diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,22255 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9987859166329423,
+  "global_step": 3705,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7857142857142858e-07,
+      "loss": 0.5232,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5714285714285716e-07,
+      "loss": 0.5888,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.357142857142857e-07,
+      "loss": 0.5995,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.142857142857143e-07,
+      "loss": 0.5232,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.928571428571429e-07,
+      "loss": 0.5183,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0714285714285714e-06,
+      "loss": 0.5118,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.25e-06,
+      "loss": 0.4769,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 0.4642,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6071428571428574e-06,
+      "loss": 0.444,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7857142857142859e-06,
+      "loss": 0.4775,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9642857142857144e-06,
+      "loss": 0.4643,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1428571428571427e-06,
+      "loss": 0.4744,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.321428571428572e-06,
+      "loss": 0.4557,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-06,
+      "loss": 0.4219,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6785714285714285e-06,
+      "loss": 0.4308,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 0.4059,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0357142857142856e-06,
+      "loss": 0.4565,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2142857142857147e-06,
+      "loss": 0.4459,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3928571428571435e-06,
+      "loss": 0.4713,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5714285714285718e-06,
+      "loss": 0.4248,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 0.4345,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.928571428571429e-06,
+      "loss": 0.4138,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.107142857142857e-06,
+      "loss": 0.4617,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 0.4337,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.464285714285715e-06,
+      "loss": 0.4091,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.642857142857144e-06,
+      "loss": 0.4319,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.821428571428572e-06,
+      "loss": 0.4037,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 0.3941,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1785714285714296e-06,
+      "loss": 0.4133,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.357142857142857e-06,
+      "loss": 0.4329,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.535714285714286e-06,
+      "loss": 0.4192,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 0.3731,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.892857142857144e-06,
+      "loss": 0.3834,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.071428571428571e-06,
+      "loss": 0.4087,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.25e-06,
+      "loss": 0.4003,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.4285714285714295e-06,
+      "loss": 0.4156,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.607142857142858e-06,
+      "loss": 0.4018,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.785714285714287e-06,
+      "loss": 0.4469,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.964285714285714e-06,
+      "loss": 0.4027,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.1428571428571436e-06,
+      "loss": 0.4048,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.321428571428572e-06,
+      "loss": 0.3918,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.3952,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.67857142857143e-06,
+      "loss": 0.4033,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.857142857142858e-06,
+      "loss": 0.3752,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.035714285714286e-06,
+      "loss": 0.3983,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.214285714285714e-06,
+      "loss": 0.4196,
+      "step": 46
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.392857142857144e-06,
+      "loss": 0.375,
+      "step": 47
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.4246,
+      "step": 48
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.750000000000001e-06,
+      "loss": 0.3932,
+      "step": 49
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.92857142857143e-06,
+      "loss": 0.3867,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.107142857142858e-06,
+      "loss": 0.4003,
+      "step": 51
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.285714285714288e-06,
+      "loss": 0.4101,
+      "step": 52
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.464285714285714e-06,
+      "loss": 0.4237,
+      "step": 53
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.642857142857144e-06,
+      "loss": 0.4034,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.821428571428573e-06,
+      "loss": 0.4163,
+      "step": 55
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 0.3688,
+      "step": 56
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0178571428571429e-05,
+      "loss": 0.3539,
+      "step": 57
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0357142857142859e-05,
+      "loss": 0.3587,
+      "step": 58
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0535714285714287e-05,
+      "loss": 0.3939,
+      "step": 59
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 0.3619,
+      "step": 60
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0892857142857142e-05,
+      "loss": 0.3772,
+      "step": 61
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1071428571428572e-05,
+      "loss": 0.4062,
+      "step": 62
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.125e-05,
+      "loss": 0.4013,
+      "step": 63
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 0.3996,
+      "step": 64
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1607142857142859e-05,
+      "loss": 0.4177,
+      "step": 65
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1785714285714287e-05,
+      "loss": 0.3784,
+      "step": 66
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1964285714285716e-05,
+      "loss": 0.4099,
+      "step": 67
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2142857142857142e-05,
+      "loss": 0.3805,
+      "step": 68
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2321428571428572e-05,
+      "loss": 0.3893,
+      "step": 69
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.25e-05,
+      "loss": 0.3808,
+      "step": 70
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2678571428571429e-05,
+      "loss": 0.3836,
+      "step": 71
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2857142857142859e-05,
+      "loss": 0.3813,
+      "step": 72
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3035714285714287e-05,
+      "loss": 0.4045,
+      "step": 73
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3214285714285716e-05,
+      "loss": 0.3922,
+      "step": 74
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3392857142857142e-05,
+      "loss": 0.3784,
+      "step": 75
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3571428571428574e-05,
+      "loss": 0.4176,
+      "step": 76
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.375e-05,
+      "loss": 0.3988,
+      "step": 77
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3928571428571429e-05,
+      "loss": 0.3907,
+      "step": 78
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4107142857142859e-05,
+      "loss": 0.4343,
+      "step": 79
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 0.4076,
+      "step": 80
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4464285714285715e-05,
+      "loss": 0.3781,
+      "step": 81
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4642857142857144e-05,
+      "loss": 0.3938,
+      "step": 82
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4821428571428574e-05,
+      "loss": 0.4115,
+      "step": 83
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.4292,
+      "step": 84
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5178571428571429e-05,
+      "loss": 0.3658,
+      "step": 85
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.535714285714286e-05,
+      "loss": 0.4083,
+      "step": 86
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.553571428571429e-05,
+      "loss": 0.399,
+      "step": 87
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5714285714285715e-05,
+      "loss": 0.4139,
+      "step": 88
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5892857142857142e-05,
+      "loss": 0.3998,
+      "step": 89
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6071428571428572e-05,
+      "loss": 0.3745,
+      "step": 90
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6250000000000002e-05,
+      "loss": 0.4073,
+      "step": 91
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.642857142857143e-05,
+      "loss": 0.409,
+      "step": 92
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.660714285714286e-05,
+      "loss": 0.4014,
+      "step": 93
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.678571428571429e-05,
+      "loss": 0.4023,
+      "step": 94
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6964285714285715e-05,
+      "loss": 0.3807,
+      "step": 95
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.3957,
+      "step": 96
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7321428571428572e-05,
+      "loss": 0.4128,
+      "step": 97
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 0.3613,
+      "step": 98
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.767857142857143e-05,
+      "loss": 0.4126,
+      "step": 99
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 0.4138,
+      "step": 100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.803571428571429e-05,
+      "loss": 0.4081,
+      "step": 101
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8214285714285715e-05,
+      "loss": 0.4098,
+      "step": 102
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8392857142857142e-05,
+      "loss": 0.43,
+      "step": 103
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8571428571428575e-05,
+      "loss": 0.3932,
+      "step": 104
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 0.4521,
+      "step": 105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.892857142857143e-05,
+      "loss": 0.3701,
+      "step": 106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.910714285714286e-05,
+      "loss": 0.4304,
+      "step": 107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.928571428571429e-05,
+      "loss": 0.3868,
+      "step": 108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9464285714285715e-05,
+      "loss": 0.3988,
+      "step": 109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9642857142857145e-05,
+      "loss": 0.398,
+      "step": 110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821428571428575e-05,
+      "loss": 0.434,
+      "step": 111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.411,
+      "step": 112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999996177431404e-05,
+      "loss": 0.4098,
+      "step": 113
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999998470972854e-05,
+      "loss": 0.3513,
+      "step": 114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999965596900167e-05,
+      "loss": 0.4108,
+      "step": 115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99999388389609e-05,
+      "loss": 0.396,
+      "step": 116
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999904435931204e-05,
+      "loss": 0.4236,
+      "step": 117
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999862387837373e-05,
+      "loss": 0.4131,
+      "step": 118
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999812694711555e-05,
+      "loss": 0.422,
+      "step": 119
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999975535659174e-05,
+      "loss": 0.4013,
+      "step": 120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999690373521767e-05,
+      "loss": 0.4046,
+      "step": 121
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999617745551313e-05,
+      "loss": 0.4071,
+      "step": 122
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999537472735903e-05,
+      "loss": 0.4226,
+      "step": 123
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999944955513691e-05,
+      "loss": 0.3875,
+      "step": 124
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999935399282155e-05,
+      "loss": 0.4064,
+      "step": 125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999925078586287e-05,
+      "loss": 0.4602,
+      "step": 126
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999139934339787e-05,
+      "loss": 0.3792,
+      "step": 127
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999902143833704e-05,
+      "loss": 0.3784,
+      "step": 128
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998895297945224e-05,
+      "loss": 0.4106,
+      "step": 129
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998761513260776e-05,
+      "loss": 0.4295,
+      "step": 130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998620084385972e-05,
+      "loss": 0.4117,
+      "step": 131
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999847101142894e-05,
+      "loss": 0.3963,
+      "step": 132
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999831429450365e-05,
+      "loss": 0.4021,
+      "step": 133
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999814993372991e-05,
+      "loss": 0.4069,
+      "step": 134
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997977929233376e-05,
+      "loss": 0.3932,
+      "step": 135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999779828114555e-05,
+      "loss": 0.4167,
+      "step": 136
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997610989603775e-05,
+      "loss": 0.4026,
+      "step": 137
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997416054751237e-05,
+      "loss": 0.4158,
+      "step": 138
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999721347673697e-05,
+      "loss": 0.392,
+      "step": 139
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997003255715842e-05,
+      "loss": 0.4101,
+      "step": 140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996785391848575e-05,
+      "loss": 0.4164,
+      "step": 141
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999655988530173e-05,
+      "loss": 0.42,
+      "step": 142
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99963267362477e-05,
+      "loss": 0.4287,
+      "step": 143
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999608594486474e-05,
+      "loss": 0.4087,
+      "step": 144
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995837511336935e-05,
+      "loss": 0.4137,
+      "step": 145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995581435854217e-05,
+      "loss": 0.423,
+      "step": 146
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995317718612356e-05,
+      "loss": 0.4255,
+      "step": 147
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995046359812976e-05,
+      "loss": 0.4333,
+      "step": 148
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994767359663522e-05,
+      "loss": 0.3923,
+      "step": 149
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99944807183773e-05,
+      "loss": 0.4213,
+      "step": 150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994186436173457e-05,
+      "loss": 0.4244,
+      "step": 151
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999388451327697e-05,
+      "loss": 0.4081,
+      "step": 152
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993574949918658e-05,
+      "loss": 0.3591,
+      "step": 153
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993257746335193e-05,
+      "loss": 0.4104,
+      "step": 154
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999293290276908e-05,
+      "loss": 0.3892,
+      "step": 155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992600419468665e-05,
+      "loss": 0.3977,
+      "step": 156
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992260296688142e-05,
+      "loss": 0.4284,
+      "step": 157
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991912534687536e-05,
+      "loss": 0.3945,
+      "step": 158
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999155713373271e-05,
+      "loss": 0.4076,
+      "step": 159
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999119409409538e-05,
+      "loss": 0.415,
+      "step": 160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990823416053088e-05,
+      "loss": 0.4527,
+      "step": 161
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990445099889234e-05,
+      "loss": 0.3773,
+      "step": 162
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990059145893034e-05,
+      "loss": 0.3933,
+      "step": 163
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998966555435956e-05,
+      "loss": 0.4066,
+      "step": 164
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989264325589717e-05,
+      "loss": 0.3938,
+      "step": 165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998885545989026e-05,
+      "loss": 0.3957,
+      "step": 166
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988438957573756e-05,
+      "loss": 0.3871,
+      "step": 167
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998801481895863e-05,
+      "loss": 0.4291,
+      "step": 168
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987583044369153e-05,
+      "loss": 0.4254,
+      "step": 169
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987143634135415e-05,
+      "loss": 0.4009,
+      "step": 170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998669658859335e-05,
+      "loss": 0.3771,
+      "step": 171
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986241908084735e-05,
+      "loss": 0.4078,
+      "step": 172
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998577959295717e-05,
+      "loss": 0.3719,
+      "step": 173
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985309643564115e-05,
+      "loss": 0.3973,
+      "step": 174
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984832060264842e-05,
+      "loss": 0.432,
+      "step": 175
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984346843424475e-05,
+      "loss": 0.4183,
+      "step": 176
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983853993413968e-05,
+      "loss": 0.4264,
+      "step": 177
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998335351061011e-05,
+      "loss": 0.4169,
+      "step": 178
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998284539539553e-05,
+      "loss": 0.3849,
+      "step": 179
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982329648158684e-05,
+      "loss": 0.4176,
+      "step": 180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981806269293873e-05,
+      "loss": 0.4276,
+      "step": 181
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981275259201226e-05,
+      "loss": 0.4179,
+      "step": 182
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980736618286708e-05,
+      "loss": 0.3981,
+      "step": 183
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980190346962115e-05,
+      "loss": 0.3927,
+      "step": 184
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979636445645077e-05,
+      "loss": 0.4061,
+      "step": 185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979074914759064e-05,
+      "loss": 0.3894,
+      "step": 186
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978505754733375e-05,
+      "loss": 0.4337,
+      "step": 187
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977928966003137e-05,
+      "loss": 0.3645,
+      "step": 188
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997734454900931e-05,
+      "loss": 0.3969,
+      "step": 189
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99767525041987e-05,
+      "loss": 0.4329,
+      "step": 190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976152832023922e-05,
+      "loss": 0.3881,
+      "step": 191
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975545532943435e-05,
+      "loss": 0.408,
+      "step": 192
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974930607421536e-05,
+      "loss": 0.3903,
+      "step": 193
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974308055928337e-05,
+      "loss": 0.4006,
+      "step": 194
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973677878939783e-05,
+      "loss": 0.3973,
+      "step": 195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973040076937666e-05,
+      "loss": 0.3886,
+      "step": 196
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997239465040958e-05,
+      "loss": 0.3932,
+      "step": 197
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971741599848973e-05,
+      "loss": 0.3739,
+      "step": 198
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971080925755103e-05,
+      "loss": 0.379,
+      "step": 199
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970412628633075e-05,
+      "loss": 0.4215,
+      "step": 200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969736708993802e-05,
+      "loss": 0.3778,
+      "step": 201
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969053167354035e-05,
+      "loss": 0.3893,
+      "step": 202
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968362004236353e-05,
+      "loss": 0.3777,
+      "step": 203
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9967663220169163e-05,
+      "loss": 0.4106,
+      "step": 204
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966956815686687e-05,
+      "loss": 0.381,
+      "step": 205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966242791328988e-05,
+      "loss": 0.411,
+      "step": 206
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965521147641947e-05,
+      "loss": 0.41,
+      "step": 207
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964791885177264e-05,
+      "loss": 0.4112,
+      "step": 208
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964055004492477e-05,
+      "loss": 0.4209,
+      "step": 209
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963310506150938e-05,
+      "loss": 0.3762,
+      "step": 210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962558390721827e-05,
+      "loss": 0.3892,
+      "step": 211
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961798658780146e-05,
+      "loss": 0.3989,
+      "step": 212
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961031310906722e-05,
+      "loss": 0.3665,
+      "step": 213
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960256347688203e-05,
+      "loss": 0.3614,
+      "step": 214
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959473769717058e-05,
+      "loss": 0.3904,
+      "step": 215
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958683577591576e-05,
+      "loss": 0.4293,
+      "step": 216
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957885771915873e-05,
+      "loss": 0.4075,
+      "step": 217
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957080353299886e-05,
+      "loss": 0.3954,
+      "step": 218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995626732235936e-05,
+      "loss": 0.3868,
+      "step": 219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955446679715874e-05,
+      "loss": 0.3795,
+      "step": 220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995461842599682e-05,
+      "loss": 0.3836,
+      "step": 221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953782561835405e-05,
+      "loss": 0.4161,
+      "step": 222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995293908787067e-05,
+      "loss": 0.3984,
+      "step": 223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995208800474745e-05,
+      "loss": 0.3684,
+      "step": 224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995122931311641e-05,
+      "loss": 0.4284,
+      "step": 225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995036301363404e-05,
+      "loss": 0.416,
+      "step": 226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949489106962635e-05,
+      "loss": 0.3849,
+      "step": 227
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948607593770303e-05,
+      "loss": 0.4009,
+      "step": 228
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994771847473098e-05,
+      "loss": 0.4141,
+      "step": 229
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946821750524406e-05,
+      "loss": 0.388,
+      "step": 230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945917421836145e-05,
+      "loss": 0.3738,
+      "step": 231
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994500548935756e-05,
+      "loss": 0.3857,
+      "step": 232
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944085953785838e-05,
+      "loss": 0.3954,
+      "step": 233
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994315881582398e-05,
+      "loss": 0.406,
+      "step": 234
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994222407618079e-05,
+      "loss": 0.3476,
+      "step": 235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99412817355709e-05,
+      "loss": 0.3928,
+      "step": 236
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940331794714732e-05,
+      "loss": 0.412,
+      "step": 237
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939374254338532e-05,
+      "loss": 0.4127,
+      "step": 238
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993840911517435e-05,
+      "loss": 0.4001,
+      "step": 239
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937436377960055e-05,
+      "loss": 0.3982,
+      "step": 240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9936456043439313e-05,
+      "loss": 0.3935,
+      "step": 241
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935468112361605e-05,
+      "loss": 0.3493,
+      "step": 242
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934472585482214e-05,
+      "loss": 0.3942,
+      "step": 243
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993346946356224e-05,
+      "loss": 0.3766,
+      "step": 244
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932458747368582e-05,
+      "loss": 0.4005,
+      "step": 245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931440437673938e-05,
+      "loss": 0.4004,
+      "step": 246
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930414535256833e-05,
+      "loss": 0.3907,
+      "step": 247
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929381040901575e-05,
+      "loss": 0.3747,
+      "step": 248
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928339955398287e-05,
+      "loss": 0.3913,
+      "step": 249
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927291279542892e-05,
+      "loss": 0.3749,
+      "step": 250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992623501413712e-05,
+      "loss": 0.4015,
+      "step": 251
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925171159988495e-05,
+      "loss": 0.4236,
+      "step": 252
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924099717910356e-05,
+      "loss": 0.4382,
+      "step": 253
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992302068872183e-05,
+      "loss": 0.3976,
+      "step": 254
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921934073247844e-05,
+      "loss": 0.386,
+      "step": 255
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920839872319145e-05,
+      "loss": 0.3864,
+      "step": 256
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991973808677225e-05,
+      "loss": 0.4325,
+      "step": 257
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99186287174495e-05,
+      "loss": 0.3603,
+      "step": 258
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917511765199018e-05,
+      "loss": 0.3922,
+      "step": 259
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916387230874726e-05,
+      "loss": 0.404,
+      "step": 260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915255115336352e-05,
+      "loss": 0.4165,
+      "step": 261
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914115419449415e-05,
+      "loss": 0.3917,
+      "step": 262
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912968144085217e-05,
+      "loss": 0.3784,
+      "step": 263
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991181329012088e-05,
+      "loss": 0.3736,
+      "step": 264
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99106508584393e-05,
+      "loss": 0.4064,
+      "step": 265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9909480849929165e-05,
+      "loss": 0.4106,
+      "step": 266
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908303265484973e-05,
+      "loss": 0.39,
+      "step": 267
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907118106006997e-05,
+      "loss": 0.3873,
+      "step": 268
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905925372401312e-05,
+      "loss": 0.4148,
+      "step": 269
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904725065579775e-05,
+      "loss": 0.4188,
+      "step": 270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903517186460043e-05,
+      "loss": 0.4318,
+      "step": 271
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990230173596555e-05,
+      "loss": 0.3814,
+      "step": 272
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901078715025527e-05,
+      "loss": 0.3697,
+      "step": 273
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899848124574986e-05,
+      "loss": 0.389,
+      "step": 274
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989860996555474e-05,
+      "loss": 0.4185,
+      "step": 275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989736423891137e-05,
+      "loss": 0.3589,
+      "step": 276
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896110945597257e-05,
+      "loss": 0.3862,
+      "step": 277
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9894850086570553e-05,
+      "loss": 0.4008,
+      "step": 278
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989358166279521e-05,
+      "loss": 0.3965,
+      "step": 279
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892305675240953e-05,
+      "loss": 0.3884,
+      "step": 280
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891022124883292e-05,
+      "loss": 0.3841,
+      "step": 281
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889731012703517e-05,
+      "loss": 0.4484,
+      "step": 282
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888432339688704e-05,
+      "loss": 0.3847,
+      "step": 283
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98871261068317e-05,
+      "loss": 0.3913,
+      "step": 284
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988581231513115e-05,
+      "loss": 0.3948,
+      "step": 285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884490965591453e-05,
+      "loss": 0.3835,
+      "step": 286
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883162059222805e-05,
+      "loss": 0.4034,
+      "step": 287
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988182559704117e-05,
+      "loss": 0.4151,
+      "step": 288
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98804815800683e-05,
+      "loss": 0.3778,
+      "step": 289
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879130009331704e-05,
+      "loss": 0.3646,
+      "step": 290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987777088586468e-05,
+      "loss": 0.4013,
+      "step": 291
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98764042107063e-05,
+      "loss": 0.3844,
+      "step": 292
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98750299849014e-05,
+      "loss": 0.3935,
+      "step": 293
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98736482095006e-05,
+      "loss": 0.3831,
+      "step": 294
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872258885560286e-05,
+      "loss": 0.4018,
+      "step": 295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987086201414261e-05,
+      "loss": 0.4351,
+      "step": 296
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869457596315504e-05,
+      "loss": 0.3913,
+      "step": 297
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868045633152662e-05,
+      "loss": 0.4092,
+      "step": 298
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866626125733552e-05,
+      "loss": 0.3893,
+      "step": 299
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865199075143404e-05,
+      "loss": 0.3742,
+      "step": 300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986376448247322e-05,
+      "loss": 0.3869,
+      "step": 301
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862322348819763e-05,
+      "loss": 0.372,
+      "step": 302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9860872675285567e-05,
+      "loss": 0.3895,
+      "step": 303
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859415462978928e-05,
+      "loss": 0.3957,
+      "step": 304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9857950713013902e-05,
+      "loss": 0.3852,
+      "step": 305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985647842651031e-05,
+      "loss": 0.4034,
+      "step": 306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854998604593737e-05,
+      "loss": 0.3838,
+      "step": 307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853511248395528e-05,
+      "loss": 0.3842,
+      "step": 308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852016359052783e-05,
+      "loss": 0.4523,
+      "step": 309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985051393770837e-05,
+      "loss": 0.3625,
+      "step": 310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849003985510912e-05,
+      "loss": 0.3986,
+      "step": 311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984748650361478e-05,
+      "loss": 0.4148,
+      "step": 312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984596149318012e-05,
+      "loss": 0.4057,
+      "step": 313
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9844428955372814e-05,
+      "loss": 0.3687,
+      "step": 314
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9842888891364517e-05,
+      "loss": 0.4026,
+      "step": 315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9841341302332623e-05,
+      "loss": 0.3672,
+      "step": 316
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9839786189460285e-05,
+      "loss": 0.3686,
+      "step": 317
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838223553936412e-05,
+      "loss": 0.3951,
+      "step": 318
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836653396955656e-05,
+      "loss": 0.3877,
+      "step": 319
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835075719718428e-05,
+      "loss": 0.3988,
+      "step": 320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983349052343088e-05,
+      "loss": 0.3978,
+      "step": 321
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831897809304918e-05,
+      "loss": 0.3633,
+      "step": 322
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9830297578558195e-05,
+      "loss": 0.3942,
+      "step": 323
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9828689832414106e-05,
+      "loss": 0.3825,
+      "step": 324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9827074572101798e-05,
+      "loss": 0.4239,
+      "step": 325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982545179885616e-05,
+      "loss": 0.4287,
+      "step": 326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982382151391782e-05,
+      "loss": 0.3972,
+      "step": 327
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9822183718533158e-05,
+      "loss": 0.3715,
+      "step": 328
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982053841395429e-05,
