,
+      "step": 1858
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6592137046682305e-06,
+      "loss": 3.19,
+      "step": 1859
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6543469682057105e-06,
+      "loss": 3.2137,
+      "step": 1860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6494859644284623e-06,
+      "loss": 3.1451,
+      "step": 1861
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.644630701665686e-06,
+      "loss": 3.2735,
+      "step": 1862
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6397811882367403e-06,
+      "loss": 3.2012,
+      "step": 1863
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6349374324511347e-06,
+      "loss": 3.2878,
+      "step": 1864
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6300994426085103e-06,
+      "loss": 3.2406,
+      "step": 1865
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6252672269986352e-06,
+      "loss": 3.1178,
+      "step": 1866
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6204407939013771e-06,
+      "loss": 3.2431,
+      "step": 1867
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6156201515866971e-06,
+      "loss": 3.2698,
+      "step": 1868
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6108053083146347e-06,
+      "loss": 3.1528,
+      "step": 1869
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6059962723352912e-06,
+      "loss": 3.155,
+      "step": 1870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6011930518888214e-06,
+      "loss": 3.3231,
+      "step": 1871
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.596395655205411e-06,
+      "loss": 3.2486,
+      "step": 1872
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5916040905052693e-06,
+      "loss": 3.2149,
+      "step": 1873
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5868183659986108e-06,
+      "loss": 3.2332,
+      "step": 1874
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5820384898856433e-06,
+      "loss": 3.1919,
+      "step": 1875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5772644703565564e-06,
+      "loss": 3.2261,
+      "step": 1876
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5724963155915035e-06,
+      "loss": 3.2443,
+      "step": 1877
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5677340337605817e-06,
+      "loss": 3.19,
+      "step": 1878
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5629776330238372e-06,
+      "loss": 3.1391,
+      "step": 1879
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5582271215312294e-06,
+      "loss": 3.3166,
+      "step": 1880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5534825074226351e-06,
+      "loss": 3.2043,
+      "step": 1881
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5487437988278141e-06,
+      "loss": 3.2406,
+      "step": 1882
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5440110038664202e-06,
+      "loss": 3.2514,
+      "step": 1883
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5392841306479667e-06,
+      "loss": 3.1941,
+      "step": 1884
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5345631872718214e-06,
+      "loss": 3.1554,
+      "step": 1885
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5298481818271932e-06,
+      "loss": 3.1826,
+      "step": 1886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5251391223931127e-06,
+      "loss": 3.2067,
+      "step": 1887
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5204360170384286e-06,
+      "loss": 3.1883,
+      "step": 1888
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.515738873821782e-06,
+      "loss": 3.2002,
+      "step": 1889
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5110477007916002e-06,
+      "loss": 3.2254,
+      "step": 1890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.50636250598608e-06,
+      "loss": 3.176,
+      "step": 1891
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5016832974331725e-06,
+      "loss": 3.2679,
+      "step": 1892
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4970100831505786e-06,
+      "loss": 3.2127,
+      "step": 1893
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4923428711457217e-06,
+      "loss": 3.2688,
+      "step": 1894
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.487681669415742e-06,
+      "loss": 3.2223,
+      "step": 1895
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4830264859474814e-06,
+      "loss": 3.1553,
+      "step": 1896
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4783773287174685e-06,
+      "loss": 3.279,
+      "step": 1897
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.473734205691913e-06,
+      "loss": 3.1856,
+      "step": 1898
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4690971248266722e-06,
+      "loss": 3.2035,
+      "step": 1899
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4644660940672628e-06,
+      "loss": 3.2718,
+      "step": 1900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4598411213488285e-06,
+      "loss": 3.1762,
+      "step": 1901
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4552222145961326e-06,
+      "loss": 3.2143,
+      "step": 1902
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4506093817235495e-06,
+      "loss": 3.1554,
+      "step": 1903
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4460026306350378e-06,
+      "loss": 3.0915,
+      "step": 1904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4414019692241437e-06,
+      "loss": 3.1692,
+      "step": 1905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4368074053739733e-06,
+      "loss": 3.193,
+      "step": 1906
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4322189469571878e-06,