+      "loss": 0.4128,
+      "step": 329
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981888560143907e-05,
+      "loss": 0.4317,
+      "step": 330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9817225282251106e-05,
+      "loss": 0.3901,
+      "step": 331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981555745765972e-05,
+      "loss": 0.3966,
+      "step": 332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981388212894e-05,
+      "loss": 0.4187,
+      "step": 333
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812199297372753e-05,
+      "loss": 0.3994,
+      "step": 334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9810508964244524e-05,
+      "loss": 0.3805,
+      "step": 335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808811130847596e-05,
+      "loss": 0.4158,
+      "step": 336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9807105798479995e-05,
+      "loss": 0.4244,
+      "step": 337
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805392968445457e-05,
+      "loss": 0.3794,
+      "step": 338
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9803672642053473e-05,
+      "loss": 0.4133,
+      "step": 339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9801944820619256e-05,
+      "loss": 0.356,
+      "step": 340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.980020950546375e-05,
+      "loss": 0.4194,
+      "step": 341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798466697913617e-05,
+      "loss": 0.3628,
+      "step": 342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979671639930127e-05,
+      "loss": 0.3597,
+      "step": 343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9794958610964832e-05,
+      "loss": 0.4201,
+      "step": 344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9793193334248152e-05,
+      "loss": 0.3865,
+      "step": 345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9791420570500813e-05,
+      "loss": 0.4007,
+      "step": 346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978964032107812e-05,
+      "loss": 0.3713,
+      "step": 347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787852587341088e-05,
+      "loss": 0.4233,
+      "step": 348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9786057370656474e-05,
+      "loss": 0.424,
+      "step": 349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9784254672396743e-05,
+      "loss": 0.4003,
+      "step": 350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978244449394008e-05,
+      "loss": 0.3781,
+      "step": 351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978062683667039e-05,
+      "loss": 0.3956,
+      "step": 352
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9778801701977305e-05,
+      "loss": 0.4061,
+      "step": 353
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776969091256158e-05,
+      "loss": 0.4425,
+      "step": 354
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775129005908006e-05,
+      "loss": 0.3692,
+      "step": 355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977328144733962e-05,
+      "loss": 0.3966,
+      "step": 356
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9771426416963488e-05,
+      "loss": 0.3655,
+      "step": 357
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.97695639161978e-05,
+      "loss": 0.382,
+      "step": 358
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9767693946466465e-05,
+      "loss": 0.403,
+      "step": 359
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9765816509199105e-05,
+      "loss": 0.3638,
+      "step": 360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9763931605831043e-05,
+      "loss": 0.3815,
+      "step": 361
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976203923780331e-05,
+      "loss": 0.381,
+      "step": 362
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9760139406562656e-05,
+      "loss": 0.4069,
+      "step": 363
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975823211356152e-05,
+      "loss": 0.3858,
+      "step": 364
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9756317360258057e-05,
+      "loss": 0.4069,
+      "step": 365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975439514811612e-05,
+      "loss": 0.4356,
+      "step": 366
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9752465478605272e-05,
+      "loss": 0.4088,
+      "step": 367
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9750528353200765e-05,
+      "loss": 0.432,
+      "step": 368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748583773383564e-05,
+      "loss": 0.397,
+      "step": 369
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9746631740640323e-05,
+      "loss": 0.4002,
+      "step": 370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.97446722564634e-05,
+      "loss": 0.3905,
+      "step": 371
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742705322350846e-05,
+      "loss": 0.4286,
+      "step": 372
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974073093980641e-05,
+      "loss": 0.4092,
+      "step": 373
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973874911033953e-05,
+      "loss": 0.4029,
+      "step": 374
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9736759835465352e-05,
+      "loss": 0.3745,
+      "step": 375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9734763116704692e-05,
+      "loss": 0.3995,
+      "step": 376
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9732758955584082e-05,
+      "loss": 0.3956,
+      "step": 377
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973074735363572e-05,
+      "loss": 0.3948,
+      "step": 378
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9728728312397507e-05,
+      "loss": 0.3947,
+      "step": 379
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9726701833413028e-05,
+      "loss": 0.4095,
+      "step": 380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972466791823155e-05,
+      "loss": 0.3958,
+      "step": 381
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9722626568408033e-05,
+      "loss": 0.4105,
+      "step": 382
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9720577785503118e-05,
+      "loss": 0.3973,
+      "step": 383
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971852157108313e-05,
+      "loss": 0.3762,
+      "step": 384
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9716457926720064e-05,
+      "loss": 0.438,
+      "step": 385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971438685399161e-05,
+      "loss": 0.3754,
+      "step": 386
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9712308354481133e-05,
+      "loss": 0.3826,
+      "step": 387
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9710222429777673e-05,
+      "loss": 0.3885,
+      "step": 388
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9708129081475945e-05,
+      "loss": 0.3697,
+      "step": 389
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9706028311176348e-05,
+      "loss": 0.3844,
+      "step": 390
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970392012048495e-05,
+      "loss": 0.4041,
+      "step": 391
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970180451101348e-05,
+      "loss": 0.3709,
+      "step": 392
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9699681484379364e-05,
+      "loss": 0.4016,
+      "step": 393
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9697551042205682e-05,
+      "loss": 0.3702,
+      "step": 394
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969541318612118e-05,
+      "loss": 0.4097,
+      "step": 395
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9693267917760283e-05,
+      "loss": 0.3505,
+      "step": 396
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969111523876308e-05,
+      "loss": 0.393,
+      "step": 397
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968895515077532e-05,
+      "loss": 0.3953,
+      "step": 398
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968678765544842e-05,
+      "loss": 0.3596,
+      "step": 399
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9684612754439463e-05,
+      "loss": 0.3938,
+      "step": 400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968243044941119e-05,
+      "loss": 0.367,
+      "step": 401
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9680240742032e-05,
+      "loss": 0.4049,
+      "step": 402
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9678043633975956e-05,
+      "loss": 0.3779,
+      "step": 403
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9675839126922777e-05,
+      "loss": 0.4081,
+      "step": 404
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9673627222557844e-05,
+      "loss": 0.41,
+      "step": 405
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9671407922572178e-05,
+      "loss": 0.3573,
+      "step": 406
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9669181228662473e-05,
+      "loss": 0.3918,
+      "step": 407
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9666947142531062e-05,
+      "loss": 0.3942,
+      "step": 408
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9664705665885938e-05,
+      "loss": 0.4162,
+      "step": 409
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9662456800440738e-05,
+      "loss": 0.4099,
+      "step": 410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9660200547914753e-05,
+      "loss": 0.4048,
+      "step": 411
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9657936910032915e-05,
+      "loss": 0.4321,
+      "step": 412
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965566588852581e-05,
+      "loss": 0.4008,
+      "step": 413
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9653387485129665e-05,
+      "loss": 0.4062,
+      "step": 414
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965110170158635e-05,
+      "loss": 0.4156,
+      "step": 415
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9648808539643375e-05,
+      "loss": 0.4046,
+      "step": 416
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.96465080010539e-05,
+      "loss": 0.4095,
+      "step": 417
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964420008757671e-05,
+      "loss": 0.421,
+      "step": 418
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9641884800976244e-05,
+      "loss": 0.3896,
+      "step": 419
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9639562143022567e-05,
+      "loss": 0.3993,
+      "step": 420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9637232115491384e-05,
+      "loss": 0.3907,
+      "step": 421
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9634894720164028e-05,
+      "loss": 0.3982,
+      "step": 422
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9632549958827476e-05,
+      "loss": 0.3999,
+      "step": 423
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963019783327433e-05,
+      "loss": 0.4422,
+      "step": 424
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962783834530282e-05,
+      "loss": 0.3832,
+      "step": 425
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9625471496716808e-05,
+      "loss": 0.4013,
+      "step": 426
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962309728932578e-05,
+      "loss": 0.3709,
+      "step": 427
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9620715724944855e-05,
+      "loss": 0.406,
+      "step": 428
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9618326805394765e-05,
+      "loss": 0.3986,
+      "step": 429
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9615930532501877e-05,
+      "loss": 0.3977,
+      "step": 430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9613526908098174e-05,
+      "loss": 0.3797,
+      "step": 431
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9611115934021256e-05,
+      "loss": 0.4138,
+      "step": 432
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960869761211435e-05,
+      "loss": 0.3998,
+      "step": 433
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9606271944226295e-05,
+      "loss": 0.4021,
+      "step": 434
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9603838932211544e-05,
+      "loss": 0.4147,
+      "step": 435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960139857793017e-05,
+      "loss": 0.3558,
+      "step": 436
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959895088324786e-05,
+      "loss": 0.4063,
+      "step": 437
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9596495850035914e-05,
+      "loss": 0.3945,
+      "step": 438
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9594033480171227e-05,
+      "loss": 0.3981,
+      "step": 439
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9591563775536322e-05,
+      "loss": 0.4146,
+      "step": 440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9589086738019323e-05,
+      "loss": 0.379,
+      "step": 441
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9586602369513958e-05,
+      "loss": 0.3739,
+      "step": 442
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9584110671919553e-05,
+      "loss": 0.421,
+      "step": 443
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9581611647141055e-05,
+      "loss": 0.3927,
+      "step": 444
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9579105297089e-05,
+      "loss": 0.3871,
+      "step": 445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9576591623679528e-05,
+      "loss": 0.3916,
+      "step": 446
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957407062883437e-05,
+      "loss": 0.3842,
+      "step": 447
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9571542314480872e-05,
+      "loss": 0.3757,
+      "step": 448
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9569006682551958e-05,
+      "loss": 0.3903,
+      "step": 449
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9566463734986154e-05,
+      "loss": 0.4201,
+      "step": 450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9563913473727577e-05,
+      "loss": 0.4047,
+      "step": 451
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.956135590072594e-05,
+      "loss": 0.3894,
+      "step": 452
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9558791017936545e-05,
+      "loss": 0.4072,
+      "step": 453
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9556218827320275e-05,
+      "loss": 0.4104,
+      "step": 454
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9553639330843604e-05,
+      "loss": 0.3914,
+      "step": 455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9551052530478597e-05,
+      "loss": 0.4231,
+      "step": 456
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9548458428202896e-05,
+      "loss": 0.4003,
+      "step": 457
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9545857025999728e-05,
+      "loss": 0.3934,
+      "step": 458
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95432483258579e-05,
+      "loss": 0.3815,
+      "step": 459
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95406323297718e-05,
+      "loss": 0.4133,
+      "step": 460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9538009039741393e-05,
+      "loss": 0.402,
+      "step": 461
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9535378457772223e-05,
+      "loss": 0.4123,
+      "step": 462
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95327405858754e-05,
+      "loss": 0.3835,
+      "step": 463
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9530095426067616e-05,
+      "loss": 0.3826,
+      "step": 464
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9527442980371133e-05,
+      "loss": 0.3697,
+      "step": 465
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952478325081378e-05,
+      "loss": 0.3898,
+      "step": 466
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952211623942896e-05,
+      "loss": 0.3621,
+      "step": 467
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9519441948255638e-05,
+      "loss": 0.3872,
+      "step": 468
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9516760379338345e-05,
+      "loss": 0.3684,
+      "step": 469
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951407153472718e-05,
+      "loss": 0.4105,
+      "step": 470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.95113754164778e-05,
+      "loss": 0.4056,
+      "step": 471
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950867202665142e-05,
+      "loss": 0.4395,
+      "step": 472
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9505961367314824e-05,
+      "loss": 0.3943,
+      "step": 473
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950324344054035e-05,
+      "loss": 0.3526,
+      "step": 474
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9500518248405886e-05,
+      "loss": 0.3952,
+      "step": 475
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9497785792994882e-05,
+      "loss": 0.3532,
+      "step": 476
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9495046076396336e-05,
+      "loss": 0.3844,
+      "step": 477
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9492299100704798e-05,
+      "loss": 0.3893,
+      "step": 478
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9489544868020368e-05,
+      "loss": 0.3901,
+      "step": 479
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.94867833804487e-05,
+      "loss": 0.3972,
+      "step": 480
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9484014640100982e-05,
+      "loss": 0.3663,
+      "step": 481
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9481238649093958e-05,
+      "loss": 0.3995,
+      "step": 482
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9478455409549907e-05,
+      "loss": 0.3933,
+      "step": 483
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947566492359666e-05,
+      "loss": 0.4002,
+      "step": 484
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947286719336758e-05,
+      "loss": 0.4059,
+      "step": 485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947006222100157e-05,
+      "loss": 0.3827,
+      "step": 486
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946725000864307e-05,
+      "loss": 0.3762,
+      "step": 487
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9464430558442047e-05,
+      "loss": 0.3948,
+      "step": 488
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9461603872554025e-05,
+      "loss": 0.3995,
+      "step": 489
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9458769953140032e-05,
+      "loss": 0.3824,
+      "step": 490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945592880236664e-05,
+      "loss": 0.3432,
+      "step": 491
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9453080422405953e-05,
+      "loss": 0.3605,
+      "step": 492
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945022481543559e-05,
+      "loss": 0.3963,
+      "step": 493
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9447361983638707e-05,
+      "loss": 0.3772,
+      "step": 494
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9444491929203978e-05,
+      "loss": 0.3855,
+      "step": 495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944161465432559e-05,
+      "loss": 0.4137,
+      "step": 496
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943873016120327e-05,
+      "loss": 0.3952,
+      "step": 497
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943583845204225e-05,
+      "loss": 0.3892,
+      "step": 498
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9432939529053277e-05,
+      "loss": 0.4005,
+      "step": 499
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9430033394452624e-05,
+      "loss": 0.3729,
+      "step": 500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9427120050462066e-05,
+      "loss": 0.389,
+      "step": 501
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9424199499308894e-05,
+      "loss": 0.3704,
+      "step": 502
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9421271743225913e-05,
+      "loss": 0.3774,
+      "step": 503
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9418336784451428e-05,
+      "loss": 0.3776,
+      "step": 504
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941539462522926e-05,
+      "loss": 0.4097,
+      "step": 505
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941244526780873e-05,
+      "loss": 0.3762,
+      "step": 506
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940948871444466e-05,
+      "loss": 0.3773,
+      "step": 507
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940652496739737e-05,
+      "loss": 0.4121,
+      "step": 508
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9403554028932696e-05,
+      "loss": 0.4038,
+      "step": 509
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9400575901321955e-05,
+      "loss": 0.3991,
+      "step": 510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9397590586841968e-05,
+      "loss": 0.4021,
+      "step": 511
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939459808777505e-05,
+      "loss": 0.4056,
+      "step": 512
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9391598406409004e-05,
+      "loss": 0.3804,
+      "step": 513
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938859154503713e-05,
+      "loss": 0.4025,
+      "step": 514
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9385577505958217e-05,
+      "loss": 0.3662,
+      "step": 515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9382556291476535e-05,
+      "loss": 0.3816,
+      "step": 516
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937952790390185e-05,
+      "loss": 0.4161,
+      "step": 517
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9376492345549396e-05,
+      "loss": 0.3846,
+      "step": 518
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9373449618739903e-05,
+      "loss": 0.4213,
+      "step": 519
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9370399725799585e-05,
+      "loss": 0.3962,
+      "step": 520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9367342669060115e-05,
+      "loss": 0.3651,
+      "step": 521
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9364278450858663e-05,
+      "loss": 0.4128,
+      "step": 522
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9361207073537865e-05,
+      "loss": 0.3837,
+      "step": 523
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935812853944583e-05,
+      "loss": 0.3643,
+      "step": 524
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935504285093614e-05,
+      "loss": 0.3743,
+      "step": 525
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9351950010367843e-05,
+      "loss": 0.3712,
+      "step": 526
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9348850020105462e-05,
+      "loss": 0.4042,
+      "step": 527
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934574288251898e-05,
+      "loss": 0.3784,
+      "step": 528
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9342628599983852e-05,
+      "loss": 0.3868,
+      "step": 529
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933950717488098e-05,
+      "loss": 0.3889,
+      "step": 530
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9336378609596737e-05,
+      "loss": 0.3817,
+      "step": 531
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9333242906522963e-05,
+      "loss": 0.3865,
+      "step": 532
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933010006805694e-05,
+      "loss": 0.3958,
+      "step": 533
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932695009660141e-05,
+      "loss": 0.3752,
+      "step": 534
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9323792994564576e-05,
+      "loss": 0.3951,
+      "step": 535
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9320628764360078e-05,
+      "loss": 0.3938,
+      "step": 536
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.931745740840702e-05,
+      "loss": 0.4152,
+      "step": 537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931427892912994e-05,
+      "loss": 0.4189,
+      "step": 538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9311093328958838e-05,
+      "loss": 0.3966,
+      "step": 539
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930790061032914e-05,
+      "loss": 0.3372,
+      "step": 540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9304700775681726e-05,
+      "loss": 0.4261,
+      "step": 541
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9301493827462917e-05,
+      "loss": 0.4144,
+      "step": 542
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9298279768124466e-05,
+      "loss": 0.4068,
+      "step": 543
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9295058600123568e-05,
+      "loss": 0.4116,
+      "step": 544
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9291830325922848e-05,
+      "loss": 0.3977,
+      "step": 545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928859494799036e-05,
+      "loss": 0.3986,
+      "step": 546
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9285352468799606e-05,
+      "loss": 0.3895,
+      "step": 547
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9282102890829502e-05,
+      "loss": 0.3903,
+      "step": 548
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927884621656439e-05,
+      "loss": 0.4133,
+      "step": 549
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9275582448494052e-05,
+      "loss": 0.3759,
+      "step": 550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9272311589113673e-05,
+      "loss": 0.3961,
+      "step": 551
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9269033640923873e-05,
+      "loss": 0.3577,
+      "step": 552
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926574860643069e-05,
+      "loss": 0.399,
+      "step": 553
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926245648814558e-05,
+      "loss": 0.3912,
+      "step": 554
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925915728858541e-05,
+      "loss": 0.3915,
+      "step": 555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925585101027246e-05,
+      "loss": 0.3823,
+      "step": 556
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9252537655734432e-05,
+      "loss": 0.4164,
+      "step": 557
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9249217227504425e-05,
+      "loss": 0.3859,
+      "step": 558
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9245889728120954e-05,
+      "loss": 0.4083,
+      "step": 559
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924255516012794e-05,
+      "loss": 0.367,
+      "step": 560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9239213526074704e-05,
+      "loss": 0.3784,
+      "step": 561
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923586482851597e-05,
+      "loss": 0.3791,
+      "step": 562
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9232509070011868e-05,
+      "loss": 0.373,
+      "step": 563
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9229146253127917e-05,
+      "loss": 0.3759,
+      "step": 564
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922577638043504e-05,
+      "loss": 0.3918,
+      "step": 565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9222399454509543e-05,
+      "loss": 0.4139,
+      "step": 566
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921901547793314e-05,
+      "loss": 0.3768,
+      "step": 567
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9215624453292926e-05,
+      "loss": 0.3829,
+      "step": 568
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9212226383181385e-05,
+      "loss": 0.3728,
+      "step": 569
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9208821270196387e-05,
+      "loss": 0.3676,
+      "step": 570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9205409116941195e-05,
+      "loss": 0.3932,
+      "step": 571
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9201989926024435e-05,
+      "loss": 0.3934,
+      "step": 572
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9198563700060132e-05,
+      "loss": 0.4269,
+      "step": 573
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9195130441667686e-05,
+      "loss": 0.3645,
+      "step": 574
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.919169015347186e-05,
+      "loss": 0.3857,
+      "step": 575
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918824283810281e-05,
+      "loss": 0.3997,
+      "step": 576
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9184788498196055e-05,
+      "loss": 0.3915,
+      "step": 577
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9181327136392484e-05,
+      "loss": 0.3711,
+      "step": 578
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9177858755338358e-05,
+      "loss": 0.3671,
+      "step": 579
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9174383357685294e-05,
+      "loss": 0.3785,
+      "step": 580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9170900946090292e-05,
+      "loss": 0.3886,
+      "step": 581
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9167411523215693e-05,
+      "loss": 0.3846,
+      "step": 582
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916391509172922e-05,
+      "loss": 0.3837,
+      "step": 583
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916041165430393e-05,
+      "loss": 0.4211,
+      "step": 584
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915690121361826e-05,
+      "loss": 0.3732,
+      "step": 585
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9153383772355984e-05,
+      "loss": 0.385,
+      "step": 586
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9149859333206233e-05,
+      "loss": 0.4041,
+      "step": 587
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9146327898863493e-05,
+      "loss": 0.4041,
+      "step": 588
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9142789472027594e-05,
+      "loss": 0.3946,
+      "step": 589
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9139244055403712e-05,
+      "loss": 0.3757,
+      "step": 590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9135691651702365e-05,
+      "loss": 0.3843,
+      "step": 591
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9132132263639415e-05,
+      "loss": 0.3953,
+      "step": 592
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9128565893936058e-05,
+      "loss": 0.3738,
+      "step": 593
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912499254531884e-05,
+      "loss": 0.4154,
+      "step": 594
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9121412220519633e-05,
+      "loss": 0.3807,
+      "step": 595
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9117824922275644e-05,
+      "loss": 0.363,
+      "step": 596
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911423065332941e-05,
+      "loss": 0.378,
+      "step": 597
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9110629416428802e-05,
+      "loss": 0.4233,
+      "step": 598
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9107021214327013e-05,
+      "loss": 0.3947,
+      "step": 599
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9103406049782563e-05,
+      "loss": 0.3626,
+      "step": 600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9099783925559295e-05,
+      "loss": 0.3739,
+      "step": 601
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9096154844426372e-05,
+      "loss": 0.3929,
+      "step": 602
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909251880915828e-05,
+      "loss": 0.3624,
+      "step": 603
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908887582253481e-05,
+      "loss": 0.3602,
+      "step": 604
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9085225887341087e-05,
+      "loss": 0.3907,
+      "step": 605
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9081569006367525e-05,
+      "loss": 0.3796,
+      "step": 606
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9077905182409866e-05,
+      "loss": 0.4068,
+      "step": 607
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9074234418269153e-05,
+      "loss": 0.3939,
+      "step": 608
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9070556716751734e-05,
+      "loss": 0.3651,
+      "step": 609
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9066872080669262e-05,
+      "loss": 0.3979,
+      "step": 610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9063180512838696e-05,
+      "loss": 0.384,
+      "step": 611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9059482016082286e-05,
+      "loss": 0.3894,
+      "step": 612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9055776593227584e-05,
+      "loss": 0.3773,
+      "step": 613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9052064247107437e-05,
+      "loss": 0.4081,
+      "step": 614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9048344980559986e-05,
+      "loss": 0.3459,
+      "step": 615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9044618796428658e-05,
+      "loss": 0.3922,
+      "step": 616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9040885697562174e-05,
+      "loss": 0.3932,
+      "step": 617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903714568681454e-05,
+      "loss": 0.3882,
+      "step": 618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9033398767045045e-05,
+      "loss": 0.3813,
+      "step": 619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9029644941118262e-05,
+      "loss": 0.3924,
+      "step": 620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9025884211904038e-05,
+      "loss": 0.3642,
+      "step": 621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9022116582277505e-05,
+      "loss": 0.3772,
+      "step": 622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901834205511907e-05,
+      "loss": 0.4028,
+      "step": 623
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.901456063331441e-05,
+      "loss": 0.3893,
+      "step": 624
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9010772319754464e-05,
+      "loss": 0.3968,
+      "step": 625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9006977117335466e-05,
+      "loss": 0.4008,
+      "step": 626
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9003175028958887e-05,
+      "loss": 0.3754,
+      "step": 627
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899936605753148e-05,
+      "loss": 0.3874,
+      "step": 628
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899555020596526e-05,
+      "loss": 0.4049,
+      "step": 629
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8991727477177492e-05,
+      "loss": 0.3805,
+      "step": 630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8987897874090713e-05,
+      "loss": 0.3681,
+      "step": 631
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8984061399632696e-05,
+      "loss": 0.4204,
+      "step": 632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8980218056736484e-05,
+      "loss": 0.3866,
+      "step": 633
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8976367848340366e-05,
+      "loss": 0.3726,
+      "step": 634
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8972510777387878e-05,
+      "loss": 0.3769,
+      "step": 635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8968646846827802e-05,
+      "loss": 0.3878,
+      "step": 636
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896477605961417e-05,
+      "loss": 0.3846,
+      "step": 637
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896089841870625e-05,
+      "loss": 0.4283,
+      "step": 638
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895701392706855e-05,
+      "loss": 0.3734,
+      "step": 639
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895312258767082e-05,
+      "loss": 0.3829,
+      "step": 640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8949224403488038e-05,
+      "loss": 0.4039,
+      "step": 641
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8945319377500425e-05,
+      "loss": 0.3636,
+      "step": 642
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894140751269342e-05,
+      "loss": 0.3752,
+      "step": 643
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8937488812057703e-05,
+      "loss": 0.4128,
+      "step": 644
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8933563278589177e-05,
+      "loss": 0.3868,
+      "step": 645
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8929630915288957e-05,
+      "loss": 0.3827,
+      "step": 646
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8925691725163395e-05,
+      "loss": 0.3577,
+      "step": 647
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8921745711224057e-05,
+      "loss": 0.3747,
+      "step": 648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891779287648772e-05,
+      "loss": 0.3869,
+      "step": 649
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8913833223976384e-05,
+      "loss": 0.3962,
+      "step": 650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8909866756717257e-05,
+      "loss": 0.3658,
+      "step": 651
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8905893477742753e-05,
+      "loss": 0.3783,
+      "step": 652
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8901913390090505e-05,
+      "loss": 0.3983,
+      "step": 653
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8897926496803345e-05,
+      "loss": 0.3994,
+      "step": 654
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.88939328009293e-05,
+      "loss": 0.3959,
+      "step": 655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8889932305521614e-05,
+      "loss": 0.3664,
+      "step": 656
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888592501363871e-05,
+      "loss": 0.3883,
+      "step": 657
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8881910928344228e-05,
+      "loss": 0.3656,
+      "step": 658
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8877890052706988e-05,
+      "loss": 0.3951,
+      "step": 659
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8873862389801003e-05,
+      "loss": 0.3888,
+      "step": 660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886982794270548e-05,
+      "loss": 0.3714,
+      "step": 661
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8865786714504803e-05,
+      "loss": 0.4063,
+      "step": 662
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886173870828855e-05,
+      "loss": 0.3772,
+      "step": 663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885768392715148e-05,
+      "loss": 0.3616,
+      "step": 664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8853622374193527e-05,
+      "loss": 0.3592,
+      "step": 665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8849554052519798e-05,
+      "loss": 0.4094,
+      "step": 666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884547896524059e-05,
+      "loss": 0.3486,
+      "step": 667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8841397115471357e-05,
+      "loss": 0.3895,
+      "step": 668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8837308506332733e-05,
+      "loss": 0.3516,
+      "step": 669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883321314095051e-05,
+      "loss": 0.3953,
+      "step": 670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882911102245566e-05,
+      "loss": 0.3759,
+      "step": 671
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.88250021539843e-05,
+      "loss": 0.4004,
+      "step": 672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8820886538677724e-05,
+      "loss": 0.3855,
+      "step": 673
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881676417968237e-05,
+      "loss": 0.3915,
+      "step": 674
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881263508014984e-05,
+      "loss": 0.4133,
+      "step": 675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880849924323689e-05,
+      "loss": 0.3674,
+      "step": 676
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880435667210542e-05,
+      "loss": 0.3536,
+      "step": 677
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8800207369922486e-05,
+      "loss": 0.3816,
+      "step": 678
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879605133986028e-05,
+      "loss": 0.3775,
+      "step": 679
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8791888585096153e-05,
+      "loss": 0.4081,
+      "step": 680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8787719108812585e-05,
+      "loss": 0.3734,
+      "step": 681
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8783542914197194e-05,
+      "loss": 0.3582,
+      "step": 682
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877936000444274e-05,
+      "loss": 0.3635,
+      "step": 683
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8775170382747115e-05,
+      "loss": 0.3591,
+      "step": 684
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8770974052313347e-05,
+      "loss": 0.3751,
+      "step": 685
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8766771016349584e-05,
+      "loss": 0.382,
+      "step": 686
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87625612780691e-05,
+      "loss": 0.3444,
+      "step": 687
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8758344840690307e-05,
+      "loss": 0.3671,
+      "step": 688
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875412170743672e-05,
+      "loss": 0.388,
+      "step": 689
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874989188153699e-05,
+      "loss": 0.384,
+      "step": 690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8745655366224872e-05,
+      "loss": 0.4147,
+      "step": 691
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8741412164739244e-05,
+      "loss": 0.3946,
+      "step": 692
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873716228032409e-05,
+      "loss": 0.4002,
+      "step": 693
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87329057162285e-05,
+      "loss": 0.4323,
+      "step": 694
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8728642475706687e-05,
+      "loss": 0.3671,
+      "step": 695
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8724372562017948e-05,
+      "loss": 0.3841,
+      "step": 696
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872009597842669e-05,
+      "loss": 0.3668,
+      "step": 697
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8715812728202425e-05,
+      "loss": 0.4067,
+      "step": 698
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8711522814619753e-05,
+      "loss": 0.4001,
+      "step": 699
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8707226240958373e-05,
+      "loss": 0.388,
+      "step": 700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8702923010503076e-05,
+      "loss": 0.3907,
+      "step": 701
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8698613126543737e-05,
+      "loss": 0.429,
+      "step": 702
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8694296592375325e-05,
+      "loss": 0.3614,
+      "step": 703
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868997341129789e-05,
+      "loss": 0.3703,
+      "step": 704
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8685643586616558e-05,
+      "loss": 0.3963,
+      "step": 705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8681307121641544e-05,
+      "loss": 0.3818,
+      "step": 706
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867696401968813e-05,
+      "loss": 0.4132,
+      "step": 707
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8672614284076686e-05,
+      "loss": 0.3913,
+      "step": 708
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866825791813264e-05,
+      "loss": 0.3642,
+      "step": 709
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866389492518649e-05,
+      "loss": 0.4126,
+      "step": 710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8659525308573806e-05,
+      "loss": 0.3717,
+      "step": 711
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8655149071635224e-05,
+      "loss": 0.3929,
+      "step": 712
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865076621771643e-05,
+      "loss": 0.3598,
+      "step": 713
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8646376750168183e-05,
+      "loss": 0.3874,
+      "step": 714
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8641980672346284e-05,
+      "loss": 0.3588,
+      "step": 715
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8637577987611603e-05,
+      "loss": 0.3836,
+      "step": 716
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8633168699330048e-05,
+      "loss": 0.3688,
+      "step": 717
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8628752810872578e-05,
+      "loss": 0.3814,
+      "step": 718
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8624330325615205e-05,
+      "loss": 0.372,
+      "step": 719
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861990124693898e-05,
+      "loss": 0.42,
+      "step": 720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861546557822999e-05,
+      "loss": 0.3489,
+      "step": 721
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8611023322879366e-05,
+      "loss": 0.3692,
+      "step": 722
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8606574484283276e-05,
+      "loss": 0.4177,
+      "step": 723
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8602119065842918e-05,
+      "loss": 0.4102,
+      "step": 724
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8597657070964517e-05,
+      "loss": 0.3674,
+      "step": 725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8593188503059333e-05,
+      "loss": 0.3632,
+      "step": 726
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8588713365543644e-05,
+      "loss": 0.3712,
+      "step": 727
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8584231661838753e-05,
+      "loss": 0.4181,
+      "step": 728
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857974339537099e-05,
+      "loss": 0.3991,
+      "step": 729
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8575248569571692e-05,
+      "loss": 0.3684,
+      "step": 730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8570747187877214e-05,
+      "loss": 0.3784,
+      "step": 731
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8566239253728928e-05,
+      "loss": 0.3875,
+      "step": 732
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856172477057321e-05,
+      "loss": 0.3889,
+      "step": 733
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8557203741861444e-05,
+      "loss": 0.3601,
+      "step": 734
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8552676171050013e-05,
+      "loss": 0.3905,
+      "step": 735
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854814206160032e-05,
+      "loss": 0.3935,
+      "step": 736
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854360141697874e-05,
+      "loss": 0.4062,
+      "step": 737
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8539054240656667e-05,
+      "loss": 0.3724,
+      "step": 738
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8534500536110476e-05,
+      "loss": 0.3845,
+      "step": 739
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8529940306821537e-05,
+      "loss": 0.3759,
+      "step": 740
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852537355627621e-05,
+      "loss": 0.3899,
+      "step": 741
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8520800287965836e-05,
+      "loss": 0.3562,
+      "step": 742
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851622050538674e-05,
+      "loss": 0.3725,
+      "step": 743
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8511634212040235e-05,
+      "loss": 0.3816,
+      "step": 744
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85070414114326e-05,
+      "loss": 0.3481,
+      "step": 745
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850244210707509e-05,
+      "loss": 0.3938,
+      "step": 746
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8497836302483945e-05,
+      "loss": 0.3871,
+      "step": 747
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849322400118036e-05,
+      "loss": 0.4019,
+      "step": 748
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8488605206690508e-05,
+      "loss": 0.3976,
+      "step": 749
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8483979922545515e-05,
+      "loss": 0.3996,
+      "step": 750
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847934815228148e-05,
+      "loss": 0.3353,
+      "step": 751
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847470989943945e-05,
+      "loss": 0.3579,
+      "step": 752
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8470065167565433e-05,
+      "loss": 0.3867,
+      "step": 753
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8465413960210393e-05,
+      "loss": 0.357,
+      "step": 754
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8460756280930238e-05,
+      "loss": 0.3736,
+      "step": 755
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8456092133285834e-05,
+      "loss": 0.3548,
+      "step": 756
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845142152084298e-05,
+      "loss": 0.3717,
+      "step": 757
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8446744447172428e-05,
+      "loss": 0.3906,
+      "step": 758
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8442060915849856e-05,
+      "loss": 0.39,
+      "step": 759
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.84373709304559e-05,
+      "loss": 0.3856,
+      "step": 760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8432674494576113e-05,
+      "loss": 0.3874,
+      "step": 761
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8427971611800982e-05,
+      "loss": 0.3781,
+      "step": 762
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842326228572593e-05,
+      "loss": 0.3421,
+      "step": 763
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8418546519951296e-05,
+      "loss": 0.4018,
+      "step": 764
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841382431808235e-05,
+      "loss": 0.4096,
+      "step": 765
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8409095683729284e-05,
+      "loss": 0.3774,
+      "step": 766
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.84043606205072e-05,
+      "loss": 0.3954,
+      "step": 767
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8399619132036117e-05,
+      "loss": 0.3601,
+      "step": 768
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8394871221940972e-05,
+      "loss": 0.3991,
+      "step": 769
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8390116893851605e-05,
+      "loss": 0.4195,
+      "step": 770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838535615140277e-05,
+      "loss": 0.3833,
+      "step": 771
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8380588998234107e-05,
+      "loss": 0.3674,
+      "step": 772
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8375815437990185e-05,
+      "loss": 0.4022,
+      "step": 773
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8371035474320453e-05,
+      "loss": 0.402,
+      "step": 774
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836624911087925e-05,
+      "loss": 0.3625,
+      "step": 775
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8361456351325824e-05,
+      "loss": 0.4088,
+      "step": 776
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8356657199324304e-05,
+      "loss": 0.3825,
+      "step": 777
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8351851658543707e-05,
+      "loss": 0.4008,
+      "step": 778
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8347039732657935e-05,
+      "loss": 0.3997,
+      "step": 779
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8342221425345773e-05,
+      "loss": 0.3739,
+      "step": 780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833739674029088e-05,
+      "loss": 0.3879,
+      "step": 781
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8332565681181794e-05,
+      "loss": 0.4081,
+      "step": 782
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832772825171193e-05,
+      "loss": 0.4104,
+      "step": 783
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8322884455579567e-05,
+      "loss": 0.3869,
+      "step": 784
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8318034296487853e-05,
+      "loss": 0.3334,
+      "step": 785
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.83131777781448e-05,
+      "loss": 0.3783,
+      "step": 786
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8308314904263288e-05,
+      "loss": 0.3853,
+      "step": 787
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8303445678561043e-05,
+      "loss": 0.3884,
+      "step": 788
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829857010476066e-05,
+      "loss": 0.3459,
+      "step": 789
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8293688186589582e-05,
+      "loss": 0.3739,
+      "step": 790
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.82887999277801e-05,
+      "loss": 0.3401,
+      "step": 791
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8283905332069357e-05,
+      "loss": 0.3884,
+      "step": 792
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827900440319934e-05,
+      "loss": 0.3425,
+      "step": 793
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827409714491687e-05,
+      "loss": 0.3561,
+      "step": 794
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8269183560973622e-05,
+      "loss": 0.3746,
+      "step": 795
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8264263655126092e-05,
+      "loss": 0.3895,
+      "step": 796
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8259337431135617e-05,
+      "loss": 0.3437,
+      "step": 797
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8254404892768366e-05,
+      "loss": 0.3775,
+      "step": 798
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8249466043795326e-05,
+      "loss": 0.3844,
+      "step": 799
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8244520887992322e-05,
+      "loss": 0.4206,
+      "step": 800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8239569429139986e-05,
+      "loss": 0.3954,
+      "step": 801
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823461167102378e-05,
+      "loss": 0.3875,
+      "step": 802
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8229647617433982e-05,
+      "loss": 0.3763,
+      "step": 803
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822467727216567e-05,
+      "loss": 0.4144,
+      "step": 804
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821970063901875e-05,
+      "loss": 0.3874,
+      "step": 805
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821471772179792e-05,
+      "loss": 0.3558,
+      "step": 806
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8209728524312696e-05,
+      "loss": 0.3897,
+      "step": 807
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820473305037738e-05,
+      "loss": 0.3937,
+      "step": 808
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8199731303811083e-05,
+      "loss": 0.3991,
+      "step": 809
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.819472328843771e-05,
+      "loss": 0.4124,
+      "step": 810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8189709008085957e-05,
+      "loss": 0.3457,
+      "step": 811
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8184688466589308e-05,
+      "loss": 0.3619,
+      "step": 812
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8179661667786042e-05,
+      "loss": 0.3932,
+      "step": 813
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817462861551921e-05,
+      "loss": 0.3765,
+      "step": 814
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816958931363665e-05,
+      "loss": 0.3728,
+      "step": 815
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8164543765990976e-05,
+      "loss": 0.3616,
+      "step": 816
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8159491976439585e-05,
+      "loss": 0.3627,
+      "step": 817
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815443394884463e-05,
+      "loss": 0.3899,
+      "step": 818
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8149369687073053e-05,
+      "loss": 0.3836,
+      "step": 819
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8144299194996545e-05,
+      "loss": 0.4012,
+      "step": 820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.813922247649157e-05,
+      "loss": 0.3745,
+      "step": 821
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.813413953543934e-05,
+      "loss": 0.4013,
+      "step": 822
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8129050375725844e-05,
+      "loss": 0.3731,
+      "step": 823
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8123955001241813e-05,
+      "loss": 0.405,
+      "step": 824
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8118853415882726e-05,
+      "loss": 0.3733,
+      "step": 825
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811374562354882e-05,
+      "loss": 0.3707,
+      "step": 826
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8108631628145068e-05,
+      "loss": 0.3546,
+      "step": 827
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810351143358119e-05,
+      "loss": 0.4078,
+      "step": 828
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8098385043771646e-05,
+      "loss": 0.3922,
+      "step": 829
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8093252462635634e-05,
+      "loss": 0.3623,
+      "step": 830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808811369409708e-05,
+      "loss": 0.3714,
+      "step": 831
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808296874208464e-05,
+      "loss": 0.3898,
+      "step": 832
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8077817610531704e-05,
+      "loss": 0.4124,
+      "step": 833
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8072660303376384e-05,
+      "loss": 0.383,
+      "step": 834
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8067496824561514e-05,
+      "loss": 0.3994,
+      "step": 835
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8062327178034635e-05,
+      "loss": 0.3507,
+      "step": 836
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8057151367748024e-05,
+      "loss": 0.3532,
+      "step": 837
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805196939765865e-05,
+      "loss": 0.3683,
+      "step": 838
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8046781271728203e-05,
+      "loss": 0.3954,
+      "step": 839
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804158699392308e-05,
+      "loss": 0.4022,
+      "step": 840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8036386568214374e-05,
+      "loss": 0.375,
+      "step": 841
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8031179998577882e-05,
+      "loss": 0.3666,
+      "step": 842
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80259672889941e-05,
+      "loss": 0.363,
+      "step": 843
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8020748443448216e-05,
+      "loss": 0.4166,
+      "step": 844
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8015523465930104e-05,
+      "loss": 0.4173,
+      "step": 845
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8010292360434336e-05,
+      "loss": 0.4118,
+      "step": 846
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8005055130960162e-05,
+      "loss": 0.3671,
+      "step": 847
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7999811781511517e-05,
+      "loss": 0.3659,
+      "step": 848
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7994562316097014e-05,
+      "loss": 0.4199,
+      "step": 849
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798930673872994e-05,
+      "loss": 0.4238,
+      "step": 850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7984045053428256e-05,
+      "loss": 0.3838,
+      "step": 851
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.797877726421459e-05,
+      "loss": 0.389,
+      "step": 852
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7973503375116243e-05,
+      "loss": 0.3652,
+      "step": 853
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796822339016518e-05,
+      "loss": 0.3909,
+      "step": 854
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796293731339801e-05,
+      "loss": 0.406,
+      "step": 855
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795764514885602e-05,
+      "loss": 0.3615,
+      "step": 856
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7952346900585142e-05,
+      "loss": 0.3644,
+      "step": 857
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7947042572635954e-05,
+      "loss": 0.3792,
+      "step": 858
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7941732169063692e-05,
+      "loss": 0.4111,
+      "step": 859
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7936415693928233e-05,
+      "loss": 0.3439,
+      "step": 860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7931093151294094e-05,
+      "loss": 0.3656,
+      "step": 861
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7925764545230428e-05,
+      "loss": 0.3596,
+      "step": 862
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7920429879811037e-05,
+      "loss": 0.3674,
+      "step": 863
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7915089159114333e-05,
+      "loss": 0.3941,
+      "step": 864
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7909742387223382e-05,
+      "loss": 0.3677,
+      "step": 865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7904389568225857e-05,
+      "loss": 0.41,
+      "step": 866
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7899030706214067e-05,
+      "loss": 0.4099,
+      "step": 867
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789366580528493e-05,
+      "loss": 0.3559,
+      "step": 868
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788829486953999e-05,
+      "loss": 0.393,
+      "step": 869
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.78829179030854e-05,
+      "loss": 0.3947,
+      "step": 870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7877534910031923e-05,
+      "loss": 0.381,
+      "step": 871
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7872145894494935e-05,
+      "loss": 0.3635,
+      "step": 872
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7866750860594412e-05,
+      "loss": 0.3597,
+      "step": 873
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7861349812454926e-05,
+      "loss": 0.3655,
+      "step": 874
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785594275420566e-05,
+      "loss": 0.3331,
+      "step": 875
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7850529689980375e-05,
+      "loss": 0.4022,
+      "step": 876
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7845110623917436e-05,
+      "loss": 0.3664,
+      "step": 877
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7839685560159802e-05,
+      "loss": 0.3513,
+      "step": 878
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7834254502855e-05,
+      "loss": 0.4183,
+      "step": 879
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7828817456155144e-05,
+      "loss": 0.3696,
+      "step": 880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7823374424216944e-05,
+      "loss": 0.3838,
+      "step": 881
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7817925411201664e-05,
+      "loss": 0.3838,
+      "step": 882
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7812470421275147e-05,
+      "loss": 0.3724,
+      "step": 883
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7807009458607814e-05,
+      "loss": 0.3779,
+      "step": 884
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7801542527374643e-05,
+      "loss": 0.3504,
+      "step": 885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7796069631755176e-05,
+      "loss": 0.3778,
+      "step": 886
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779059077593352e-05,
+      "loss": 0.3671,
+      "step": 887
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7785105964098336e-05,
+      "loss": 0.3778,
+      "step": 888
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777961520044283e-05,
+      "loss": 0.4103,
+      "step": 889
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7774118489164775e-05,
+      "loss": 0.3618,
+      "step": 890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7768615834466484e-05,
+      "loss": 0.401,
+      "step": 891
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77631072405548e-05,
+      "loss": 0.3751,
+      "step": 892
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775759271164113e-05,
+      "loss": 0.4115,
+      "step": 893
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7752072251941404e-05,
+      "loss": 0.3707,
+      "step": 894
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7746545865676085e-05,
+      "loss": 0.42,
+      "step": 895
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7741013557070175e-05,
+      "loss": 0.417,
+      "step": 896
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7735475330353198e-05,
+      "loss": 0.3436,
+      "step": 897
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7729931189759208e-05,
+      "loss": 0.3867,
+      "step": 898
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7724381139526774e-05,
+      "loss": 0.3859,
+      "step": 899
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7718825183898983e-05,
+      "loss": 0.393,
+      "step": 900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771326332712344e-05,
+      "loss": 0.4093,
+      "step": 901
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770769557345227e-05,
+      "loss": 0.3822,
+      "step": 902
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7702121927142083e-05,
+      "loss": 0.3856,
+      "step": 903
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7696542392454018e-05,
+      "loss": 0.3852,
+      "step": 904
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.76909569736537e-05,
+      "loss": 0.4009,
+      "step": 905
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7685365675011265e-05,
+      "loss": 0.3801,
+      "step": 906
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.767976850080133e-05,
+      "loss": 0.367,
+      "step": 907
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7674165455303013e-05,
+      "loss": 0.3534,
+      "step": 908
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7668556542799928e-05,
+      "loss": 0.391,
+      "step": 909
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766294176758015e-05,
+      "loss": 0.3784,
+      "step": 910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7657321133936263e-05,
+      "loss": 0.3514,
+      "step": 911
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7651694646165317e-05,
+      "loss": 0.3836,
+      "step": 912
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764606230856884e-05,
+      "loss": 0.3929,
+      "step": 913
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7640424125452827e-05,
+      "loss": 0.3852,
+      "step": 914
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763478010112775e-05,
+      "loss": 0.3806,
+      "step": 915
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.762913023990854e-05,
+      "loss": 0.3934,
+      "step": 916
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7623474546114593e-05,
+      "loss": 0.3609,
+      "step": 917
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761781302406977e-05,
+      "loss": 0.3899,
+      "step": 918
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7612145678102377e-05,
+      "loss": 0.3777,
+      "step": 919
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7606472512545178e-05,
+      "loss": 0.3825,
+      "step": 920
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7600793531735392e-05,
+      "loss": 0.3438,
+      "step": 921
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759510874001467e-05,
+      "loss": 0.3914,
+      "step": 922
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7589418141729114e-05,
+      "loss": 0.3826,
+      "step": 923
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7583721741229272e-05,
+      "loss": 0.3692,
+      "step": 924
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7578019542870113e-05,
+      "loss": 0.3877,
+      "step": 925
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7572311551011047e-05,
+      "loss": 0.3789,
+      "step": 926
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7566597770015915e-05,
+      "loss": 0.4146,
+      "step": 927
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756087820425298e-05,
+      "loss": 0.3772,
+      "step": 928
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7555152858094923e-05,
+      "loss": 0.3907,
+      "step": 929
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7549421735918856e-05,
+      "loss": 0.3623,
+      "step": 930
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7543684842106304e-05,
+      "loss": 0.3807,
+      "step": 931
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753794218104319e-05,
+      "loss": 0.3733,
+      "step": 932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7532193757119867e-05,
+      "loss": 0.3641,
+      "step": 933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7526439574731074e-05,
+      "loss": 0.386,
+      "step": 934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7520679638275974e-05,
+      "loss": 0.3874,
+      "step": 935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751491395215811e-05,
+      "loss": 0.3703,
+      "step": 936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750914252078543e-05,
+      "loss": 0.3527,
+      "step": 937
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7503365348570273e-05,
+      "loss": 0.3826,
+      "step": 938
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7497582439929367e-05,
+      "loss": 0.3726,
+      "step": 939
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.749179379928382e-05,
+      "loss": 0.3673,
+      "step": 940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7485999431059135e-05,
+      "loss": 0.3954,
+      "step": 941
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748019933968518e-05,
+      "loss": 0.3509,
+      "step": 942
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7474393529596208e-05,
+      "loss": 0.3456,
+      "step": 943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7468582005230837e-05,
+      "loss": 0.3649,
+      "step": 944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.746276477103206e-05,
+      "loss": 0.3884,
+      "step": 945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745694183144723e-05,
+      "loss": 0.395,
+      "step": 946
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7451113190928065e-05,
+      "loss": 0.3925,
+      "step": 947
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7445278853930645e-05,
+      "loss": 0.3581,
+      "step": 948
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7439438824915394e-05,
+      "loss": 0.3864,
+      "step": 949
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7433593108347097e-05,
+      "loss": 0.3928,
+      "step": 950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7427741708694887e-05,
+      "loss": 0.3695,
+      "step": 951
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7421884630432236e-05,
+      "loss": 0.3885,
+      "step": 952
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741602187803696e-05,
+      "loss": 0.4005,
+      "step": 953
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7410153455991218e-05,
+      "loss": 0.3631,
+      "step": 954
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7404279368781494e-05,
+      "loss": 0.3966,
+      "step": 955
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7398399620898613e-05,
+      "loss": 0.3975,
+      "step": 956
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.739251421683772e-05,
+      "loss": 0.3554,
+      "step": 957