+      "loss": 3.1874,
+      "step": 1907
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4276366018359845e-06,
+      "loss": 3.1268,
+      "step": 1908
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4230603778620855e-06,
+      "loss": 3.1813,
+      "step": 1909
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4184902828767288e-06,
+      "loss": 3.1587,
+      "step": 1910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4139263247106466e-06,
+      "loss": 3.2151,
+      "step": 1911
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4093685111840567e-06,
+      "loss": 3.1365,
+      "step": 1912
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4048168501066478e-06,
+      "loss": 3.1456,
+      "step": 1913
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.400271349277566e-06,
+      "loss": 3.2352,
+      "step": 1914
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.395732016485406e-06,
+      "loss": 3.2583,
+      "step": 1915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3911988595081894e-06,
+      "loss": 3.201,
+      "step": 1916
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3866718861133572e-06,
+      "loss": 3.1647,
+      "step": 1917
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.382151104057754e-06,
+      "loss": 3.1635,
+      "step": 1918
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3776365210876164e-06,
+      "loss": 3.2086,
+      "step": 1919
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.373128144938563e-06,
+      "loss": 3.1588,
+      "step": 1920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.368625983335568e-06,
+      "loss": 3.2522,
+      "step": 1921
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3641300439929666e-06,
+      "loss": 3.1903,
+      "step": 1922
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3596403346144255e-06,
+      "loss": 3.1662,
+      "step": 1923
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3551568628929434e-06,
+      "loss": 3.1815,
+      "step": 1924
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3506796365108232e-06,
+      "loss": 3.1658,
+      "step": 1925
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3462086631396693e-06,
+      "loss": 3.3063,
+      "step": 1926
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3417439504403769e-06,
+      "loss": 3.2498,
+      "step": 1927
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3372855060631067e-06,
+      "loss": 3.2307,
+      "step": 1928
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3328333376472823e-06,
+      "loss": 3.1924,
+      "step": 1929
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3283874528215735e-06,
+      "loss": 3.2383,
+      "step": 1930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3239478592038802e-06,
+      "loss": 3.1362,
+      "step": 1931
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3195145644013286e-06,
+      "loss": 3.1956,
+      "step": 1932
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3150875760102467e-06,
+      "loss": 3.2433,
+      "step": 1933
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3106669016161588e-06,
+      "loss": 3.1637,
+      "step": 1934
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.30625254879377e-06,
+      "loss": 3.2802,
+      "step": 1935
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.301844525106951e-06,
+      "loss": 3.3231,
+      "step": 1936
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2974428381087351e-06,
+      "loss": 3.2267,
+      "step": 1937
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2930474953412897e-06,
+      "loss": 3.1715,
+      "step": 1938
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2886585043359156e-06,
+      "loss": 3.2458,
+      "step": 1939
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2842758726130283e-06,
+      "loss": 3.1021,
+      "step": 1940
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2798996076821497e-06,
+      "loss": 3.2047,
+      "step": 1941
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2755297170418913e-06,
+      "loss": 3.0505,
+      "step": 1942
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2711662081799365e-06,
+      "loss": 3.221,
+      "step": 1943
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2668090885730439e-06,
+      "loss": 3.1784,
+      "step": 1944
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2624583656870153e-06,
+      "loss": 3.1562,
+      "step": 1945
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2581140469767012e-06,
+      "loss": 3.1553,
+      "step": 1946
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2537761398859682e-06,
+      "loss": 3.2368,
+      "step": 1947
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2494446518477022e-06,
+      "loss": 3.1872,
+      "step": 1948
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2451195902837932e-06,
+      "loss": 3.2235,
+      "step": 1949
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2408009626051137e-06,
+      "loss": 3.1909,
+      "step": 1950
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2364887762115152e-06,
+      "loss": 3.1115,
+      "step": 1951
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2321830384918116e-06,
+      "loss": 3.2265,
+      "step": 1952
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2278837568237657e-06,
+      "loss": 3.2144,
+      "step": 1953
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2235909385740825e-06,
+      "loss": 3.2181,
+      "step": 1954
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2193045910983864e-06,
+      "loss": 3.2139,