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7386623161098287e-05,
+      "loss": 0.3669,
+      "step": 958
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7380726458184108e-05,
+      "loss": 0.4055,
+      "step": 959
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7374824112603294e-05,
+      "loss": 0.3465,
+      "step": 960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7368916128868266e-05,
+      "loss": 0.378,
+      "step": 961
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7363002511495765e-05,
+      "loss": 0.369,
+      "step": 962
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7357083265006825e-05,
+      "loss": 0.3907,
+      "step": 963
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7351158393926793e-05,
+      "loss": 0.3715,
+      "step": 964
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7345227902785318e-05,
+      "loss": 0.3704,
+      "step": 965
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.733929179611634e-05,
+      "loss": 0.3474,
+      "step": 966
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7333350078458096e-05,
+      "loss": 0.38,
+      "step": 967
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7327402754353108e-05,
+      "loss": 0.364,
+      "step": 968
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7321449828348184e-05,
+      "loss": 0.3482,
+      "step": 969
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7315491304994422e-05,
+      "loss": 0.3619,
+      "step": 970
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7309527188847193e-05,
+      "loss": 0.3839,
+      "step": 971
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7303557484466146e-05,
+      "loss": 0.3756,
+      "step": 972
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7297582196415203e-05,
+      "loss": 0.3452,
+      "step": 973
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7291601329262552e-05,
+      "loss": 0.395,
+      "step": 974
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7285614887580653e-05,
+      "loss": 0.4041,
+      "step": 975
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7279622875946214e-05,
+      "loss": 0.3628,
+      "step": 976
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7273625298940216e-05,
+      "loss": 0.3975,
+      "step": 977
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7267622161147887e-05,
+      "loss": 0.411,
+      "step": 978
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726161346715871e-05,
+      "loss": 0.3803,
+      "step": 979
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7255599221566416e-05,
+      "loss": 0.3769,
+      "step": 980
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724957942896897e-05,
+      "loss": 0.4096,
+      "step": 981
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7243554093968595e-05,
+      "loss": 0.4003,
+      "step": 982
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7237523221171733e-05,
+      "loss": 0.4229,
+      "step": 983
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7231486815189077e-05,
+      "loss": 0.3814,
+      "step": 984
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7225444880635544e-05,
+      "loss": 0.3793,
+      "step": 985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7219397422130265e-05,
+      "loss": 0.3709,
+      "step": 986
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7213344444296615e-05,
+      "loss": 0.3437,
+      "step": 987
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7207285951762177e-05,
+      "loss": 0.3877,
+      "step": 988
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7201221949158745e-05,
+      "loss": 0.3813,
+      "step": 989
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7195152441122338e-05,
+      "loss": 0.3697,
+      "step": 990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7189077432293176e-05,
+      "loss": 0.3462,
+      "step": 991
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7182996927315693e-05,
+      "loss": 0.362,
+      "step": 992
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7176910930838508e-05,
+      "loss": 0.4052,
+      "step": 993
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7170819447514456e-05,
+      "loss": 0.3817,
+      "step": 994
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7164722482000557e-05,
+      "loss": 0.3697,
+      "step": 995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7158620038958023e-05,
+      "loss": 0.3606,
+      "step": 996
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7152512123052262e-05,
+      "loss": 0.3682,
+      "step": 997
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714639873895285e-05,
+      "loss": 0.3576,
+      "step": 998
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7140279891333565e-05,
+      "loss": 0.3742,
+      "step": 999
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713415558487234e-05,
+      "loss": 0.3949,
+      "step": 1000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7128025824251296e-05,
+      "loss": 0.3793,
+      "step": 1001
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7121890614156714e-05,
+      "loss": 0.3807,
+      "step": 1002
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711574995927905e-05,
+      "loss": 0.3666,
+      "step": 1003
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710960386431292e-05,
+      "loss": 0.3673,
+      "step": 1004
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7103452333957097e-05,
+      "loss": 0.3547,
+      "step": 1005
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7097295372914515e-05,
+      "loss": 0.3937,
+      "step": 1006
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.709113298589224e-05,
+      "loss": 0.3684,
+      "step": 1007
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.708496517760152e-05,
+      "loss": 0.3589,
+      "step": 1008
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7078791952757718e-05,
+      "loss": 0.3609,
+      "step": 1009
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707261331608035e-05,
+      "loss": 0.3586,
+      "step": 1010
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7066429272293072e-05,
+      "loss": 0.3479,
+      "step": 1011
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7060239826123668e-05,
+      "loss": 0.3667,
+      "step": 1012
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7054044982304058e-05,
+      "loss": 0.3921,
+      "step": 1013
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7047844745570276e-05,
+      "loss": 0.3878,
+      "step": 1014
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.70416391206625e-05,
+      "loss": 0.3787,
+      "step": 1015
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7035428112325008e-05,
+      "loss": 0.3785,
+      "step": 1016
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7029211725306205e-05,
+      "loss": 0.3697,
+      "step": 1017
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7022989964358597e-05,
+      "loss": 0.3518,
+      "step": 1018
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7016762834238815e-05,
+      "loss": 0.4002,
+      "step": 1019
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.701053033970758e-05,
+      "loss": 0.337,
+      "step": 1020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7004292485529718e-05,
+      "loss": 0.3914,
+      "step": 1021
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699804927647416e-05,
+      "loss": 0.3746,
+      "step": 1022
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6991800717313913e-05,
+      "loss": 0.3819,
+      "step": 1023
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6985546812826102e-05,
+      "loss": 0.3684,
+      "step": 1024
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6979287567791917e-05,
+      "loss": 0.3435,
+      "step": 1025
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6973022986996633e-05,
+      "loss": 0.4142,
+      "step": 1026
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696675307522961e-05,
+      "loss": 0.37,
+      "step": 1027
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696047783728428e-05,
+      "loss": 0.3665,
+      "step": 1028
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6954197277958154e-05,
+      "loss": 0.3662,
+      "step": 1029
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.69479114020528e-05,
+      "loss": 0.3411,
+      "step": 1030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6941620214373856e-05,
+      "loss": 0.4115,
+      "step": 1031
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.693532371973103e-05,
+      "loss": 0.3372,
+      "step": 1032
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6929021922938064e-05,
+      "loss": 0.3767,
+      "step": 1033
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6922714828812785e-05,
+      "loss": 0.3576,
+      "step": 1034
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691640244217704e-05,
+      "loss": 0.3582,
+      "step": 1035
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6910084767856738e-05,
+      "loss": 0.3654,
+      "step": 1036
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6903761810681833e-05,
+      "loss": 0.3959,
+      "step": 1037
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6897433575486302e-05,
+      "loss": 0.3438,
+      "step": 1038
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6891100067108182e-05,
+      "loss": 0.3361,
+      "step": 1039
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6884761290389516e-05,
+      "loss": 0.3759,
+      "step": 1040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6878417250176393e-05,
+      "loss": 0.381,
+      "step": 1041
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687206795131891e-05,
+      "loss": 0.3822,
+      "step": 1042
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6865713398671197e-05,
+      "loss": 0.3854,
+      "step": 1043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.68593535970914e-05,
+      "loss": 0.3682,
+      "step": 1044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6852988551441674e-05,
+      "loss": 0.3776,
+      "step": 1045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6846618266588182e-05,
+      "loss": 0.3391,
+      "step": 1046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684024274740109e-05,
+      "loss": 0.3458,
+      "step": 1047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6833861998754578e-05,
+      "loss": 0.3903,
+      "step": 1048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6827476025526807e-05,
+      "loss": 0.3513,
+      "step": 1049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6821084832599953e-05,
+      "loss": 0.3838,
+      "step": 1050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681468842486016e-05,
+      "loss": 0.3688,
+      "step": 1051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6808286807197572e-05,
+      "loss": 0.3883,
+      "step": 1052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680187998450632e-05,
+      "loss": 0.353,
+      "step": 1053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6795467961684496e-05,
+      "loss": 0.4125,
+      "step": 1054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6789050743634187e-05,
+      "loss": 0.3713,
+      "step": 1055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6782628335261445e-05,
+      "loss": 0.3958,
+      "step": 1056
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6776200741476288e-05,
+      "loss": 0.3519,
+      "step": 1057
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6769767967192702e-05,
+      "loss": 0.3584,
+      "step": 1058
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6763330017328622e-05,
+      "loss": 0.3906,
+      "step": 1059
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675688689680596e-05,
+      "loss": 0.3627,
+      "step": 1060
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6750438610550563e-05,
+      "loss": 0.3738,
+      "step": 1061
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6743985163492236e-05,
+      "loss": 0.4075,
+      "step": 1062
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6737526560564732e-05,
+      "loss": 0.3789,
+      "step": 1063
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6731062806705735e-05,
+      "loss": 0.3559,
+      "step": 1064
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672459390685688e-05,
+      "loss": 0.3981,
+      "step": 1065
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671811986596372e-05,
+      "loss": 0.3699,
+      "step": 1066
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671164068897576e-05,
+      "loss": 0.3974,
+      "step": 1067
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6705156380846413e-05,
+      "loss": 0.3303,
+      "step": 1068
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.669866694653302e-05,
+      "loss": 0.3866,
+      "step": 1069
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.669217239099685e-05,
+      "loss": 0.3726,
+      "step": 1070
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6685672719203067e-05,
+      "loss": 0.3818,
+      "step": 1071
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6679167936120773e-05,
+      "loss": 0.3591,
+      "step": 1072
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667265804672296e-05,
+      "loss": 0.3389,
+      "step": 1073
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666614305598652e-05,
+      "loss": 0.351,
+      "step": 1074
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6659622968892263e-05,
+      "loss": 0.3622,
+      "step": 1075
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6653097790424878e-05,
+      "loss": 0.3924,
+      "step": 1076
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664656752557296e-05,
+      "loss": 0.373,
+      "step": 1077
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6640032179328978e-05,
+      "loss": 0.3973,
+      "step": 1078
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66334917566893e-05,
+      "loss": 0.3747,
+      "step": 1079
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6626946262654167e-05,
+      "loss": 0.3739,
+      "step": 1080
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6620395702227695e-05,
+      "loss": 0.3395,
+      "step": 1081
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6613840080417887e-05,
+      "loss": 0.3969,
+      "step": 1082
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.66072794022366e-05,
+      "loss": 0.3492,
+      "step": 1083
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6600713672699558e-05,
+      "loss": 0.3752,
+      "step": 1084
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6594142896826358e-05,
+      "loss": 0.4046,
+      "step": 1085
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6587567079640447e-05,
+      "loss": 0.4329,
+      "step": 1086
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6580986226169128e-05,
+      "loss": 0.3809,
+      "step": 1087
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657440034144355e-05,
+      "loss": 0.3588,
+      "step": 1088
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6567809430498714e-05,
+      "loss": 0.3809,
+      "step": 1089
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656121349837346e-05,
+      "loss": 0.361,
+      "step": 1090
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6554612550110476e-05,
+      "loss": 0.3999,
+      "step": 1091
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.654800659075627e-05,
+      "loss": 0.3666,
+      "step": 1092
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6541395625361193e-05,
+      "loss": 0.3598,
+      "step": 1093
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6534779658979414e-05,
+      "loss": 0.3571,
+      "step": 1094
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6528158696668936e-05,
+      "loss": 0.3834,
+      "step": 1095
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.652153274349157e-05,
+      "loss": 0.3617,
+      "step": 1096
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6514901804512953e-05,
+      "loss": 0.4006,
+      "step": 1097
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6508265884802523e-05,
+      "loss": 0.3654,
+      "step": 1098
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6501624989433538e-05,
+      "loss": 0.3722,
+      "step": 1099
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6494979123483048e-05,
+      "loss": 0.3452,
+      "step": 1100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.648832829203191e-05,
+      "loss": 0.3664,
+      "step": 1101
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.648167250016478e-05,
+      "loss": 0.3779,
+      "step": 1102
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6475011752970102e-05,
+      "loss": 0.3358,
+      "step": 1103
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.64683460555401e-05,
+      "loss": 0.376,
+      "step": 1104
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6461675412970802e-05,
+      "loss": 0.3824,
+      "step": 1105
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6454999830362e-05,
+      "loss": 0.3704,
+      "step": 1106
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644831931281727e-05,
+      "loss": 0.3686,
+      "step": 1107
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6441633865443956e-05,
+      "loss": 0.3695,
+      "step": 1108
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6434943493353176e-05,
+      "loss": 0.3518,
+      "step": 1109
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6428248201659816e-05,
+      "loss": 0.3608,
+      "step": 1110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6421547995482516e-05,
+      "loss": 0.3627,
+      "step": 1111
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641484287994367e-05,
+      "loss": 0.3491,
+      "step": 1112
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6408132860169433e-05,
+      "loss": 0.3719,
+      "step": 1113
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640141794128971e-05,
+      "loss": 0.3755,
+      "step": 1114
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6394698128438152e-05,
+      "loss": 0.3421,
+      "step": 1115
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638797342675214e-05,
+      "loss": 0.3436,
+      "step": 1116
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6381243841372808e-05,
+      "loss": 0.3612,
+      "step": 1117
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.637450937744501e-05,
+      "loss": 0.3252,
+      "step": 1118
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636777004011734e-05,
+      "loss": 0.3499,
+      "step": 1119
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636102583454211e-05,
+      "loss": 0.408,
+      "step": 1120
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6354276765875364e-05,
+      "loss": 0.3817,
+      "step": 1121
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6347522839276852e-05,
+      "loss": 0.4034,
+      "step": 1122
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6340764059910047e-05,
+      "loss": 0.3438,
+      "step": 1123
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6334000432942127e-05,
+      "loss": 0.3654,
+      "step": 1124
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6327231963543976e-05,
+      "loss": 0.3773,
+      "step": 1125
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6320458656890187e-05,
+      "loss": 0.3515,
+      "step": 1126
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6313680518159037e-05,
+      "loss": 0.3878,
+      "step": 1127
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6306897552532517e-05,
+      "loss": 0.3719,
+      "step": 1128
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6300109765196288e-05,
+      "loss": 0.3947,
+      "step": 1129
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6293317161339714e-05,
+      "loss": 0.3624,
+      "step": 1130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6286519746155825e-05,
+      "loss": 0.397,
+      "step": 1131
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.627971752484135e-05,
+      "loss": 0.4146,
+      "step": 1132
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6272910502596664e-05,
+      "loss": 0.3314,
+      "step": 1133
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6266098684625846e-05,
+      "loss": 0.3656,
+      "step": 1134
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6259282076136616e-05,
+      "loss": 0.3522,
+      "step": 1135
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625246068234036e-05,
+      "loss": 0.3474,
+      "step": 1136
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.624563450845214e-05,
+      "loss": 0.3591,
+      "step": 1137
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6238803559690646e-05,
+      "loss": 0.3808,
+      "step": 1138
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6231967841278246e-05,
+      "loss": 0.3786,
+      "step": 1139
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622512735844093e-05,
+      "loss": 0.3748,
+      "step": 1140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6218282116408342e-05,
+      "loss": 0.3778,
+      "step": 1141
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6211432120413766e-05,
+      "loss": 0.3829,
+      "step": 1142
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6204577375694116e-05,
+      "loss": 0.3952,
+      "step": 1143
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.619771788748994e-05,
+      "loss": 0.4033,
+      "step": 1144
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6190853661045413e-05,
+      "loss": 0.3862,
+      "step": 1145
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6183984701608326e-05,
+      "loss": 0.4143,
+      "step": 1146
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6177111014430092e-05,
+      "loss": 0.4007,
+      "step": 1147
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6170232604765746e-05,
+      "loss": 0.3873,
+      "step": 1148
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.616334947787392e-05,
+      "loss": 0.3412,
+      "step": 1149
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615646163901686e-05,
+      "loss": 0.3978,
+      "step": 1150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6149569093460417e-05,
+      "loss": 0.377,
+      "step": 1151
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6142671846474032e-05,
+      "loss": 0.3704,
+      "step": 1152
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6135769903330746e-05,
+      "loss": 0.3789,
+      "step": 1153
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6128863269307193e-05,
+      "loss": 0.375,
+      "step": 1154
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6121951949683584e-05,
+      "loss": 0.3655,
+      "step": 1155
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6115035949743725e-05,
+      "loss": 0.3814,
+      "step": 1156
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.610811527477498e-05,
+      "loss": 0.3612,
+      "step": 1157
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6101189930068316e-05,
+      "loss": 0.3721,
+      "step": 1158
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6094259920918246e-05,
+      "loss": 0.3458,
+      "step": 1159
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6087325252622853e-05,
+      "loss": 0.3571,
+      "step": 1160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608038593048379e-05,
+      "loss": 0.3697,
+      "step": 1161
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607344195980627e-05,
+      "loss": 0.3999,
+      "step": 1162
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6066493345899043e-05,
+      "loss": 0.3557,
+      "step": 1163
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6059540094074427e-05,
+      "loss": 0.405,
+      "step": 1164
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6052582209648275e-05,
+      "loss": 0.3668,
+      "step": 1165
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6045619697939986e-05,
+      "loss": 0.3952,
+      "step": 1166
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6038652564272497e-05,
+      "loss": 0.3823,
+      "step": 1167
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.603168081397228e-05,
+      "loss": 0.3478,
+      "step": 1168
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6024704452369323e-05,
+      "loss": 0.3974,
+      "step": 1169
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6017723484797163e-05,
+      "loss": 0.3898,
+      "step": 1170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6010737916592835e-05,
+      "loss": 0.3933,
+      "step": 1171
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6003747753096908e-05,
+      "loss": 0.3738,
+      "step": 1172
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599675299965346e-05,
+      "loss": 0.3951,
+      "step": 1173
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5989753661610068e-05,
+      "loss": 0.318,
+      "step": 1174
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5982749744317825e-05,
+      "loss": 0.3811,
+      "step": 1175
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5975741253131326e-05,
+      "loss": 0.3624,
+      "step": 1176
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.596872819340866e-05,
+      "loss": 0.3732,
+      "step": 1177
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5961710570511395e-05,
+      "loss": 0.3697,
+      "step": 1178
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5954688389804612e-05,
+      "loss": 0.3556,
+      "step": 1179
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.594766165665686e-05,
+      "loss": 0.3856,
+      "step": 1180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5940630376440175e-05,
+      "loss": 0.3558,
+      "step": 1181
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5933594554530064e-05,
+      "loss": 0.3813,
+      "step": 1182
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592655419630551e-05,
+      "loss": 0.3514,
+      "step": 1183
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5919509307148967e-05,
+      "loss": 0.3706,
+      "step": 1184
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5912459892446346e-05,
+      "loss": 0.3603,
+      "step": 1185
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5905405957587022e-05,
+      "loss": 0.3738,
+      "step": 1186
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5898347507963827e-05,
+      "loss": 0.4019,
+      "step": 1187
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589128454897304e-05,
+      "loss": 0.3714,
+      "step": 1188
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588421708601439e-05,
+      "loss": 0.4093,
+      "step": 1189
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.587714512449105e-05,
+      "loss": 0.3788,
+      "step": 1190
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5870068669809634e-05,
+      "loss": 0.4068,
+      "step": 1191
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586298772738018e-05,
+      "loss": 0.3921,
+      "step": 1192
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585590230261618e-05,
+      "loss": 0.3576,
+      "step": 1193
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5848812400934527e-05,
+      "loss": 0.3767,
+      "step": 1194
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584171802775555e-05,
+      "loss": 0.3412,
+      "step": 1195
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5834619188503e-05,
+      "loss": 0.3768,
+      "step": 1196
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.582751588860403e-05,
+      "loss": 0.3808,
+      "step": 1197
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5820408133489215e-05,
+      "loss": 0.3763,
+      "step": 1198
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5813295928592533e-05,
+      "loss": 0.3959,
+      "step": 1199
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580617927935135e-05,
+      "loss": 0.3795,
+      "step": 1200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.579905819120646e-05,
+      "loss": 0.3436,
+      "step": 1201
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5791932669602023e-05,
+      "loss": 0.3536,
+      "step": 1202
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5784802719985595e-05,
+      "loss": 0.3726,
+      "step": 1203
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5777668347808122e-05,
+      "loss": 0.3664,
+      "step": 1204
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5770529558523937e-05,
+      "loss": 0.363,
+      "step": 1205
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5763386357590734e-05,
+      "loss": 0.3424,
+      "step": 1206
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5756238750469593e-05,
+      "loss": 0.3638,
+      "step": 1207
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5749086742624952e-05,
+      "loss": 0.3812,
+      "step": 1208
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5741930339524623e-05,
+      "loss": 0.4336,
+      "step": 1209
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5734769546639776e-05,
+      "loss": 0.3588,
+      "step": 1210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5727604369444932e-05,
+      "loss": 0.3634,
+      "step": 1211
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572043481341797e-05,
+      "loss": 0.3867,
+      "step": 1212
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571326088404011e-05,
+      "loss": 0.3711,
+      "step": 1213
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5706082586795924e-05,
+      "loss": 0.3336,
+      "step": 1214
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5698899927173316e-05,
+      "loss": 0.3834,
+      "step": 1215
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5691712910663526e-05,
+      "loss": 0.4138,
+      "step": 1216
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5684521542761134e-05,
+      "loss": 0.3803,
+      "step": 1217
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5677325828964035e-05,
+      "loss": 0.3849,
+      "step": 1218
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5670125774773447e-05,
+      "loss": 0.3497,
+      "step": 1219
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5662921385693914e-05,
+      "loss": 0.3643,
+      "step": 1220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5655712667233293e-05,
+      "loss": 0.3825,
+      "step": 1221
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5648499624902743e-05,
+      "loss": 0.3809,
+      "step": 1222
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.564128226421674e-05,
+      "loss": 0.3553,
+      "step": 1223
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5634060590693046e-05,
+      "loss": 0.3703,
+      "step": 1224
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5626834609852738e-05,
+      "loss": 0.3737,
+      "step": 1225
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561960432722017e-05,
+      "loss": 0.4092,
+      "step": 1226
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5612369748323e-05,
+      "loss": 0.3649,
+      "step": 1227
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5605130878692162e-05,
+      "loss": 0.373,
+      "step": 1228
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5597887723861864e-05,
+      "loss": 0.3482,
+      "step": 1229
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5590640289369605e-05,
+      "loss": 0.3938,
+      "step": 1230
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5583388580756144e-05,
+      "loss": 0.3305,
+      "step": 1231
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557613260356551e-05,
+      "loss": 0.3994,
+      "step": 1232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5568872363345e-05,
+      "loss": 0.3754,
+      "step": 1233
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5561607865645174e-05,
+      "loss": 0.3399,
+      "step": 1234
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.555433911601983e-05,
+      "loss": 0.4098,
+      "step": 1235
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5547066120026025e-05,
+      "loss": 0.3576,
+      "step": 1236
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553978888322407e-05,
+      "loss": 0.2817,
+      "step": 1237
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5532507411177514e-05,
+      "loss": 0.2958,
+      "step": 1238
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5525221709453142e-05,
+      "loss": 0.2582,
+      "step": 1239
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5517931783620968e-05,
+      "loss": 0.2672,
+      "step": 1240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5510637639254245e-05,
+      "loss": 0.2622,
+      "step": 1241
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5503339281929442e-05,
+      "loss": 0.2843,
+      "step": 1242
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5496036717226257e-05,
+      "loss": 0.2503,
+      "step": 1243
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54887299507276e-05,
+      "loss": 0.2565,
+      "step": 1244
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5481418988019592e-05,
+      "loss": 0.2538,
+      "step": 1245
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547410383469157e-05,
+      "loss": 0.2726,
+      "step": 1246
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546678449633606e-05,
+      "loss": 0.2559,
+      "step": 1247
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5459460978548806e-05,
+      "loss": 0.2724,
+      "step": 1248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5452133286928732e-05,
+      "loss": 0.2759,
+      "step": 1249
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5444801427077964e-05,
+      "loss": 0.2381,
+      "step": 1250
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54374654046018e-05,
+      "loss": 0.2598,
+      "step": 1251
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543012522510874e-05,
+      "loss": 0.2636,
+      "step": 1252
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5422780894210446e-05,
+      "loss": 0.2443,
+      "step": 1253
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5415432417521766e-05,
+      "loss": 0.2636,
+      "step": 1254
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5408079800660697e-05,
+      "loss": 0.2701,
+      "step": 1255
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5400723049248433e-05,
+      "loss": 0.2622,
+      "step": 1256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5393362168909303e-05,
+      "loss": 0.2975,
+      "step": 1257
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.53859971652708e-05,
+      "loss": 0.2653,
+      "step": 1258
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5378628043963567e-05,
+      "loss": 0.2594,
+      "step": 1259
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5371254810621414e-05,
+      "loss": 0.2575,
+      "step": 1260
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536387747088126e-05,
+      "loss": 0.2431,
+      "step": 1261
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.535649603038319e-05,
+      "loss": 0.2384,
+      "step": 1262
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5349110494770423e-05,
+      "loss": 0.2618,
+      "step": 1263
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5341720869689294e-05,
+      "loss": 0.2556,
+      "step": 1264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5334327160789273e-05,
+      "loss": 0.2373,
+      "step": 1265
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5326929373722957e-05,
+      "loss": 0.2479,
+      "step": 1266
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.531952751414605e-05,
+      "loss": 0.2657,
+      "step": 1267
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.531212158771738e-05,
+      "loss": 0.2664,
+      "step": 1268
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5304711600098882e-05,
+      "loss": 0.2606,
+      "step": 1269
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.529729755695558e-05,
+      "loss": 0.25,
+      "step": 1270
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5289879463955622e-05,
+      "loss": 0.2556,
+      "step": 1271
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5282457326770245e-05,
+      "loss": 0.2526,
+      "step": 1272
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5275031151073768e-05,
+      "loss": 0.2498,
+      "step": 1273
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5267600942543606e-05,
+      "loss": 0.2502,
+      "step": 1274
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5260166706860254e-05,
+      "loss": 0.252,
+      "step": 1275
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5252728449707288e-05,
+      "loss": 0.2791,
+      "step": 1276
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5245286176771361e-05,
+      "loss": 0.2596,
+      "step": 1277
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5237839893742188e-05,
+      "loss": 0.2435,
+      "step": 1278
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5230389606312557e-05,
+      "loss": 0.2494,
+      "step": 1279
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5222935320178315e-05,
+      "loss": 0.2601,
+      "step": 1280
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5215477041038364e-05,
+      "loss": 0.2423,
+      "step": 1281
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5208014774594663e-05,
+      "loss": 0.268,
+      "step": 1282
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5200548526552219e-05,
+      "loss": 0.2286,
+      "step": 1283
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5193078302619073e-05,
+      "loss": 0.2545,
+      "step": 1284
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5185604108506327e-05,
+      "loss": 0.2526,
+      "step": 1285
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.517812594992809e-05,
+      "loss": 0.2859,
+      "step": 1286
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5170643832601526e-05,
+      "loss": 0.2301,
+      "step": 1287
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5163157762246814e-05,
+      "loss": 0.2507,
+      "step": 1288
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.515566774458716e-05,
+      "loss": 0.2658,
+      "step": 1289
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5148173785348777e-05,
+      "loss": 0.2764,
+      "step": 1290
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5140675890260912e-05,
+      "loss": 0.2545,
+      "step": 1291
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.51331740650558e-05,
+      "loss": 0.2489,
+      "step": 1292
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5125668315468689e-05,
+      "loss": 0.2533,
+      "step": 1293
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5118158647237833e-05,
+      "loss": 0.2412,
+      "step": 1294
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5110645066104473e-05,
+      "loss": 0.2507,
+      "step": 1295
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5103127577812844e-05,
+      "loss": 0.2493,
+      "step": 1296
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5095606188110173e-05,
+      "loss": 0.2302,
+      "step": 1297
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.508808090274666e-05,
+      "loss": 0.2739,
+      "step": 1298
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5080551727475494e-05,
+      "loss": 0.2551,
+      "step": 1299
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.507301866805283e-05,
+      "loss": 0.2585,
+      "step": 1300
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5065481730237795e-05,
+      "loss": 0.2396,
+      "step": 1301
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5057940919792482e-05,
+      "loss": 0.2543,
+      "step": 1302
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5050396242481946e-05,
+      "loss": 0.2495,
+      "step": 1303
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.504284770407419e-05,
+      "loss": 0.2775,
+      "step": 1304
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5035295310340186e-05,
+      "loss": 0.2718,
+      "step": 1305
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5027739067053832e-05,
+      "loss": 0.2459,
+      "step": 1306
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5020178979991986e-05,
+      "loss": 0.2697,
+      "step": 1307
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5012615054934436e-05,
+      "loss": 0.2599,
+      "step": 1308
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5005047297663905e-05,
+      "loss": 0.2589,
+      "step": 1309
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.499747571396605e-05,
+      "loss": 0.2562,
+      "step": 1310
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4989900309629453e-05,
+      "loss": 0.2642,
+      "step": 1311
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4982321090445609e-05,
+      "loss": 0.2367,
+      "step": 1312
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4974738062208935e-05,
+      "loss": 0.2489,
+      "step": 1313
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4967151230716763e-05,
+      "loss": 0.2432,
+      "step": 1314
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.495956060176933e-05,
+      "loss": 0.2344,
+      "step": 1315
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4951966181169775e-05,
+      "loss": 0.2463,
+      "step": 1316
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4944367974724137e-05,
+      "loss": 0.2577,
+      "step": 1317
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4936765988241346e-05,
+      "loss": 0.2486,
+      "step": 1318
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4929160227533233e-05,
+      "loss": 0.2579,
+      "step": 1319
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4921550698414497e-05,
+      "loss": 0.2448,
+      "step": 1320
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4913937406702735e-05,
+      "loss": 0.2502,
+      "step": 1321
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4906320358218407e-05,
+      "loss": 0.2629,
+      "step": 1322
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4898699558784858e-05,
+      "loss": 0.2693,
+      "step": 1323
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4891075014228282e-05,
+      "loss": 0.2378,
+      "step": 1324
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4883446730377762e-05,
+      "loss": 0.2425,
+      "step": 1325
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4875814713065216e-05,
+      "loss": 0.2524,
+      "step": 1326
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4868178968125427e-05,
+      "loss": 0.2436,
+      "step": 1327
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4860539501396031e-05,
+      "loss": 0.2471,
+      "step": 1328
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4852896318717504e-05,
+      "loss": 0.2409,
+      "step": 1329
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.484524942593316e-05,
+      "loss": 0.2423,
+      "step": 1330
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4837598828889159e-05,
+      "loss": 0.2413,
+      "step": 1331
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4829944533434484e-05,
+      "loss": 0.2683,
+      "step": 1332
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4822286545420948e-05,
+      "loss": 0.2395,
+      "step": 1333
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4814624870703194e-05,
+      "loss": 0.269,
+      "step": 1334
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4806959515138667e-05,
+      "loss": 0.2594,
+      "step": 1335
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4799290484587646e-05,
+      "loss": 0.2601,
+      "step": 1336
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4791617784913203e-05,
+      "loss": 0.2478,
+      "step": 1337
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.478394142198123e-05,
+      "loss": 0.2592,
+      "step": 1338
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4776261401660404e-05,
+      "loss": 0.2303,
+      "step": 1339
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4768577729822214e-05,
+      "loss": 0.2442,
+      "step": 1340
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.476089041234092e-05,
+      "loss": 0.238,
+      "step": 1341
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4753199455093596e-05,
+      "loss": 0.2579,
+      "step": 1342
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4745504863960076e-05,
+      "loss": 0.2637,
+      "step": 1343
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4737806644822984e-05,
+      "loss": 0.2487,
+      "step": 1344
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4730104803567708e-05,
+      "loss": 0.2597,
+      "step": 1345
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4722399346082421e-05,
+      "loss": 0.2803,
+      "step": 1346
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4714690278258039e-05,
+      "loss": 0.2428,
+      "step": 1347
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4706977605988264e-05,
+      "loss": 0.2506,
+      "step": 1348
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.469926133516953e-05,
+      "loss": 0.2639,
+      "step": 1349
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4691541471701034e-05,
+      "loss": 0.2637,
+      "step": 1350
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4683818021484717e-05,
+      "loss": 0.2428,
+      "step": 1351
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4676090990425265e-05,
+      "loss": 0.2394,
+      "step": 1352
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4668360384430098e-05,
+      "loss": 0.2574,
+      "step": 1353
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4660626209409371e-05,
+      "loss": 0.2283,
+      "step": 1354
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4652888471275963e-05,
+      "loss": 0.2822,
+      "step": 1355
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4645147175945489e-05,
+      "loss": 0.2662,
+      "step": 1356
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4637402329336268e-05,
+      "loss": 0.268,
+      "step": 1357
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4629653937369343e-05,
+      "loss": 0.2274,
+      "step": 1358
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4621902005968469e-05,
+      "loss": 0.2734,
+      "step": 1359
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4614146541060103e-05,
+      "loss": 0.2671,
+      "step": 1360
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.46063875485734e-05,
+      "loss": 0.2288,
+      "step": 1361
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4598625034440224e-05,
+      "loss": 0.2223,
+      "step": 1362
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4590859004595115e-05,
+      "loss": 0.2533,
+      "step": 1363
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4583089464975315e-05,
+      "loss": 0.2728,
+      "step": 1364
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4575316421520742e-05,
+      "loss": 0.2614,
+      "step": 1365
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4567539880173995e-05,
+      "loss": 0.2615,
+      "step": 1366
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4559759846880344e-05,
+      "loss": 0.2683,
+      "step": 1367
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4551976327587736e-05,
+      "loss": 0.2565,
+      "step": 1368
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4544189328246774e-05,
+      "loss": 0.2689,
+      "step": 1369
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4536398854810727e-05,
+      "loss": 0.2921,
+      "step": 1370
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.452860491323552e-05,
+      "loss": 0.2423,
+      "step": 1371
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4520807509479726e-05,
+      "loss": 0.2513,
+      "step": 1372
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4513006649504569e-05,
+      "loss": 0.2688,
+      "step": 1373
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4505202339273916e-05,
+      "loss": 0.2416,
+      "step": 1374
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.449739458475426e-05,
+      "loss": 0.2452,
+      "step": 1375
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.448958339191475e-05,
+      "loss": 0.2444,
+      "step": 1376
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4481768766727138e-05,
+      "loss": 0.2496,
+      "step": 1377
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4473950715165817e-05,
+      "loss": 0.2605,
+      "step": 1378
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4466129243207796e-05,
+      "loss": 0.223,
+      "step": 1379
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4458304356832695e-05,
+      "loss": 0.2534,
+      "step": 1380
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4450476062022749e-05,
+      "loss": 0.2574,
+      "step": 1381
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4442644364762797e-05,
+      "loss": 0.2549,
+      "step": 1382
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.443480927104028e-05,
+      "loss": 0.2622,
+      "step": 1383
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4426970786845228e-05,
+      "loss": 0.2346,
+      "step": 1384
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4419128918170278e-05,
+      "loss": 0.2627,
+      "step": 1385
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4411283671010638e-05,
+      "loss": 0.2466,
+      "step": 1386
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4403435051364116e-05,
+      "loss": 0.2358,
+      "step": 1387
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4395583065231083e-05,
+      "loss": 0.2766,
+      "step": 1388
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4387727718614492e-05,
+      "loss": 0.2694,
+      "step": 1389
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4379869017519865e-05,
+      "loss": 0.2558,
+      "step": 1390
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4372006967955285e-05,
+      "loss": 0.2525,
+      "step": 1391
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4364141575931395e-05,
+      "loss": 0.2284,
+      "step": 1392
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4356272847461399e-05,
+      "loss": 0.2753,
+      "step": 1393
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4348400788561046e-05,
+      "loss": 0.2335,
+      "step": 1394
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.434052540524863e-05,
+      "loss": 0.2534,
+      "step": 1395
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4332646703544995e-05,
+      "loss": 0.2574,
+      "step": 1396
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4324764689473517e-05,
+      "loss": 0.2412,
+      "step": 1397
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4316879369060095e-05,
+      "loss": 0.2534,
+      "step": 1398
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4308990748333176e-05,
+      "loss": 0.244,
+      "step": 1399
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.430109883332371e-05,
+      "loss": 0.2519,
+      "step": 1400
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4293203630065177e-05,
+      "loss": 0.2483,
+      "step": 1401
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4285305144593571e-05,
+      "loss": 0.2586,
+      "step": 1402
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.427740338294739e-05,
+      "loss": 0.2648,
+      "step": 1403
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.426949835116764e-05,
+      "loss": 0.254,
+      "step": 1404
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4261590055297828e-05,
+      "loss": 0.2508,
+      "step": 1405
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.425367850138395e-05,
+      "loss": 0.2574,
+      "step": 1406
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4245763695474498e-05,
+      "loss": 0.25,
+      "step": 1407
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4237845643620454e-05,
+      "loss": 0.2477,
+      "step": 1408
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4229924351875272e-05,
+      "loss": 0.2519,
+      "step": 1409
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4221999826294897e-05,
+      "loss": 0.2671,
+      "step": 1410
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.421407207293773e-05,
+      "loss": 0.2653,
+      "step": 1411
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4206141097864645e-05,
+      "loss": 0.2478,
+      "step": 1412
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.419820690713899e-05,
+      "loss": 0.2463,
+      "step": 1413
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4190269506826556e-05,
+      "loss": 0.2394,
+      "step": 1414
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4182328902995594e-05,
+      "loss": 0.2787,
+      "step": 1415
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.417438510171681e-05,
+      "loss": 0.2605,
+      "step": 1416
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4166438109063347e-05,
+      "loss": 0.2503,
+      "step": 1417
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4158487931110786e-05,
+      "loss": 0.2635,
+      "step": 1418
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.415053457393715e-05,
+      "loss": 0.2555,
+      "step": 1419
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4142578043622892e-05,
+      "loss": 0.2404,
+      "step": 1420
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4134618346250883e-05,
+      "loss": 0.2583,
+      "step": 1421
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4126655487906426e-05,
+      "loss": 0.2686,
+      "step": 1422
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4118689474677234e-05,
+      "loss": 0.2884,
+      "step": 1423
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4110720312653432e-05,
+      "loss": 0.273,
+      "step": 1424
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4102748007927555e-05,
+      "loss": 0.2619,
+      "step": 1425
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4094772566594537e-05,
+      "loss": 0.2939,
+      "step": 1426
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.408679399475172e-05,
+      "loss": 0.2518,
+      "step": 1427
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4078812298498821e-05,
+      "loss": 0.2518,
+      "step": 1428
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4070827483937963e-05,
+      "loss": 0.2852,
+      "step": 1429
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4062839557173641e-05,
+      "loss": 0.2548,
+      "step": 1430
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4054848524312741e-05,
+      "loss": 0.2609,
+      "step": 1431
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4046854391464515e-05,
+      "loss": 0.2484,
+      "step": 1432
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4038857164740584e-05,
+      "loss": 0.2678,
+      "step": 1433
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4030856850254946e-05,
+      "loss": 0.271,
+      "step": 1434
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4022853454123939e-05,
+      "loss": 0.2583,
+      "step": 1435
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.401484698246628e-05,
+      "loss": 0.264,
+      "step": 1436
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4006837441403018e-05,
+      "loss": 0.2527,
+      "step": 1437
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3998824837057563e-05,
+      "loss": 0.2691,
+      "step": 1438
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3990809175555657e-05,
+      "loss": 0.2586,
+      "step": 1439
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3982790463025389e-05,
+      "loss": 0.2732,
+      "step": 1440
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3974768705597166e-05,
+      "loss": 0.2444,
+      "step": 1441
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.396674390940374e-05,
+      "loss": 0.2684,
+      "step": 1442
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.395871608058017e-05,
+      "loss": 0.2565,
+      "step": 1443
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.395068522526385e-05,
+      "loss": 0.2508,
+      "step": 1444
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3942651349594472e-05,
+      "loss": 0.2385,
+      "step": 1445
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3934614459714046e-05,
+      "loss": 0.2661,
+      "step": 1446
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3926574561766888e-05,
+      "loss": 0.2665,
+      "step": 1447
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3918531661899606e-05,
+      "loss": 0.2645,
+      "step": 1448
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3910485766261106e-05,
+      "loss": 0.2765,
+      "step": 1449
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3902436881002595e-05,
+      "loss": 0.2228,
+      "step": 1450
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3894385012277545e-05,
+      "loss": 0.2493,
+      "step": 1451
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3886330166241725e-05,
+      "loss": 0.272,
+      "step": 1452
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3878272349053176e-05,
+      "loss": 0.2626,
+      "step": 1453
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3870211566872209e-05,
+      "loss": 0.249,
+      "step": 1454
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.38621478258614e-05,
+      "loss": 0.2555,
+      "step": 1455
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3854081132185597e-05,
+      "loss": 0.2571,
+      "step": 1456
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3846011492011889e-05,
+      "loss": 0.251,
+      "step": 1457
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3837938911509631e-05,
+      "loss": 0.2489,
+      "step": 1458
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3829863396850423e-05,
+      "loss": 0.259,
+      "step": 1459
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3821784954208104e-05,
+      "loss": 0.2598,
+      "step": 1460
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3813703589758754e-05,
+      "loss": 0.2537,
+      "step": 1461
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3805619309680691e-05,
+      "loss": 0.2498,
+      "step": 1462
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3797532120154453e-05,
+      "loss": 0.2399,
+      "step": 1463
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3789442027362809e-05,
+      "loss": 0.2731,
+      "step": 1464
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3781349037490746e-05,
+      "loss": 0.2416,
+      "step": 1465
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3773253156725468e-05,
+      "loss": 0.2667,
+      "step": 1466
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3765154391256386e-05,
+      "loss": 0.244,
+      "step": 1467
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3757052747275115e-05,
+      "loss": 0.261,
+      "step": 1468
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3748948230975472e-05,
+      "loss": 0.2576,
+      "step": 1469
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3740840848553478e-05,
+      "loss": 0.2424,
+      "step": 1470
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3732730606207332e-05,
+      "loss": 0.2656,
+      "step": 1471
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3724617510137426e-05,
+      "loss": 0.2583,
+      "step": 1472
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3716501566546334e-05,
+      "loss": 0.2532,
+      "step": 1473
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3708382781638809e-05,
+      "loss": 0.2466,
+      "step": 1474
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3700261161621767e-05,
+      "loss": 0.2664,
+      "step": 1475
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3692136712704305e-05,
+      "loss": 0.2484,
+      "step": 1476
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.368400944109767e-05,
+      "loss": 0.2546,
+      "step": 1477
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3675879353015275e-05,
+      "loss": 0.2766,
+      "step": 1478
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3667746454672685e-05,
+      "loss": 0.2642,
+      "step": 1479
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3659610752287614e-05,
+      "loss": 0.2605,
+      "step": 1480
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3651472252079912e-05,
+      "loss": 0.2524,
+      "step": 1481
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3643330960271577e-05,
+      "loss": 0.2308,
+      "step": 1482
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3635186883086742e-05,
+      "loss": 0.2758,
+      "step": 1483
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3627040026751658e-05,
+      "loss": 0.2515,
+      "step": 1484
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3618890397494719e-05,
+      "loss": 0.28,
+      "step": 1485
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3610738001546417e-05,
+      "loss": 0.2554,
+      "step": 1486
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.360258284513938e-05,
+      "loss": 0.272,
+      "step": 1487
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.359442493450833e-05,
+      "loss": 0.248,
+      "step": 1488
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3586264275890107e-05,
+      "loss": 0.2611,
+      "step": 1489
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.357810087552364e-05,
+      "loss": 0.23,
+      "step": 1490
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3569934739649968e-05,
+      "loss": 0.252,
+      "step": 1491
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3561765874512208e-05,
+      "loss": 0.2584,
+      "step": 1492
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3553594286355575e-05,
+      "loss": 0.2601,
+      "step": 1493
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3545419981427355e-05,
+      "loss": 0.2415,
+      "step": 1494
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3537242965976918e-05,
+      "loss": 0.2605,
+      "step": 1495
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3529063246255704e-05,
+      "loss": 0.2349,
+      "step": 1496
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3520880828517224e-05,
+      "loss": 0.244,
+      "step": 1497
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3512695719017046e-05,
+      "loss": 0.257,
+      "step": 1498
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3504507924012802e-05,
+      "loss": 0.2666,
+      "step": 1499
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3496317449764168e-05,
+      "loss": 0.2614,
+      "step": 1500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3488124302532875e-05,
+      "loss": 0.264,
+      "step": 1501
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3479928488582696e-05,
+      "loss": 0.2912,
+      "step": 1502
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3471730014179449e-05,
+      "loss": 0.2401,
+      "step": 1503
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3463528885590975e-05,
+      "loss": 0.2697,
+      "step": 1504
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3455325109087154e-05,
+      "loss": 0.2695,
+      "step": 1505
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3447118690939876e-05,
+      "loss": 0.2604,
+      "step": 1506
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3438909637423071e-05,
+      "loss": 0.2674,
+      "step": 1507
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3430697954812669e-05,
+      "loss": 0.2475,
+      "step": 1508
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.342248364938661e-05,
+      "loss": 0.2463,
+      "step": 1509
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.341426672742485e-05,
+      "loss": 0.2579,
+      "step": 1510
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3406047195209336e-05,
+      "loss": 0.2656,
+      "step": 1511
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3397825059024008e-05,
+      "loss": 0.2593,
+      "step": 1512
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3389600325154813e-05,
+      "loss": 0.2537,
+      "step": 1513
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3381372999889662e-05,
+      "loss": 0.2544,
+      "step": 1514
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.337314308951846e-05,
+      "loss": 0.2471,
+      "step": 1515
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3364910600333094e-05,
+      "loss": 0.2672,
+      "step": 1516
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3356675538627406e-05,
+      "loss": 0.2453,
+      "step": 1517
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3348437910697221e-05,
+      "loss": 0.2633,
+      "step": 1518
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3340197722840314e-05,
+      "loss": 0.2535,
+      "step": 1519
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.333195498135642e-05,
+      "loss": 0.2316,
+      "step": 1520
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3323709692547233e-05,
+      "loss": 0.2681,
+      "step": 1521
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3315461862716385e-05,
+      "loss": 0.2622,
+      "step": 1522
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3307211498169459e-05,
+      "loss": 0.2437,
+      "step": 1523
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.329895860521397e-05,
+      "loss": 0.2764,
+      "step": 1524
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3290703190159367e-05,
+      "loss": 0.259,
+      "step": 1525
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3282445259317028e-05,
+      "loss": 0.2389,
+      "step": 1526
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3274184819000258e-05,
+      "loss": 0.2452,
+      "step": 1527
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.326592187552427e-05,
+      "loss": 0.2402,
+      "step": 1528
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3257656435206204e-05,
+      "loss": 0.2612,
+      "step": 1529
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3249388504365102e-05,
+      "loss": 0.2774,
+      "step": 1530
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3241118089321905e-05,
+      "loss": 0.2715,
+      "step": 1531
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3232845196399466e-05,
+      "loss": 0.2424,
+      "step": 1532
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.322456983192252e-05,
+      "loss": 0.2381,
+      "step": 1533
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.32162920022177e-05,
+      "loss": 0.2414,
+      "step": 1534
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3208011713613518e-05,
+      "loss": 0.2757,
+      "step": 1535
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3199728972440373e-05,
+      "loss": 0.2658,
+      "step": 1536
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3191443785030527e-05,
+      "loss": 0.2639,
+      "step": 1537
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3183156157718122e-05,
+      "loss": 0.2554,
+      "step": 1538
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3174866096839165e-05,
+      "loss": 0.2534,
+      "step": 1539
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3166573608731518e-05,
+      "loss": 0.2699,
+      "step": 1540
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3158278699734906e-05,
+      "loss": 0.2568,
+      "step": 1541
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3149981376190899e-05,
+      "loss": 0.2684,
+      "step": 1542
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3141681644442912e-05,
+      "loss": 0.275,
+      "step": 1543
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3133379510836209e-05,
+      "loss": 0.2566,
+      "step": 1544
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3125074981717879e-05,
+      "loss": 0.2711,
+      "step": 1545
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.311676806343685e-05,
+      "loss": 0.256,
+      "step": 1546
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3108458762343881e-05,
+      "loss": 0.2677,
+      "step": 1547
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3100147084791544e-05,
+      "loss": 0.2688,
+      "step": 1548
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.309183303713422e-05,
+      "loss": 0.2614,
+      "step": 1549
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.308351662572813e-05,
+      "loss": 0.2454,
+      "step": 1550
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3075197856931273e-05,
+      "loss": 0.2452,
+      "step": 1551
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3066876737103463e-05,
+      "loss": 0.2511,
+      "step": 1552
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3058553272606314e-05,
+      "loss": 0.2568,
+      "step": 1553
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3050227469803222e-05,
+      "loss": 0.2543,
+      "step": 1554
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3041899335059387e-05,
+      "loss": 0.2337,
+      "step": 1555
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3033568874741778e-05,
+      "loss": 0.2532,
+      "step": 1556
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3025236095219145e-05,
+      "loss": 0.2517,
+      "step": 1557
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3016901002862013e-05,
+      "loss": 0.235,
+      "step": 1558
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3008563604042673e-05,
+      "loss": 0.269,
+      "step": 1559
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3000223905135183e-05,
+      "loss": 0.2493,
+      "step": 1560
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2991881912515357e-05,
+      "loss": 0.2539,
+      "step": 1561
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.298353763256076e-05,
+      "loss": 0.2703,
+      "step": 1562
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.297519107165071e-05,
+      "loss": 0.2589,
+      "step": 1563
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2966842236166273e-05,
+      "loss": 0.2527,
+      "step": 1564
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.295849113249024e-05,
+      "loss": 0.2735,
+      "step": 1565
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.295013776700715e-05,
+      "loss": 0.2619,
+      "step": 1566
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.294178214610326e-05,
+      "loss": 0.2473,
+      "step": 1567
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2933424276166564e-05,
+      "loss": 0.2651,
+      "step": 1568
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.292506416358676e-05,
+      "loss": 0.2458,
+      "step": 1569
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2916701814755275e-05,
+      "loss": 0.2598,
+      "step": 1570
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2908337236065237e-05,
+      "loss": 0.2377,
+      "step": 1571
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2899970433911482e-05,
+      "loss": 0.2521,
+      "step": 1572
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2891601414690544e-05,
+      "loss": 0.2427,
+      "step": 1573
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2883230184800655e-05,
+      "loss": 0.2963,
+      "step": 1574
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2874856750641734e-05,
+      "loss": 0.2432,
+      "step": 1575
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2866481118615385e-05,
+      "loss": 0.2678,
+      "step": 1576
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2858103295124891e-05,
+      "loss": 0.2597,
+      "step": 1577
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2849723286575222e-05,
+      "loss": 0.2519,
+      "step": 1578
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2841341099373001e-05,