+      "step": 1955
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2150247217412186e-06,
+      "loss": 3.1882,
+      "step": 1956
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2107513378360163e-06,
+      "loss": 3.1288,
+      "step": 1957
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.206484446705109e-06,
+      "loss": 3.1334,
+      "step": 1958
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2022240556596977e-06,
+      "loss": 3.1356,
+      "step": 1959
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1979701719998454e-06,
+      "loss": 3.2018,
+      "step": 1960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.193722803014467e-06,
+      "loss": 3.242,
+      "step": 1961
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1894819559813108e-06,
+      "loss": 3.1353,
+      "step": 1962
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1852476381669558e-06,
+      "loss": 3.1224,
+      "step": 1963
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1810198568267906e-06,
+      "loss": 3.2461,
+      "step": 1964
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1767986192049986e-06,
+      "loss": 3.2068,
+      "step": 1965
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1725839325345601e-06,
+      "loss": 3.329,
+      "step": 1966
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1683758040372222e-06,
+      "loss": 3.2015,
+      "step": 1967
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.164174240923503e-06,
+      "loss": 3.1989,
+      "step": 1968
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.159979250392661e-06,
+      "loss": 3.1735,
+      "step": 1969
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1557908396327028e-06,
+      "loss": 3.211,
+      "step": 1970
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.151609015820353e-06,
+      "loss": 3.2783,
+      "step": 1971
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1474337861210543e-06,
+      "loss": 3.2622,
+      "step": 1972
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1432651576889487e-06,
+      "loss": 3.1636,
+      "step": 1973
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1391031376668655e-06,
+      "loss": 3.2525,
+      "step": 1974
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.134947733186315e-06,
+      "loss": 3.1336,
+      "step": 1975
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1307989513674695e-06,
+      "loss": 3.2125,
+      "step": 1976
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1266567993191523e-06,
+      "loss": 3.1958,
+      "step": 1977
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1225212841388282e-06,
+      "loss": 3.2015,
+      "step": 1978
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.118392412912589e-06,
+      "loss": 3.1868,
+      "step": 1979
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1142701927151456e-06,
+      "loss": 3.156,
+      "step": 1980
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1101546306098092e-06,
+      "loss": 3.1232,
+      "step": 1981
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1060457336484802e-06,
+      "loss": 3.0993,
+      "step": 1982
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1019435088716446e-06,
+      "loss": 3.1109,
+      "step": 1983
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.097847963308351e-06,
+      "loss": 3.2362,
+      "step": 1984
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0937591039762086e-06,
+      "loss": 3.1643,
+      "step": 1985
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0896769378813609e-06,
+      "loss": 3.1893,
+      "step": 1986
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0856014720184927e-06,
+      "loss": 3.1978,
+      "step": 1987
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0815327133708015e-06,
+      "loss": 3.2592,
+      "step": 1988
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0774706689099956e-06,
+      "loss": 3.1309,
+      "step": 1989
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0734153455962765e-06,
+      "loss": 3.1571,
+      "step": 1990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.069366750378329e-06,
+      "loss": 3.16,
+      "step": 1991
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.065324890193314e-06,
+      "loss": 3.1874,
+      "step": 1992
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0612897719668457e-06,
+      "loss": 3.3061,
+      "step": 1993
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0572614026129912e-06,
+      "loss": 3.2346,
+      "step": 1994
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0532397890342506e-06,
+      "loss": 3.2171,
+      "step": 1995
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.049224938121548e-06,
+      "loss": 3.1514,
+      "step": 1996
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0452168567542237e-06,
+      "loss": 3.1468,
+      "step": 1997
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0412155518000138e-06,
+      "loss": 3.2062,
+      "step": 1998
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0372210301150464e-06,
+      "loss": 3.2015,
+      "step": 1999
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0332332985438248e-06,
+      "loss": 3.1211,
+      "step": 2000
+    },
+    {
+      "epoch": 0.8,
+      "eval_loss": 3.1742944717407227,
+      "eval_runtime": 18.5942,
+      "eval_samples_per_second": 17.425,
+      "eval_steps_per_second": 1.129,
+      "step": 2000
+    }
+  ],
+  "max_steps": 2500,
+  "num_train_epochs": 9223372036854775807,
+  "total_flos": 1.3414994804736e+17,
+  "trial_name": null,
+  "trial_params": null
+}