+      "loss": 0.2594,
+      "step": 1579
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2832956739926527e-05,
+      "loss": 0.2628,
+      "step": 1580
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2824570214645764e-05,
+      "loss": 0.2541,
+      "step": 1581
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.281618152994232e-05,
+      "loss": 0.2561,
+      "step": 1582
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2807790692229455e-05,
+      "loss": 0.2589,
+      "step": 1583
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.279939770792209e-05,
+      "loss": 0.2703,
+      "step": 1584
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.279100258343677e-05,
+      "loss": 0.2787,
+      "step": 1585
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2782605325191682e-05,
+      "loss": 0.2525,
+      "step": 1586
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.277420593960665e-05,
+      "loss": 0.2689,
+      "step": 1587
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2765804433103118e-05,
+      "loss": 0.2534,
+      "step": 1588
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2757400812104149e-05,
+      "loss": 0.2304,
+      "step": 1589
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.274899508303443e-05,
+      "loss": 0.251,
+      "step": 1590
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2740587252320255e-05,
+      "loss": 0.2582,
+      "step": 1591
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2732177326389528e-05,
+      "loss": 0.2463,
+      "step": 1592
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2723765311671752e-05,
+      "loss": 0.2507,
+      "step": 1593
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2715351214598026e-05,
+      "loss": 0.2692,
+      "step": 1594
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2706935041601043e-05,
+      "loss": 0.2515,
+      "step": 1595
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2698516799115083e-05,
+      "loss": 0.2619,
+      "step": 1596
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.269009649357601e-05,
+      "loss": 0.2397,
+      "step": 1597
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.268167413142126e-05,
+      "loss": 0.2642,
+      "step": 1598
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2673249719089849e-05,
+      "loss": 0.2516,
+      "step": 1599
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2664823263022345e-05,
+      "loss": 0.2764,
+      "step": 1600
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2656394769660904e-05,
+      "loss": 0.2435,
+      "step": 1601
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2647964245449216e-05,
+      "loss": 0.2139,
+      "step": 1602
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2639531696832531e-05,
+      "loss": 0.251,
+      "step": 1603
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2631097130257652e-05,
+      "loss": 0.2393,
+      "step": 1604
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2622660552172926e-05,
+      "loss": 0.254,
+      "step": 1605
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2614221969028217e-05,
+      "loss": 0.2767,
+      "step": 1606
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2605781387274956e-05,
+      "loss": 0.2459,
+      "step": 1607
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.259733881336607e-05,
+      "loss": 0.2398,
+      "step": 1608
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2588894253756027e-05,
+      "loss": 0.2538,
+      "step": 1609
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2580447714900813e-05,
+      "loss": 0.2466,
+      "step": 1610
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2571999203257917e-05,
+      "loss": 0.2611,
+      "step": 1611
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2563548725286343e-05,
+      "loss": 0.2536,
+      "step": 1612
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2555096287446599e-05,
+      "loss": 0.2843,
+      "step": 1613
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2546641896200686e-05,
+      "loss": 0.2626,
+      "step": 1614
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2538185558012111e-05,
+      "loss": 0.2435,
+      "step": 1615
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2529727279345852e-05,
+      "loss": 0.2454,
+      "step": 1616
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.252126706666838e-05,
+      "loss": 0.2765,
+      "step": 1617
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2512804926447649e-05,
+      "loss": 0.2386,
+      "step": 1618
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2504340865153077e-05,
+      "loss": 0.2556,
+      "step": 1619
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.249587488925555e-05,
+      "loss": 0.2578,
+      "step": 1620
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2487407005227436e-05,
+      "loss": 0.2252,
+      "step": 1621
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2478937219542537e-05,
+      "loss": 0.2659,
+      "step": 1622
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2470465538676125e-05,
+      "loss": 0.2741,
+      "step": 1623
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2461991969104914e-05,
+      "loss": 0.2641,
+      "step": 1624
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2453516517307067e-05,
+      "loss": 0.273,
+      "step": 1625
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2445039189762182e-05,
+      "loss": 0.2489,
+      "step": 1626
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2436559992951296e-05,
+      "loss": 0.2585,
+      "step": 1627
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2428078933356862e-05,
+      "loss": 0.2549,
+      "step": 1628
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2419596017462775e-05,
+      "loss": 0.2599,
+      "step": 1629
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2411111251754334e-05,
+      "loss": 0.2533,
+      "step": 1630
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2402624642718264e-05,
+      "loss": 0.244,
+      "step": 1631
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2394136196842692e-05,
+      "loss": 0.2488,
+      "step": 1632
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2385645920617151e-05,
+      "loss": 0.2549,
+      "step": 1633
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2377153820532575e-05,
+      "loss": 0.259,
+      "step": 1634
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.236865990308129e-05,
+      "loss": 0.231,
+      "step": 1635
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2360164174757014e-05,
+      "loss": 0.2546,
+      "step": 1636
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2351666642054844e-05,
+      "loss": 0.2649,
+      "step": 1637
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2343167311471267e-05,
+      "loss": 0.2478,
+      "step": 1638
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.233466618950413e-05,
+      "loss": 0.2183,
+      "step": 1639
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2326163282652663e-05,
+      "loss": 0.2709,
+      "step": 1640
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.231765859741745e-05,
+      "loss": 0.2465,
+      "step": 1641
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2309152140300446e-05,
+      "loss": 0.2435,
+      "step": 1642
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2300643917804949e-05,
+      "loss": 0.2619,
+      "step": 1643
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2292133936435615e-05,
+      "loss": 0.243,
+      "step": 1644
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2283622202698439e-05,
+      "loss": 0.2467,
+      "step": 1645
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2275108723100759e-05,
+      "loss": 0.2778,
+      "step": 1646
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2266593504151247e-05,
+      "loss": 0.267,
+      "step": 1647
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2258076552359905e-05,
+      "loss": 0.2438,
+      "step": 1648
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2249557874238058e-05,
+      "loss": 0.2727,
+      "step": 1649
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2241037476298359e-05,
+      "loss": 0.2552,
+      "step": 1650
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.223251536505476e-05,
+      "loss": 0.2749,
+      "step": 1651
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2223991547022536e-05,
+      "loss": 0.2523,
+      "step": 1652
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2215466028718261e-05,
+      "loss": 0.2611,
+      "step": 1653
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2206938816659815e-05,
+      "loss": 0.2486,
+      "step": 1654
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2198409917366365e-05,
+      "loss": 0.2603,
+      "step": 1655
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2189879337358372e-05,
+      "loss": 0.2454,
+      "step": 1656
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2181347083157582e-05,
+      "loss": 0.2507,
+      "step": 1657
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.217281316128702e-05,
+      "loss": 0.2613,
+      "step": 1658
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2164277578270989e-05,
+      "loss": 0.2361,
+      "step": 1659
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2155740340635054e-05,
+      "loss": 0.2578,
+      "step": 1660
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2147201454906055e-05,
+      "loss": 0.2432,
+      "step": 1661
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2138660927612084e-05,
+      "loss": 0.2762,
+      "step": 1662
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2130118765282493e-05,
+      "loss": 0.2542,
+      "step": 1663
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2121574974447885e-05,
+      "loss": 0.2549,
+      "step": 1664
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.21130295616401e-05,
+      "loss": 0.257,
+      "step": 1665
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2104482533392225e-05,
+      "loss": 0.2835,
+      "step": 1666
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2095933896238583e-05,
+      "loss": 0.2742,
+      "step": 1667
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2087383656714721e-05,
+      "loss": 0.2522,
+      "step": 1668
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2078831821357417e-05,
+      "loss": 0.255,
+      "step": 1669
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2070278396704667e-05,
+      "loss": 0.2473,
+      "step": 1670
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2061723389295676e-05,
+      "loss": 0.2329,
+      "step": 1671
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.205316680567087e-05,
+      "loss": 0.2183,
+      "step": 1672
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2044608652371874e-05,
+      "loss": 0.2471,
+      "step": 1673
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.203604893594151e-05,
+      "loss": 0.2231,
+      "step": 1674
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2027487662923805e-05,
+      "loss": 0.2516,
+      "step": 1675
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2018924839863965e-05,
+      "loss": 0.2564,
+      "step": 1676
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2010360473308384e-05,
+      "loss": 0.2651,
+      "step": 1677
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2001794569804643e-05,
+      "loss": 0.2685,
+      "step": 1678
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1993227135901491e-05,
+      "loss": 0.2555,
+      "step": 1679
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1984658178148844e-05,
+      "loss": 0.2518,
+      "step": 1680
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1976087703097794e-05,
+      "loss": 0.2527,
+      "step": 1681
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1967515717300584e-05,
+      "loss": 0.2686,
+      "step": 1682
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1958942227310615e-05,
+      "loss": 0.2441,
+      "step": 1683
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1950367239682436e-05,
+      "loss": 0.2561,
+      "step": 1684
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1941790760971748e-05,
+      "loss": 0.241,
+      "step": 1685
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1933212797735381e-05,
+      "loss": 0.2555,
+      "step": 1686
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.192463335653131e-05,
+      "loss": 0.2551,
+      "step": 1687
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.191605244391863e-05,
+      "loss": 0.2587,
+      "step": 1688
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1907470066457572e-05,
+      "loss": 0.2768,
+      "step": 1689
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1898886230709476e-05,
+      "loss": 0.2408,
+      "step": 1690
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1890300943236808e-05,
+      "loss": 0.2664,
+      "step": 1691
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1881714210603135e-05,
+      "loss": 0.224,
+      "step": 1692
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1873126039373133e-05,
+      "loss": 0.2425,
+      "step": 1693
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1864536436112573e-05,
+      "loss": 0.2535,
+      "step": 1694
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.185594540738833e-05,
+      "loss": 0.2412,
+      "step": 1695
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1847352959768358e-05,
+      "loss": 0.2557,
+      "step": 1696
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1838759099821707e-05,
+      "loss": 0.2799,
+      "step": 1697
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1830163834118496e-05,
+      "loss": 0.2636,
+      "step": 1698
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1821567169229926e-05,
+      "loss": 0.2612,
+      "step": 1699
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1812969111728263e-05,
+      "loss": 0.2468,
+      "step": 1700
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1804369668186843e-05,
+      "loss": 0.2653,
+      "step": 1701
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1795768845180055e-05,
+      "loss": 0.2711,
+      "step": 1702
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.178716664928335e-05,
+      "loss": 0.2287,
+      "step": 1703
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1778563087073223e-05,
+      "loss": 0.2531,
+      "step": 1704
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1769958165127214e-05,
+      "loss": 0.2723,
+      "step": 1705
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1761351890023904e-05,
+      "loss": 0.2304,
+      "step": 1706
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1752744268342913e-05,
+      "loss": 0.231,
+      "step": 1707
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1744135306664876e-05,
+      "loss": 0.2407,
+      "step": 1708
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1735525011571476e-05,
+      "loss": 0.263,
+      "step": 1709
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1726913389645391e-05,
+      "loss": 0.2317,
+      "step": 1710
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1718300447470327e-05,
+      "loss": 0.2858,
+      "step": 1711
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1709686191630998e-05,
+      "loss": 0.281,
+      "step": 1712
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.170107062871312e-05,
+      "loss": 0.2588,
+      "step": 1713
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1692453765303406e-05,
+      "loss": 0.247,
+      "step": 1714
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1683835607989576e-05,
+      "loss": 0.2392,
+      "step": 1715
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1675216163360316e-05,
+      "loss": 0.2638,
+      "step": 1716
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1666595438005317e-05,
+      "loss": 0.2732,
+      "step": 1717
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1657973438515242e-05,
+      "loss": 0.2523,
+      "step": 1718
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1649350171481728e-05,
+      "loss": 0.2598,
+      "step": 1719
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.164072564349738e-05,
+      "loss": 0.2719,
+      "step": 1720
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1632099861155767e-05,
+      "loss": 0.2412,
+      "step": 1721
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1623472831051419e-05,
+      "loss": 0.2449,
+      "step": 1722
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1614844559779819e-05,
+      "loss": 0.269,
+      "step": 1723
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1606215053937398e-05,
+      "loss": 0.2415,
+      "step": 1724
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.159758432012153e-05,
+      "loss": 0.2537,
+      "step": 1725
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1588952364930534e-05,
+      "loss": 0.2417,
+      "step": 1726
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1580319194963656e-05,
+      "loss": 0.2309,
+      "step": 1727
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1571684816821072e-05,
+      "loss": 0.2455,
+      "step": 1728
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1563049237103883e-05,
+      "loss": 0.234,
+      "step": 1729
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1554412462414109e-05,
+      "loss": 0.2629,
+      "step": 1730
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.154577449935468e-05,
+      "loss": 0.2457,
+      "step": 1731
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.153713535452944e-05,
+      "loss": 0.2519,
+      "step": 1732
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1528495034543133e-05,
+      "loss": 0.243,
+      "step": 1733
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1519853546001403e-05,
+      "loss": 0.2523,
+      "step": 1734
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1511210895510782e-05,
+      "loss": 0.2344,
+      "step": 1735
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1502567089678704e-05,
+      "loss": 0.2458,
+      "step": 1736
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1493922135113467e-05,
+      "loss": 0.2708,
+      "step": 1737
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1485276038424267e-05,
+      "loss": 0.2584,
+      "step": 1738
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1476628806221154e-05,
+      "loss": 0.2484,
+      "step": 1739
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1467980445115063e-05,
+      "loss": 0.238,
+      "step": 1740
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1459330961717782e-05,
+      "loss": 0.2352,
+      "step": 1741
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.145068036264196e-05,
+      "loss": 0.2462,
+      "step": 1742
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1442028654501096e-05,
+      "loss": 0.2605,
+      "step": 1743
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1433375843909547e-05,
+      "loss": 0.2602,
+      "step": 1744
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1424721937482495e-05,
+      "loss": 0.2487,
+      "step": 1745
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1416066941835978e-05,
+      "loss": 0.2387,
+      "step": 1746
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1407410863586854e-05,
+      "loss": 0.2581,
+      "step": 1747
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1398753709352819e-05,
+      "loss": 0.2414,
+      "step": 1748
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1390095485752381e-05,
+      "loss": 0.2728,
+      "step": 1749
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1381436199404876e-05,
+      "loss": 0.2606,
+      "step": 1750
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.137277585693044e-05,
+      "loss": 0.2539,
+      "step": 1751
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.136411446495003e-05,
+      "loss": 0.2404,
+      "step": 1752
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.13554520300854e-05,
+      "loss": 0.2549,
+      "step": 1753
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1346788558959092e-05,
+      "loss": 0.2504,
+      "step": 1754
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1338124058194455e-05,
+      "loss": 0.2603,
+      "step": 1755
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.132945853441562e-05,
+      "loss": 0.2401,
+      "step": 1756
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.132079199424749e-05,
+      "loss": 0.2653,
+      "step": 1757
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1312124444315766e-05,
+      "loss": 0.2539,
+      "step": 1758
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1303455891246901e-05,
+      "loss": 0.2411,
+      "step": 1759
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1294786341668125e-05,
+      "loss": 0.2536,
+      "step": 1760
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1286115802207426e-05,
+      "loss": 0.2523,
+      "step": 1761
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1277444279493553e-05,
+      "loss": 0.2344,
+      "step": 1762
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1268771780156e-05,
+      "loss": 0.2525,
+      "step": 1763
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1260098310825019e-05,
+      "loss": 0.2711,
+      "step": 1764
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1251423878131586e-05,
+      "loss": 0.2512,
+      "step": 1765
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1242748488707436e-05,
+      "loss": 0.2382,
+      "step": 1766
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1234072149185011e-05,
+      "loss": 0.2601,
+      "step": 1767
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.12253948661975e-05,
+      "loss": 0.2503,
+      "step": 1768
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.12167166463788e-05,
+      "loss": 0.2621,
+      "step": 1769
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1208037496363534e-05,
+      "loss": 0.2519,
+      "step": 1770
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1199357422787026e-05,
+      "loss": 0.2491,
+      "step": 1771
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1190676432285317e-05,
+      "loss": 0.2617,
+      "step": 1772
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.118199453149514e-05,
+      "loss": 0.2651,
+      "step": 1773
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1173311727053924e-05,
+      "loss": 0.2457,
+      "step": 1774
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1164628025599797e-05,
+      "loss": 0.2793,
+      "step": 1775
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1155943433771568e-05,
+      "loss": 0.2591,
+      "step": 1776
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1147257958208723e-05,
+      "loss": 0.2571,
+      "step": 1777
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1138571605551434e-05,
+      "loss": 0.2266,
+      "step": 1778
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1129884382440525e-05,
+      "loss": 0.2655,
+      "step": 1779
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1121196295517512e-05,
+      "loss": 0.2707,
+      "step": 1780
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1112507351424544e-05,
+      "loss": 0.2443,
+      "step": 1781
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1103817556804443e-05,
+      "loss": 0.2538,
+      "step": 1782
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1095126918300676e-05,
+      "loss": 0.2298,
+      "step": 1783
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1086435442557358e-05,
+      "loss": 0.2665,
+      "step": 1784
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1077743136219237e-05,
+      "loss": 0.2451,
+      "step": 1785
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1069050005931703e-05,
+      "loss": 0.2782,
+      "step": 1786
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1060356058340776e-05,
+      "loss": 0.2612,
+      "step": 1787
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1051661300093088e-05,
+      "loss": 0.2627,
+      "step": 1788
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1042965737835913e-05,
+      "loss": 0.2504,
+      "step": 1789
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1034269378217122e-05,
+      "loss": 0.2455,
+      "step": 1790
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1025572227885202e-05,
+      "loss": 0.2452,
+      "step": 1791
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1016874293489244e-05,
+      "loss": 0.2424,
+      "step": 1792
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1008175581678937e-05,
+      "loss": 0.2701,
+      "step": 1793
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0999476099104564e-05,
+      "loss": 0.2503,
+      "step": 1794
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0990775852417005e-05,
+      "loss": 0.2542,
+      "step": 1795
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0982074848267711e-05,
+      "loss": 0.2422,
+      "step": 1796
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0973373093308726e-05,
+      "loss": 0.233,
+      "step": 1797
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0964670594192652e-05,
+      "loss": 0.2516,
+      "step": 1798
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0955967357572676e-05,
+      "loss": 0.2544,
+      "step": 1799
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0947263390102543e-05,
+      "loss": 0.2655,
+      "step": 1800
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.093855869843655e-05,
+      "loss": 0.2595,
+      "step": 1801
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0929853289229552e-05,
+      "loss": 0.2364,
+      "step": 1802
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0921147169136965e-05,
+      "loss": 0.2679,
+      "step": 1803
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0912440344814726e-05,
+      "loss": 0.2532,
+      "step": 1804
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0903732822919325e-05,
+      "loss": 0.2533,
+      "step": 1805
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.089502461010778e-05,
+      "loss": 0.2421,
+      "step": 1806
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0886315713037648e-05,
+      "loss": 0.2471,
+      "step": 1807
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0877606138366985e-05,
+      "loss": 0.2396,
+      "step": 1808
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0868895892754397e-05,
+      "loss": 0.2449,
+      "step": 1809
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0860184982858974e-05,
+      "loss": 0.2547,
+      "step": 1810
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.085147341534033e-05,
+      "loss": 0.2675,
+      "step": 1811
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.084276119685858e-05,
+      "loss": 0.2692,
+      "step": 1812
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0834048334074332e-05,
+      "loss": 0.2444,
+      "step": 1813
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.082533483364869e-05,
+      "loss": 0.2518,
+      "step": 1814
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0816620702243248e-05,
+      "loss": 0.2454,
+      "step": 1815
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0807905946520073e-05,
+      "loss": 0.2553,
+      "step": 1816
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0799190573141716e-05,
+      "loss": 0.2737,
+      "step": 1817
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0790474588771202e-05,
+      "loss": 0.2827,
+      "step": 1818
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0781758000072022e-05,
+      "loss": 0.2655,
+      "step": 1819
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0773040813708124e-05,
+      "loss": 0.2662,
+      "step": 1820
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0764323036343921e-05,
+      "loss": 0.2518,
+      "step": 1821
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0755604674644267e-05,
+      "loss": 0.2267,
+      "step": 1822
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0746885735274478e-05,
+      "loss": 0.2567,
+      "step": 1823
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0738166224900292e-05,
+      "loss": 0.2453,
+      "step": 1824
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0729446150187903e-05,
+      "loss": 0.2387,
+      "step": 1825
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0720725517803925e-05,
+      "loss": 0.2634,
+      "step": 1826
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.07120043344154e-05,
+      "loss": 0.2802,
+      "step": 1827
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0703282606689792e-05,
+      "loss": 0.2597,
+      "step": 1828
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0694560341294987e-05,
+      "loss": 0.2498,
+      "step": 1829
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0685837544899266e-05,
+      "loss": 0.2613,
+      "step": 1830
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0677114224171335e-05,
+      "loss": 0.2394,
+      "step": 1831
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0668390385780292e-05,
+      "loss": 0.2539,
+      "step": 1832
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0659666036395626e-05,
+      "loss": 0.2291,
+      "step": 1833
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0650941182687226e-05,
+      "loss": 0.2399,
+      "step": 1834
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0642215831325362e-05,
+      "loss": 0.2457,
+      "step": 1835
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0633489988980682e-05,
+      "loss": 0.2443,
+      "step": 1836
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0624763662324217e-05,
+      "loss": 0.2566,
+      "step": 1837
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0616036858027364e-05,
+      "loss": 0.2503,
+      "step": 1838
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0607309582761878e-05,
+      "loss": 0.2716,
+      "step": 1839
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0598581843199883e-05,
+      "loss": 0.2701,
+      "step": 1840
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0589853646013856e-05,
+      "loss": 0.2574,
+      "step": 1841
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0581124997876627e-05,
+      "loss": 0.2501,
+      "step": 1842
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0572395905461363e-05,
+      "loss": 0.2564,
+      "step": 1843
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.056366637544158e-05,
+      "loss": 0.2445,
+      "step": 1844
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0554936414491112e-05,
+      "loss": 0.2436,
+      "step": 1845
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0546206029284149e-05,
+      "loss": 0.2554,
+      "step": 1846
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0537475226495179e-05,
+      "loss": 0.2663,
+      "step": 1847
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0528744012799025e-05,
+      "loss": 0.2452,
+      "step": 1848
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0520012394870823e-05,
+      "loss": 0.261,
+      "step": 1849
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0511280379386009e-05,
+      "loss": 0.2294,
+      "step": 1850
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0502547973020328e-05,
+      "loss": 0.2459,
+      "step": 1851
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0493815182449828e-05,
+      "loss": 0.239,
+      "step": 1852
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0485082014350847e-05,
+      "loss": 0.2467,
+      "step": 1853
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0476348475400011e-05,
+      "loss": 0.2532,
+      "step": 1854
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0467614572274229e-05,
+      "loss": 0.2585,
+      "step": 1855
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0458880311650692e-05,
+      "loss": 0.2508,
+      "step": 1856
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0450145700206862e-05,
+      "loss": 0.2617,
+      "step": 1857
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.044141074462047e-05,
+      "loss": 0.2529,
+      "step": 1858
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0432675451569503e-05,
+      "loss": 0.2569,
+      "step": 1859
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0423939827732221e-05,
+      "loss": 0.2635,
+      "step": 1860
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0415203879787123e-05,
+      "loss": 0.2449,
+      "step": 1861
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0406467614412966e-05,
+      "loss": 0.2703,
+      "step": 1862
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0397731038288737e-05,
+      "loss": 0.2534,
+      "step": 1863
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0388994158093675e-05,
+      "loss": 0.2358,
+      "step": 1864
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.038025698050724e-05,
+      "loss": 0.2626,
+      "step": 1865
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.037151951220913e-05,
+      "loss": 0.2624,
+      "step": 1866
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0362781759879252e-05,
+      "loss": 0.2474,
+      "step": 1867
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0354043730197744e-05,
+      "loss": 0.2693,
+      "step": 1868
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0345305429844946e-05,
+      "loss": 0.2678,
+      "step": 1869
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.033656686550141e-05,
+      "loss": 0.2745,
+      "step": 1870
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0327828043847887e-05,
+      "loss": 0.2736,
+      "step": 1871
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.031908897156533e-05,
+      "loss": 0.2607,
+      "step": 1872
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0310349655334872e-05,
+      "loss": 0.2481,
+      "step": 1873
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0301610101837848e-05,
+      "loss": 0.2495,
+      "step": 1874
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0292870317755763e-05,
+      "loss": 0.2788,
+      "step": 1875
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0284130309770302e-05,
+      "loss": 0.247,
+      "step": 1876
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.027539008456332e-05,
+      "loss": 0.2338,
+      "step": 1877
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0266649648816843e-05,
+      "loss": 0.2641,
+      "step": 1878
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0257909009213045e-05,
+      "loss": 0.2468,
+      "step": 1879
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0249168172434276e-05,
+      "loss": 0.2597,
+      "step": 1880
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0240427145163019e-05,
+      "loss": 0.2711,
+      "step": 1881
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0231685934081911e-05,
+      "loss": 0.226,
+      "step": 1882
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0222944545873729e-05,
+      "loss": 0.2495,
+      "step": 1883
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0214202987221383e-05,
+      "loss": 0.2607,
+      "step": 1884
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0205461264807914e-05,
+      "loss": 0.2575,
+      "step": 1885
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.019671938531649e-05,
+      "loss": 0.2781,
+      "step": 1886
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0187977355430397e-05,
+      "loss": 0.2476,
+      "step": 1887
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0179235181833034e-05,
+      "loss": 0.2602,
+      "step": 1888
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.017049287120792e-05,
+      "loss": 0.2722,
+      "step": 1889
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0161750430238664e-05,
+      "loss": 0.2417,
+      "step": 1890
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0153007865608986e-05,
+      "loss": 0.2484,
+      "step": 1891
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0144265184002695e-05,
+      "loss": 0.2406,
+      "step": 1892
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0135522392103691e-05,
+      "loss": 0.268,
+      "step": 1893
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.012677949659596e-05,
+      "loss": 0.257,
+      "step": 1894
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0118036504163567e-05,
+      "loss": 0.2622,
+      "step": 1895
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0109293421490641e-05,
+      "loss": 0.274,
+      "step": 1896
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0100550255261401e-05,
+      "loss": 0.264,
+      "step": 1897
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0091807012160106e-05,
+      "loss": 0.2577,
+      "step": 1898
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0083063698871093e-05,
+      "loss": 0.2166,
+      "step": 1899
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0074320322078743e-05,
+      "loss": 0.2489,
+      "step": 1900
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.006557688846749e-05,
+      "loss": 0.2428,
+      "step": 1901
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0056833404721802e-05,
+      "loss": 0.2764,
+      "step": 1902
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0048089877526199e-05,
+      "loss": 0.2529,
+      "step": 1903
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0039346313565224e-05,
+      "loss": 0.2695,
+      "step": 1904
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.003060271952345e-05,
+      "loss": 0.2544,
+      "step": 1905
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0021859102085479e-05,
+      "loss": 0.258,
+      "step": 1906
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0013115467935925e-05,
+      "loss": 0.2624,
+      "step": 1907
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0004371823759415e-05,
+      "loss": 0.2606,
+      "step": 1908
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.995628176240588e-06,
+      "loss": 0.2702,
+      "step": 1909
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.986884532064078e-06,
+      "loss": 0.2492,
+      "step": 1910
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.978140897914521e-06,
+      "loss": 0.2425,
+      "step": 1911
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.969397280476553e-06,
+      "loss": 0.261,
+      "step": 1912
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.96065368643478e-06,
+      "loss": 0.2775,
+      "step": 1913
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.951910122473806e-06,
+      "loss": 0.2666,
+      "step": 1914
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.9431665952782e-06,
+      "loss": 0.2562,
+      "step": 1915
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.934423111532513e-06,
+      "loss": 0.239,
+      "step": 1916
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.925679677921259e-06,
+      "loss": 0.2358,
+      "step": 1917
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.916936301128908e-06,
+      "loss": 0.2285,
+      "step": 1918
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.908192987839894e-06,
+      "loss": 0.2532,
+      "step": 1919
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.899449744738602e-06,
+      "loss": 0.2346,
+      "step": 1920
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.890706578509362e-06,
+      "loss": 0.2574,
+      "step": 1921
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.88196349583644e-06,
+      "loss": 0.2499,
+      "step": 1922
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.873220503404042e-06,
+      "loss": 0.2854,
+      "step": 1923
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.864477607896312e-06,
+      "loss": 0.2677,
+      "step": 1924
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.855734815997309e-06,
+      "loss": 0.2714,
+      "step": 1925
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.84699213439102e-06,
+      "loss": 0.2566,
+      "step": 1926
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.83824956976134e-06,
+      "loss": 0.2473,
+      "step": 1927
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.829507128792083e-06,
+      "loss": 0.2544,
+      "step": 1928
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.82076481816697e-06,
+      "loss": 0.2432,
+      "step": 1929
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.812022644569608e-06,
+      "loss": 0.248,
+      "step": 1930
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.803280614683512e-06,
+      "loss": 0.2629,
+      "step": 1931
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.794538735192089e-06,
+      "loss": 0.2344,
+      "step": 1932
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.785797012778619e-06,
+      "loss": 0.2491,
+      "step": 1933
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.777055454126271e-06,
+      "loss": 0.2442,
+      "step": 1934
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.76831406591809e-06,
+      "loss": 0.2358,
+      "step": 1935
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.759572854836983e-06,
+      "loss": 0.27,
+      "step": 1936
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.75083182756573e-06,
+      "loss": 0.2773,
+      "step": 1937
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.742090990786956e-06,
+      "loss": 0.2459,
+      "step": 1938
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.733350351183162e-06,
+      "loss": 0.2374,
+      "step": 1939
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.724609915436685e-06,
+      "loss": 0.2643,
+      "step": 1940
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.715869690229703e-06,
+      "loss": 0.2291,
+      "step": 1941
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.707129682244239e-06,
+      "loss": 0.2542,
+      "step": 1942
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.698389898162153e-06,
+      "loss": 0.2624,
+      "step": 1943
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.689650344665131e-06,
+      "loss": 0.2353,
+      "step": 1944
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.680911028434675e-06,
+      "loss": 0.2474,
+      "step": 1945
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.672171956152113e-06,
+      "loss": 0.2421,
+      "step": 1946
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.663433134498593e-06,
+      "loss": 0.2497,
+      "step": 1947
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.654694570155056e-06,
+      "loss": 0.2683,
+      "step": 1948
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.64595626980226e-06,
+      "loss": 0.2447,
+      "step": 1949
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.63721824012075e-06,
+      "loss": 0.2421,
+      "step": 1950
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.628480487790873e-06,
+      "loss": 0.2356,
+      "step": 1951
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.619743019492764e-06,
+      "loss": 0.2636,
+      "step": 1952
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.61100584190633e-06,
+      "loss": 0.2543,
+      "step": 1953
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.602268961711264e-06,
+      "loss": 0.2401,
+      "step": 1954
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.59353238558704e-06,
+      "loss": 0.2532,
+      "step": 1955
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.584796120212878e-06,
+      "loss": 0.23,
+      "step": 1956
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.576060172267784e-06,
+      "loss": 0.2399,
+      "step": 1957
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.567324548430497e-06,
+      "loss": 0.2739,
+      "step": 1958
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.558589255379534e-06,
+      "loss": 0.2619,
+      "step": 1959
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.549854299793141e-06,
+      "loss": 0.2539,
+      "step": 1960
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.54111968834931e-06,
+      "loss": 0.2505,
+      "step": 1961
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.532385427725771e-06,
+      "loss": 0.2366,
+      "step": 1962
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.523651524599992e-06,
+      "loss": 0.236,
+      "step": 1963
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.514917985649156e-06,
+      "loss": 0.2537,
+      "step": 1964
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.506184817550177e-06,
+      "loss": 0.2454,
+      "step": 1965
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.497452026979674e-06,
+      "loss": 0.2413,
+      "step": 1966
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.488719620613995e-06,
+      "loss": 0.2467,
+      "step": 1967
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.479987605129182e-06,
+      "loss": 0.2703,
+      "step": 1968
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.47125598720098e-06,
+      "loss": 0.2456,
+      "step": 1969
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.462524773504823e-06,
+      "loss": 0.2284,
+      "step": 1970
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.453793970715855e-06,
+      "loss": 0.2299,
+      "step": 1971
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.44506358550889e-06,
+      "loss": 0.2538,
+      "step": 1972
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.436333624558427e-06,
+      "loss": 0.2527,
+      "step": 1973
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.427604094538638e-06,
+      "loss": 0.2422,
+      "step": 1974
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.418875002123375e-06,
+      "loss": 0.2368,
+      "step": 1975
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.410146353986145e-06,
+      "loss": 0.2389,
+      "step": 1976
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.401418156800122e-06,
+      "loss": 0.2542,
+      "step": 1977
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.392690417238127e-06,
+      "loss": 0.2592,
+      "step": 1978
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.38396314197264e-06,
+      "loss": 0.2335,
+      "step": 1979
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.375236337675785e-06,
+      "loss": 0.2671,
+      "step": 1980
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.366510011019319e-06,
+      "loss": 0.2241,
+      "step": 1981
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.357784168674638e-06,
+      "loss": 0.2615,
+      "step": 1982
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.349058817312775e-06,
+      "loss": 0.2538,
+      "step": 1983
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.340333963604377e-06,
+      "loss": 0.2413,
+      "step": 1984
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.33160961421971e-06,
+      "loss": 0.2474,
+      "step": 1985
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.322885775828667e-06,
+      "loss": 0.238,
+      "step": 1986
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.314162455100737e-06,
+      "loss": 0.251,
+      "step": 1987
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.30543965870502e-06,
+      "loss": 0.2469,
+      "step": 1988
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.29671739331021e-06,
+      "loss": 0.2551,
+      "step": 1989
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.287995665584601e-06,
+      "loss": 0.278,
+      "step": 1990
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.279274482196079e-06,
+      "loss": 0.2627,
+      "step": 1991
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.270553849812102e-06,
+      "loss": 0.244,
+      "step": 1992
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.26183377509971e-06,
+      "loss": 0.2504,
+      "step": 1993
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.253114264725526e-06,
+      "loss": 0.2401,
+      "step": 1994
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.244395325355735e-06,
+      "loss": 0.2415,
+      "step": 1995
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.235676963656082e-06,
+      "loss": 0.2421,
+      "step": 1996
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.226959186291876e-06,
+      "loss": 0.2506,
+      "step": 1997
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.218241999927981e-06,
+      "loss": 0.2532,
+      "step": 1998
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.209525411228801e-06,
+      "loss": 0.2394,
+      "step": 1999
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.20080942685829e-06,
+      "loss": 0.2393,
+      "step": 2000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.19209405347993e-06,
+      "loss": 0.2513,
+      "step": 2001
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.183379297756755e-06,
+      "loss": 0.271,
+      "step": 2002
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.174665166351312e-06,
+      "loss": 0.2465,
+      "step": 2003
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.165951665925671e-06,
+      "loss": 0.2688,
+      "step": 2004
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.157238803141421e-06,
+      "loss": 0.2638,
+      "step": 2005
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.148526584659672e-06,
+      "loss": 0.241,
+      "step": 2006
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.13981501714103e-06,
+      "loss": 0.2543,
+      "step": 2007
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.131104107245609e-06,
+      "loss": 0.2769,
+      "step": 2008
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.122393861633015e-06,
+      "loss": 0.241,
+      "step": 2009
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.113684286962357e-06,
+      "loss": 0.2679,
+      "step": 2010
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.104975389892222e-06,
+      "loss": 0.2392,
+      "step": 2011
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.096267177080679e-06,
+      "loss": 0.2683,
+      "step": 2012
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.087559655185276e-06,
+      "loss": 0.255,
+      "step": 2013
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.078852830863038e-06,
+      "loss": 0.2508,
+      "step": 2014
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.07014671077045e-06,
+      "loss": 0.2502,
+      "step": 2015
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.061441301563455e-06,
+      "loss": 0.2337,
+      "step": 2016
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.052736609897458e-06,
+      "loss": 0.2656,
+      "step": 2017
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.044032642427325e-06,
+      "loss": 0.2704,
+      "step": 2018
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.03532940580735e-06,
+      "loss": 0.2391,
+      "step": 2019
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.02662690669128e-06,
+      "loss": 0.2431,
+      "step": 2020
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.01792515173229e-06,
+      "loss": 0.252,
+      "step": 2021
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.009224147582998e-06,
+      "loss": 0.2561,
+      "step": 2022
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.000523900895438e-06,
+      "loss": 0.2462,
+      "step": 2023
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.991824418321068e-06,
+      "loss": 0.2363,
+      "step": 2024
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.983125706510758e-06,
+      "loss": 0.2304,
+      "step": 2025
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.974427772114801e-06,
+      "loss": 0.2316,
+      "step": 2026
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.96573062178288e-06,
+      "loss": 0.224,
+      "step": 2027
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.95703426216409e-06,
+      "loss": 0.2664,
+      "step": 2028
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.948338699906912e-06,
+      "loss": 0.2605,
+      "step": 2029
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.939643941659229e-06,
+      "loss": 0.2407,
+      "step": 2030
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.930949994068299e-06,
+      "loss": 0.2502,
+      "step": 2031
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.922256863780766e-06,
+      "loss": 0.2628,
+      "step": 2032
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.913564557442644e-06,
+      "loss": 0.2494,
+      "step": 2033
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.904873081699326e-06,
+      "loss": 0.2464,
+      "step": 2034
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.89618244319556e-06,
+      "loss": 0.2425,
+      "step": 2035
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.887492648575458e-06,
+      "loss": 0.2391,
+      "step": 2036
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.878803704482493e-06,
+      "loss": 0.2413,
+      "step": 2037
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.870115617559476e-06,
+      "loss": 0.2272,
+      "step": 2038
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.86142839444857e-06,
+      "loss": 0.2447,
+      "step": 2039
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.852742041791276e-06,
+      "loss": 0.273,
+      "step": 2040
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.844056566228435e-06,
+      "loss": 0.2563,
+      "step": 2041
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.835371974400205e-06,
+      "loss": 0.2725,
+      "step": 2042
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.82668827294608e-06,
+      "loss": 0.2519,
+      "step": 2043
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.818005468504863e-06,
+      "loss": 0.2498,
+      "step": 2044
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.809323567714685e-06,
+      "loss": 0.2733,
+      "step": 2045
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.800642577212977e-06,
+      "loss": 0.2565,
+      "step": 2046
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.791962503636469e-06,
+      "loss": 0.2592,
+      "step": 2047
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.7832833536212e-06,
+      "loss": 0.2714,
+      "step": 2048
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.774605133802502e-06,
+      "loss": 0.2351,
+      "step": 2049
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.76592785081499e-06,
+      "loss": 0.2563,
+      "step": 2050
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.757251511292571e-06,
+      "loss": 0.2634,
+      "step": 2051
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.748576121868414e-06,
+      "loss": 0.2376,
+      "step": 2052
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.739901689174983e-06,
+      "loss": 0.256,
+      "step": 2053
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.731228219844002e-06,
+      "loss": 0.2385,
+      "step": 2054
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.72255572050645e-06,
+      "loss": 0.2405,
+      "step": 2055
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.713884197792575e-06,
+      "loss": 0.2596,
+      "step": 2056
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.705213658331879e-06,
+      "loss": 0.2568,
+      "step": 2057
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.6965441087531e-06,
+      "loss": 0.2462,
+      "step": 2058
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.687875555684237e-06,
+      "loss": 0.2537,
+      "step": 2059
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.679208005752509e-06,
+      "loss": 0.2442,
+      "step": 2060
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.670541465584383e-06,
+      "loss": 0.2683,
+      "step": 2061
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.661875941805547e-06,
+      "loss": 0.2604,
+      "step": 2062
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.653211441040913e-06,
+      "loss": 0.2576,
+      "step": 2063
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.644547969914606e-06,
+      "loss": 0.2325,
+      "step": 2064
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.635885535049971e-06,
+      "loss": 0.2583,
+      "step": 2065
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.627224143069564e-06,
+      "loss": 0.2437,
+      "step": 2066
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.61856380059513e-06,
+      "loss": 0.248,
+      "step": 2067
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.60990451424762e-06,
+      "loss": 0.2512,
+      "step": 2068
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.601246290647185e-06,
+      "loss": 0.251,
+      "step": 2069
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.592589136413148e-06,
+      "loss": 0.2306,
+      "step": 2070
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.583933058164027e-06,
+      "loss": 0.263,
+      "step": 2071
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.575278062517508e-06,
+      "loss": 0.2534,
+      "step": 2072
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.566624156090457e-06,
+      "loss": 0.2324,
+      "step": 2073
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.557971345498907e-06,
+      "loss": 0.2422,
+      "step": 2074
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.549319637358044e-06,
+      "loss": 0.2416,
+      "step": 2075
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.54066903828222e-06,
+      "loss": 0.2518,
+      "step": 2076
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.532019554884939e-06,
+      "loss": 0.2712,
+      "step": 2077
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.523371193778847e-06,
+      "loss": 0.2464,
+      "step": 2078
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.514723961575738e-06,
+      "loss": 0.2378,
+      "step": 2079
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.506077864886533e-06,
+      "loss": 0.2544,
+      "step": 2080
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.497432910321298e-06,
+      "loss": 0.2422,
+      "step": 2081
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.48878910448922e-06,
+      "loss": 0.2564,
+      "step": 2082
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.4801464539986e-06,
+      "loss": 0.253,
+      "step": 2083
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.471504965456869e-06,
+      "loss": 0.2366,
+      "step": 2084
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.462864645470563e-06,
+      "loss": 0.2603,
+      "step": 2085
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.454225500645325e-06,
+      "loss": 0.2557,
+      "step": 2086
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.445587537585896e-06,
+      "loss": 0.2651,
+      "step": 2087
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.43695076289612e-06,
+      "loss": 0.2545,
+      "step": 2088
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.428315183178931e-06,
+      "loss": 0.2546,
+      "step": 2089
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.419680805036348e-06,
+      "loss": 0.2579,
+      "step": 2090
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.411047635069467e-06,
+      "loss": 0.2469,
+      "step": 2091
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.402415679878471e-06,
+      "loss": 0.2389,
+      "step": 2092
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.393784946062605e-06,
+      "loss": 0.254,
+      "step": 2093
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.385155440220186e-06,
+      "loss": 0.2487,
+      "step": 2094
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.376527168948584e-06,
+      "loss": 0.2419,
+      "step": 2095
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.367900138844235e-06,
+      "loss": 0.235,
+      "step": 2096
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.359274356502625e-06,
+      "loss": 0.2112,
+      "step": 2097
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.350649828518274e-06,
+      "loss": 0.2493,
+      "step": 2098
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.342026561484758e-06,
+      "loss": 0.2459,
+      "step": 2099
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.333404561994685e-06,
+      "loss": 0.2349,
+      "step": 2100
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.324783836639687e-06,
+      "loss": 0.2301,
+      "step": 2101
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.316164392010431e-06,
+      "loss": 0.2619,
+      "step": 2102
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.307546234696594e-06,
+      "loss": 0.2337,
+      "step": 2103
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.298929371286881e-06,
+      "loss": 0.271,
+      "step": 2104
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.290313808369005e-06,
+      "loss": 0.2594,
+      "step": 2105
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.281699552529676e-06,
+      "loss": 0.24,
+      "step": 2106
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.27308661035461e-06,
+      "loss": 0.2667,
+      "step": 2107
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.264474988428527e-06,
+      "loss": 0.2616,
+      "step": 2108
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.255864693335126e-06,
+      "loss": 0.2642,
+      "step": 2109
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.247255731657092e-06,
+      "loss": 0.2451,
+      "step": 2110
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.238648109976097e-06,
+      "loss": 0.2496,
+      "step": 2111
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.230041834872791e-06,
+      "loss": 0.2554,
+      "step": 2112
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.221436912926782e-06,
+      "loss": 0.2347,
+      "step": 2113
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.212833350716654e-06,
+      "loss": 0.2659,
+      "step": 2114
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.204231154819947e-06,
+      "loss": 0.2562,
+      "step": 2115
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.195630331813158e-06,
+      "loss": 0.2662,
+      "step": 2116
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.18703088827174e-06,
+      "loss": 0.2541,
+      "step": 2117
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.178432830770078e-06,
+      "loss": 0.2437,
+      "step": 2118
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.169836165881504e-06,
+      "loss": 0.2186,
+      "step": 2119
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.161240900178295e-06,
+      "loss": 0.2493,
+      "step": 2120
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.152647040231643e-06,
+      "loss": 0.2488,
+      "step": 2121
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.144054592611675e-06,
+      "loss": 0.2581,
+      "step": 2122
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.135463563887428e-06,
+      "loss": 0.249,
+      "step": 2123
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.126873960626868e-06,
+      "loss": 0.2183,
+      "step": 2124
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.118285789396869e-06,
+      "loss": 0.2685,
+      "step": 2125
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.109699056763193e-06,
+      "loss": 0.2568,
+      "step": 2126
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.101113769290524e-06,
+      "loss": 0.2445,
+      "step": 2127
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.092529933542432e-06,
+      "loss": 0.2352,
+      "step": 2128
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.083947556081371e-06,
+      "loss": 0.2595,
+      "step": 2129
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.075366643468695e-06,
+      "loss": 0.2284,
+      "step": 2130
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.06678720226462e-06,
+      "loss": 0.2442,
+      "step": 2131
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.058209239028256e-06,
+      "loss": 0.2481,
+      "step": 2132
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.049632760317566e-06,
+      "loss": 0.2311,
+      "step": 2133
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.041057772689391e-06,
+      "loss": 0.2565,
+      "step": 2134
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.03248428269942e-06,
+      "loss": 0.2591,
+      "step": 2135
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.023912296902208e-06,
+      "loss": 0.2539,
+      "step": 2136
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.01534182185116e-06,
+      "loss": 0.2408,
+      "step": 2137
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.006772864098512e-06,
+      "loss": 0.2537,
+      "step": 2138
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.998205430195358e-06,
+      "loss": 0.2397,
+      "step": 2139
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.98963952669162e-06,
+      "loss": 0.2642,
+      "step": 2140
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.98107516013604e-06,
+      "loss": 0.243,
+      "step": 2141
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.972512337076195e-06,
+      "loss": 0.2375,
+      "step": 2142
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.963951064058491e-06,
+      "loss": 0.2532,
+      "step": 2143
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.95539134762813e-06,
+      "loss": 0.2692,
+      "step": 2144
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.946833194329133e-06,
+      "loss": 0.2447,
+      "step": 2145
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.938276610704325e-06,
+      "loss": 0.2362,
+      "step": 2146
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.929721603295336e-06,
+      "loss": 0.2648,
+      "step": 2147
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.921168178642586e-06,
+      "loss": 0.2476,
+      "step": 2148
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.912616343285282e-06,
+      "loss": 0.2564,
+      "step": 2149
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.904066103761417e-06,
+      "loss": 0.2496,
+      "step": 2150
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.895517466607778e-06,
+      "loss": 0.2557,
+      "step": 2151
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.886970438359902e-06,
+      "loss": 0.258,
+      "step": 2152
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.87842502555212e-06,
+      "loss": 0.2558,
+      "step": 2153
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.869881234717507e-06,
+      "loss": 0.251,
+      "step": 2154
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.861339072387917e-06,
+      "loss": 0.268,
+      "step": 2155
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.85279854509395e-06,
+      "loss": 0.2346,
+      "step": 2156
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.84425965936495e-06,
+      "loss": 0.2506,
+      "step": 2157
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.835722421729014e-06,
+      "loss": 0.226,
+      "step": 2158
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.827186838712982e-06,
+      "loss": 0.2466,
+      "step": 2159
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.818652916842422e-06,
+      "loss": 0.2455,
+      "step": 2160
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.810120662641631e-06,
+      "loss": 0.2495,
+      "step": 2161
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.801590082633637e-06,
+      "loss": 0.2402,
+      "step": 2162
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.79306118334019e-06,
+      "loss": 0.2362,
+      "step": 2163
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.784533971281742e-06,
+      "loss": 0.2334,
+      "step": 2164
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.77600845297747e-06,
+      "loss": 0.2612,
+      "step": 2165
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.767484634945244e-06,
+      "loss": 0.2275,
+      "step": 2166
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.758962523701644e-06,
+      "loss": 0.2488,
+      "step": 2167
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.750442125761944e-06,
+      "loss": 0.2379,
+      "step": 2168
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.741923447640098e-06,
+      "loss": 0.2358,
+      "step": 2169
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.733406495848755e-06,
+      "loss": 0.2562,
+      "step": 2170
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.724891276899245e-06,
+      "loss": 0.2369,
+      "step": 2171
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.716377797301564e-06,
+      "loss": 0.2612,
+      "step": 2172
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.70786606356439e-06,
+      "loss": 0.251,
+      "step": 2173
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.699356082195051e-06,
+      "loss": 0.2514,
+      "step": 2174
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.690847859699555e-06,
+      "loss": 0.2495,
+      "step": 2175
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.682341402582552e-06,
+      "loss": 0.246,
+      "step": 2176
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.67383671734734e-06,
+      "loss": 0.2587,
+      "step": 2177
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.665333810495869e-06,
+      "loss": 0.2654,
+      "step": 2178
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.656832688528736e-06,
+      "loss": 0.2428,
+      "step": 2179
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.648333357945157e-06,
+      "loss": 0.2457,
+      "step": 2180
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.639835825242989e-06,
+      "loss": 0.249,
+      "step": 2181
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.63134009691871e-06,
+      "loss": 0.2433,
+      "step": 2182
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.622846179467427e-06,
+      "loss": 0.2336,
+      "step": 2183
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.614354079382851e-06,
+      "loss": 0.2541,
+      "step": 2184
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.605863803157313e-06,
+      "loss": 0.2331,
+      "step": 2185
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.597375357281739e-06,
+      "loss": 0.2683,
+      "step": 2186
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.588888748245669e-06,
+      "loss": 0.2484,
+      "step": 2187
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.580403982537231e-06,
+      "loss": 0.2328,
+      "step": 2188
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5719210666431405e-06,
+      "loss": 0.2429,
+      "step": 2189
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.563440007048708e-06,
+      "loss": 0.2459,
+      "step": 2190
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.55496081023782e-06,
+      "loss": 0.2491,
+      "step": 2191
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5464834826929345e-06,
+      "loss": 0.2543,
+      "step": 2192
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.538008030895085e-06,
+      "loss": 0.2421,
+      "step": 2193
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5295344613238775e-06,
+      "loss": 0.2371,
+      "step": 2194
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.521062780457466e-06,
+      "loss": 0.2644,
+      "step": 2195
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.51259299477257e-06,
+      "loss": 0.2691,
+      "step": 2196
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.50412511074445e-06,
+      "loss": 0.2423,
+      "step": 2197
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4956591348469265e-06,
+      "loss": 0.2398,
+      "step": 2198
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4871950735523556e-06,
+      "loss": 0.2597,
+      "step": 2199
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.478732933331623e-06,
+      "loss": 0.2455,
+      "step": 2200
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.47027272065415e-06,
+      "loss": 0.2602,
+      "step": 2201
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.461814441987892e-06,
+      "loss": 0.2684,
+      "step": 2202
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.453358103799315e-06,
+      "loss": 0.2576,
+      "step": 2203
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.444903712553405e-06,
+      "loss": 0.2483,
+      "step": 2204
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.436451274713658e-06,
+      "loss": 0.248,
+      "step": 2205
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.428000796742088e-06,
+      "loss": 0.2486,
+      "step": 2206
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.419552285099191e-06,
+      "loss": 0.2801,
+      "step": 2207
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.411105746243976e-06,
+      "loss": 0.2519,
+      "step": 2208
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.402661186633933e-06,
+      "loss": 0.2519,
+      "step": 2209
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.394218612725047e-06,
+      "loss": 0.2429,
+      "step": 2210
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.385778030971784e-06,
+      "loss": 0.2456,
+      "step": 2211
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3773394478270795e-06,
+      "loss": 0.2402,
+      "step": 2212
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.368902869742347e-06,
+      "loss": 0.2544,
+      "step": 2213
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.360468303167472e-06,
+      "loss": 0.2374,
+      "step": 2214
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.352035754550788e-06,
+      "loss": 0.2463,
+      "step": 2215
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.343605230339101e-06,
+      "loss": 0.2322,
+      "step": 2216
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.335176736977655e-06,
+      "loss": 0.2582,
+      "step": 2217
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3267502809101555e-06,
+      "loss": 0.2562,
+      "step": 2218
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.318325868578744e-06,
+      "loss": 0.2599,
+      "step": 2219
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.309903506423994e-06,
+      "loss": 0.2305,
+      "step": 2220
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.301483200884918e-06,
+      "loss": 0.2286,
+      "step": 2221
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.29306495839896e-06,
+      "loss": 0.2464,
+      "step": 2222
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.284648785401977e-06,
+      "loss": 0.2331,
+      "step": 2223
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.276234688328253e-06,
+      "loss": 0.2289,
+      "step": 2224
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.267822673610474e-06,
+      "loss": 0.2597,
+      "step": 2225
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.259412747679748e-06,
+      "loss": 0.2504,
+      "step": 2226
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.251004916965574e-06,
+      "loss": 0.249,
+      "step": 2227
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.242599187895857e-06,
+      "loss": 0.2441,
+      "step": 2228
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.234195566896886e-06,
+      "loss": 0.2409,
+      "step": 2229
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2257940603933515e-06,
+      "loss": 0.2528,
+      "step": 2230
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.21739467480832e-06,
+      "loss": 0.2619,
+      "step": 2231
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2089974165632344e-06,
+      "loss": 0.2459,
+      "step": 2232
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.200602292077911e-06,
+      "loss": 0.257,
+      "step": 2233
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.192209307770545e-06,
+      "loss": 0.2404,
+      "step": 2234
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.183818470057684e-06,
+      "loss": 0.2481,
+      "step": 2235
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.17542978535424e-06,
+      "loss": 0.2537,
+      "step": 2236
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.167043260073472e-06,
+      "loss": 0.2558,
+      "step": 2237
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.158658900627e-06,
+      "loss": 0.2416,
+      "step": 2238
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.150276713424782e-06,
+      "loss": 0.2529,
+      "step": 2239
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.141896704875109e-06,
+      "loss": 0.2419,
+      "step": 2240
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.133518881384618e-06,
+      "loss": 0.2506,
+      "step": 2241
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.125143249358271e-06,
+      "loss": 0.2385,
+      "step": 2242
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.116769815199348e-06,
+      "loss": 0.2306,
+      "step": 2243
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.108398585309455e-06,
+      "loss": 0.2524,
+      "step": 2244
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.10002956608852e-06,
+      "loss": 0.2522,
+      "step": 2245
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.091662763934764e-06,
+      "loss": 0.2435,
+      "step": 2246
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.083298185244729e-06,
+      "loss": 0.2305,
+      "step": 2247
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.074935836413242e-06,
+      "loss": 0.2376,
+      "step": 2248
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0665757238334384e-06,
+      "loss": 0.237,
+      "step": 2249
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.058217853896743e-06,
+      "loss": 0.2416,
+      "step": 2250
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.049862232992856e-06,
+      "loss": 0.2593,
+      "step": 2251
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.041508867509762e-06,
+      "loss": 0.2569,
+      "step": 2252
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.033157763833729e-06,
+      "loss": 0.2167,
+      "step": 2253
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.024808928349291e-06,
+      "loss": 0.2579,
+      "step": 2254
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0164623674392425e-06,
+      "loss": 0.2534,
+      "step": 2255
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.008118087484645e-06,
+      "loss": 0.2221,
+      "step": 2256
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.999776094864821e-06,
+      "loss": 0.2225,
+      "step": 2257
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.991436395957331e-06,
+      "loss": 0.2529,
+      "step": 2258
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.983098997137993e-06,
+      "loss": 0.2336,
+      "step": 2259
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.974763904780858e-06,
+      "loss": 0.2668,
+      "step": 2260
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.966431125258224e-06,
+      "loss": 0.2677,
+      "step": 2261
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.958100664940615e-06,
+      "loss": 0.2338,
+      "step": 2262
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9497725301967785e-06,
+      "loss": 0.2393,
+      "step": 2263
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.941446727393688e-06,
+      "loss": 0.2403,
+      "step": 2264
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.933123262896539e-06,
+      "loss": 0.2468,
+      "step": 2265
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.924802143068731e-06,
+      "loss": 0.2368,
+      "step": 2266
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.916483374271875e-06,
+      "loss": 0.2488,
+      "step": 2267
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.908166962865779e-06,
+      "loss": 0.2502,
+      "step": 2268
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.89985291520846e-06,
+      "loss": 0.2527,
+      "step": 2269
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.891541237656121e-06,
+      "loss": 0.2373,
+      "step": 2270
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.883231936563151e-06,
+      "loss": 0.2563,
+      "step": 2271
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.874925018282121e-06,
+      "loss": 0.2225,
+      "step": 2272
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8666204891637956e-06,
+      "loss": 0.2743,
+      "step": 2273
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.858318355557092e-06,
+      "loss": 0.2575,
+      "step": 2274
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.850018623809106e-06,
+      "loss": 0.2501,
+      "step": 2275
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.841721300265097e-06,
+      "loss": 0.2353,
+      "step": 2276
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.833426391268485e-06,
+      "loss": 0.2367,
+      "step": 2277
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.825133903160839e-06,
+      "loss": 0.2514,
+      "step": 2278
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.816843842281884e-06,
+      "loss": 0.2386,
+      "step": 2279
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.808556214969478e-06,
+      "loss": 0.2345,
+      "step": 2280
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.80027102755963e-06,
+      "loss": 0.2681,
+      "step": 2281
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.791988286386484e-06,
+      "loss": 0.2462,
+      "step": 2282
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.783707997782303e-06,
+      "loss": 0.2269,
+      "step": 2283
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7754301680774795e-06,
+      "loss": 0.2581,
+      "step": 2284
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.767154803600538e-06,
+      "loss": 0.237,
+      "step": 2285
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.758881910678097e-06,
+      "loss": 0.2436,
+      "step": 2286
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.750611495634905e-06,
+      "loss": 0.2659,
+      "step": 2287
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7423435647937985e-06,
+      "loss": 0.2488,
+      "step": 2288
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.734078124475731e-06,
+      "loss": 0.2458,
+      "step": 2289
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.725815180999748e-06,
+      "loss": 0.2519,
+      "step": 2290
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.717554740682974e-06,
+      "loss": 0.2405,
+      "step": 2291
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.709296809840635e-06,
+      "loss": 0.2384,
+      "step": 2292
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.701041394786035e-06,
+      "loss": 0.2645,
+      "step": 2293
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.692788501830544e-06,
+      "loss": 0.2433,
+      "step": 2294
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.684538137283614e-06,
+      "loss": 0.229,
+      "step": 2295
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.676290307452769e-06,
+      "loss": 0.2375,
+      "step": 2296
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.668045018643583e-06,
+      "loss": 0.2374,
+      "step": 2297
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.65980227715969e-06,
+      "loss": 0.2584,
+      "step": 2298
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.651562089302781e-06,
+      "loss": 0.2473,
+      "step": 2299
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6433244613725955e-06,
+      "loss": 0.2745,
+      "step": 2300
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.635089399666908e-06,
+      "loss": 0.2398,
+      "step": 2301
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.626856910481543e-06,
+      "loss": 0.243,
+      "step": 2302
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6186270001103414e-06,
+      "loss": 0.2529,
+      "step": 2303
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.610399674845191e-06,
+      "loss": 0.2719,
+      "step": 2304
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.602174940975994e-06,
+      "loss": 0.238,
+      "step": 2305
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.593952804790669e-06,
+      "loss": 0.2683,
+      "step": 2306
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.585733272575151e-06,
+      "loss": 0.238,
+      "step": 2307
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.577516350613392e-06,
+      "loss": 0.258,
+      "step": 2308
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.569302045187335e-06,
+      "loss": 0.253,
+      "step": 2309
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.561090362576934e-06,
+      "loss": 0.2365,
+      "step": 2310
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.552881309060125e-06,
+      "loss": 0.2282,
+      "step": 2311
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5446748909128495e-06,
+      "loss": 0.238,
+      "step": 2312
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.536471114409027e-06,
+      "loss": 0.2465,
+      "step": 2313
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.528269985820554e-06,
+      "loss": 0.2584,
+      "step": 2314
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.520071511417303e-06,
+      "loss": 0.2645,
+      "step": 2315
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.511875697467129e-06,
+      "loss": 0.2466,
+      "step": 2316
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.5036825502358365e-06,
+      "loss": 0.2539,
+      "step": 2317
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.495492075987204e-06,
+      "loss": 0.2489,
+      "step": 2318
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.487304280982955e-06,
+      "loss": 0.2289,
+      "step": 2319
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.479119171482777e-06,
+      "loss": 0.234,
+      "step": 2320
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.470936753744297e-06,
+      "loss": 0.2451,
+      "step": 2321
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.462757034023086e-06,
+      "loss": 0.2398,
+      "step": 2322
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.454580018572647e-06,
+      "loss": 0.2507,
+      "step": 2323
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.446405713644428e-06,
+      "loss": 0.2603,
+      "step": 2324
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.438234125487795e-06,
+      "loss": 0.2407,
+      "step": 2325
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.430065260350036e-06,
+      "loss": 0.2524,
+      "step": 2326
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.42189912447636e-06,
+      "loss": 0.2432,
+      "step": 2327
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.413735724109897e-06,
+      "loss": 0.231,
+      "step": 2328
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.405575065491672e-06,
+      "loss": 0.263,
+      "step": 2329
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.397417154860625e-06,
+      "loss": 0.2528,
+      "step": 2330
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.389261998453584e-06,
+      "loss": 0.2545,
+      "step": 2331
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3811096025052845e-06,
+      "loss": 0.2523,
+      "step": 2332
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.372959973248343e-06,
+      "loss": 0.2545,
+      "step": 2333
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.364813116913262e-06,
+      "loss": 0.2307,
+      "step": 2334
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.356669039728423e-06,
+      "loss": 0.2604,
+      "step": 2335
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3485277479200915e-06,
+      "loss": 0.2279,
+      "step": 2336
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.340389247712391e-06,
+      "loss": 0.2554,
+      "step": 2337
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.332253545327318e-06,
+      "loss": 0.2653,
+      "step": 2338
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.324120646984726e-06,
+      "loss": 0.2314,
+      "step": 2339
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.315990558902333e-06,
+      "loss": 0.2585,
+      "step": 2340
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3078632872957014e-06,
+      "loss": 0.2473,
+      "step": 2341
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2997388383782354e-06,
+      "loss": 0.2293,
+      "step": 2342
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.291617218361195e-06,
+      "loss": 0.238,
+      "step": 2343
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.28349843345367e-06,
+      "loss": 0.234,
+      "step": 2344
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.275382489862581e-06,
+      "loss": 0.2579,
+      "step": 2345
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2672693937926724e-06,
+      "loss": 0.234,
+      "step": 2346
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.259159151446524e-06,
+      "loss": 0.2294,
+      "step": 2347
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.25105176902453e-06,
+      "loss": 0.2286,
+      "step": 2348
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.242947252724889e-06,
+      "loss": 0.2623,
+      "step": 2349
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.234845608743616e-06,
+      "loss": 0.2488,
+      "step": 2350
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.226746843274535e-06,
+      "loss": 0.2505,
+      "step": 2351
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.218650962509256e-06,
+      "loss": 0.2476,
+      "step": 2352
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2105579726371965e-06,
+      "loss": 0.2603,
+      "step": 2353
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.202467879845551e-06,
+      "loss": 0.2397,
+      "step": 2354
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.194380690319312e-06,
+      "loss": 0.245,
+      "step": 2355
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.186296410241249e-06,
+      "loss": 0.2415,
+      "step": 2356
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1782150457919e-06,
+      "loss": 0.2594,
+      "step": 2357
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.170136603149579e-06,
+      "loss": 0.259,
+      "step": 2358
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.162061088490371e-06,
+      "loss": 0.2163,
+      "step": 2359
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.153988507988114e-06,
+      "loss": 0.247,
+      "step": 2360
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1459188678144085e-06,
+      "loss": 0.2507,
+      "step": 2361
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1378521741386e-06,
+      "loss": 0.2501,
+      "step": 2362
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.129788433127793e-06,
+      "loss": 0.2638,
+      "step": 2363
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.121727650946827e-06,
+      "loss": 0.2596,
+      "step": 2364
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.113669833758278e-06,
+      "loss": 0.2383,
+      "step": 2365
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.1056149877224555e-06,
+      "loss": 0.232,
+      "step": 2366
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0975631189974085e-06,
+      "loss": 0.2503,
+      "step": 2367
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.089514233738896e-06,
+      "loss": 0.2331,
+      "step": 2368
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.081468338100398e-06,
+      "loss": 0.2494,
+      "step": 2369
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.073425438233114e-06,
+      "loss": 0.2364,
+      "step": 2370
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.065385540285957e-06,
+      "loss": 0.232,
+      "step": 2371
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.057348650405531e-06,
+      "loss": 0.2522,
+      "step": 2372
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.049314774736156e-06,
+      "loss": 0.2431,
+      "step": 2373
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.041283919419832e-06,
+      "loss": 0.2418,
+      "step": 2374
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.033256090596265e-06,
+      "loss": 0.2578,
+      "step": 2375
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.025231294402838e-06,
+      "loss": 0.2518,
+      "step": 2376
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.017209536974618e-06,
+      "loss": 0.2495,
+      "step": 2377
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.009190824444344e-06,
+      "loss": 0.2473,
+      "step": 2378
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.001175162942441e-06,
+      "loss": 0.2518,
+      "step": 2379
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.993162558596986e-06,
+      "loss": 0.2549,
+      "step": 2380
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9851530175337265e-06,
+      "loss": 0.2629,
+      "step": 2381
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.977146545876062e-06,
+      "loss": 0.2587,
+      "step": 2382
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.969143149745058e-06,
+      "loss": 0.2575,
+      "step": 2383
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.961142835259417e-06,
+      "loss": 0.2379,
+      "step": 2384
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9531456085354875e-06,
+      "loss": 0.2287,
+      "step": 2385
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9451514756872585e-06,
+      "loss": 0.2213,
+      "step": 2386
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.937160442826359e-06,
+      "loss": 0.2532,
+      "step": 2387
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.929172516062041e-06,
+      "loss": 0.2487,
+      "step": 2388
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.921187701501183e-06,
+      "loss": 0.2445,
+      "step": 2389
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.913206005248284e-06,
+      "loss": 0.2804,
+      "step": 2390
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.905227433405465e-06,
+      "loss": 0.2406,
+      "step": 2391
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.897251992072448e-06,
+      "loss": 0.2526,
+      "step": 2392
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.88927968734657e-06,
+      "loss": 0.2384,
+      "step": 2393
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.881310525322769e-06,
+      "loss": 0.2588,
+      "step": 2394
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.873344512093576e-06,
+      "loss": 0.2581,
+      "step": 2395
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.86538165374912e-06,
+      "loss": 0.2521,
+      "step": 2396
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.857421956377111e-06,
+      "loss": 0.2388,
+      "step": 2397
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.849465426062852e-06,
+      "loss": 0.2355,
+      "step": 2398
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.841512068889218e-06,
+      "loss": 0.2553,
+      "step": 2399
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8335618909366575e-06,
+      "loss": 0.2265,
+      "step": 2400
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8256148982831904e-06,
+      "loss": 0.2207,
+      "step": 2401
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.817671097004407e-06,
+      "loss": 0.2606,
+      "step": 2402
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.809730493173449e-06,
+      "loss": 0.2641,
+      "step": 2403
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.801793092861012e-06,
+      "loss": 0.2465,
+      "step": 2404
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.793858902135354e-06,
+      "loss": 0.2617,
+      "step": 2405
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.785927927062274e-06,
+      "loss": 0.2401,
+      "step": 2406
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.778000173705108e-06,
+      "loss": 0.2583,
+      "step": 2407
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.770075648124731e-06,
+      "loss": 0.2279,
+      "step": 2408
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.762154356379549e-06,
+      "loss": 0.2372,
+      "step": 2409
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.754236304525502e-06,
+      "loss": 0.257,
+      "step": 2410
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.746321498616054e-06,
+      "loss": 0.2616,
+      "step": 2411
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.738409944702178e-06,
+      "loss": 0.2714,
+      "step": 2412
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.73050164883236e-06,
+      "loss": 0.2461,
+      "step": 2413
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7225966170526126e-06,
+      "loss": 0.2491,
+      "step": 2414
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.714694855406433e-06,
+      "loss": 0.237,
+      "step": 2415
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.706796369934825e-06,
+      "loss": 0.242,
+      "step": 2416
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.698901166676291e-06,
+      "loss": 0.2298,
+      "step": 2417
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.691009251666828e-06,
+      "loss": 0.247,
+      "step": 2418
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.683120630939908e-06,
+      "loss": 0.253,
+      "step": 2419
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.67523531052649e-06,
+      "loss": 0.2464,
+      "step": 2420
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.667353296455007e-06,
+      "loss": 0.2537,
+      "step": 2421
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.65947459475137e-06,
+      "loss": 0.2353,
+      "step": 2422
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.651599211438959e-06,
+      "loss": 0.249,
+      "step": 2423
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.643727152538605e-06,
+      "loss": 0.2328,
+      "step": 2424
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.635858424068607e-06,
+      "loss": 0.2452,
+      "step": 2425
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.627993032044719e-06,
+      "loss": 0.2665,
+      "step": 2426
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.620130982480136e-06,
+      "loss": 0.2354,
+      "step": 2427
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.612272281385511e-06,
+      "loss": 0.2519,
+      "step": 2428
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.604416934768917e-06,
+      "loss": 0.2328,
+      "step": 2429
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.596564948635887e-06,
+      "loss": 0.2676,
+      "step": 2430
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.588716328989365e-06,
+      "loss": 0.26,
+      "step": 2431
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.580871081829726e-06,
+      "loss": 0.2408,
+      "step": 2432
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.573029213154772e-06,
+      "loss": 0.2468,
+      "step": 2433
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.565190728959724e-06,
+      "loss": 0.2512,
+      "step": 2434
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.557355635237205e-06,
+      "loss": 0.2212,
+      "step": 2435
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.549523937977255e-06,
+      "loss": 0.2589,
+      "step": 2436
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.541695643167306e-06,
+      "loss": 0.2348,
+      "step": 2437
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.533870756792208e-06,
+      "loss": 0.2491,
+      "step": 2438
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.526049284834185e-06,
+      "loss": 0.2537,
+      "step": 2439
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5182312332728616e-06,
+      "loss": 0.2312,
+      "step": 2440
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.510416608085254e-06,
+      "loss": 0.2392,
+      "step": 2441
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.502605415245741e-06,
+      "loss": 0.2412,
+      "step": 2442
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.49479766072609e-06,
+      "loss": 0.2468,
+      "step": 2443
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.486993350495433e-06,
+      "loss": 0.2166,
+      "step": 2444
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.479192490520274e-06,
+      "loss": 0.2565,
+      "step": 2445
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.471395086764483e-06,
+      "loss": 0.2676,
+      "step": 2446
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.463601145189277e-06,
+      "loss": 0.2537,
+      "step": 2447
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.455810671753228e-06,
+      "loss": 0.2389,
+      "step": 2448
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.448023672412268e-06,
+      "loss": 0.2397,
+      "step": 2449
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4402401531196556e-06,
+      "loss": 0.2399,
+      "step": 2450
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.432460119826009e-06,
+      "loss": 0.2444,
+      "step": 2451
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.424683578479259e-06,
+      "loss": 0.2421,
+      "step": 2452
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.416910535024687e-06,
+      "loss": 0.2476,
+      "step": 2453
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.409140995404889e-06,
+      "loss": 0.2374,
+      "step": 2454
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.401374965559779e-06,
+      "loss": 0.2557,
+      "step": 2455
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.393612451426599e-06,
+      "loss": 0.2515,
+      "step": 2456
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.385853458939899e-06,
+      "loss": 0.2491,
+      "step": 2457
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.378097994031533e-06,
+      "loss": 0.2624,
+      "step": 2458
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3703460626306606e-06,
+      "loss": 0.2277,
+      "step": 2459
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.362597670663734e-06,
+      "loss": 0.2353,
+      "step": 2460
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.354852824054515e-06,
+      "loss": 0.229,
+      "step": 2461
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.347111528724037e-06,
+      "loss": 0.2466,
+      "step": 2462
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.339373790590633e-06,
+      "loss": 0.2348,
+      "step": 2463
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.331639615569903e-06,
+      "loss": 0.2228,
+      "step": 2464
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.323909009574737e-06,
+      "loss": 0.2449,
+      "step": 2465
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.316181978515287e-06,
+      "loss": 0.2251,
+      "step": 2466
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.30845852829897e-06,
+      "loss": 0.2469,
+      "step": 2467
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.300738664830471e-06,
+      "loss": 0.2327,
+      "step": 2468
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.293022394011739e-06,
+      "loss": 0.2654,
+      "step": 2469
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.285309721741963e-06,
+      "loss": 0.2408,
+      "step": 2470
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.277600653917586e-06,
+      "loss": 0.2438,
+      "step": 2471
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.269895196432294e-06,
+      "loss": 0.1525,
+      "step": 2472
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.262193355177019e-06,
+      "loss": 0.1508,
+      "step": 2473
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.254495136039926e-06,
+      "loss": 0.1521,
+      "step": 2474
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.246800544906409e-06,
+      "loss": 0.1428,
+      "step": 2475
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.23910958765908e-06,
+      "loss": 0.1505,
+      "step": 2476
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.231422270177793e-06,
+      "loss": 0.1257,
+      "step": 2477
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.223738598339596e-06,
+      "loss": 0.1423,
+      "step": 2478
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.216058578018775e-06,
+      "loss": 0.1407,
+      "step": 2479
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.208382215086796e-06,
+      "loss": 0.1535,
+      "step": 2480
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.200709515412358e-06,
+      "loss": 0.1297,
+      "step": 2481
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.193040484861337e-06,
+      "loss": 0.1383,
+      "step": 2482
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.185375129296815e-06,
+      "loss": 0.1352,
+      "step": 2483
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.177713454579051e-06,
+      "loss": 0.1451,
+      "step": 2484
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.170055466565518e-06,
+      "loss": 0.1227,
+      "step": 2485
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1624011711108444e-06,
+      "loss": 0.132,
+      "step": 2486
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1547505740668434e-06,
+      "loss": 0.115,
+      "step": 2487
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.147103681282499e-06,
+      "loss": 0.138,
+      "step": 2488
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.139460498603972e-06,
+      "loss": 0.1352,
+      "step": 2489
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1318210318745745e-06,
+      "loss": 0.1144,
+      "step": 2490
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.124185286934785e-06,
+      "loss": 0.1379,
+      "step": 2491
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.116553269622242e-06,
+      "loss": 0.1434,
+      "step": 2492
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1089249857717205e-06,
+      "loss": 0.1342,
+      "step": 2493
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.10130044121515e-06,
+      "loss": 0.1306,
+      "step": 2494
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.093679641781595e-06,
+      "loss": 0.1213,
+      "step": 2495
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.086062593297267e-06,
+      "loss": 0.1471,
+      "step": 2496
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.078449301585504e-06,
+      "loss": 0.1373,
+      "step": 2497
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0708397724667716e-06,
+      "loss": 0.128,
+      "step": 2498
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.063234011758654e-06,
+      "loss": 0.1281,
+      "step": 2499
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0556320252758676e-06,
+      "loss": 0.1399,
+      "step": 2500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.048033818830227e-06,
+      "loss": 0.132,
+      "step": 2501
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.040439398230673e-06,
+      "loss": 0.1227,
+      "step": 2502
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.032848769283238e-06,
+      "loss": 0.12,
+      "step": 2503
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.025261937791068e-06,
+      "loss": 0.1279,
+      "step": 2504
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.017678909554397e-06,
+      "loss": 0.1359,
+      "step": 2505
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0100996903705515e-06,
+      "loss": 0.1256,
+      "step": 2506
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.002524286033948e-06,
+      "loss": 0.129,
+      "step": 2507
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.994952702336095e-06,
+      "loss": 0.134,
+      "step": 2508
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.987384945065568e-06,
+      "loss": 0.1225,
+      "step": 2509
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.979821020008019e-06,
+      "loss": 0.1269,
+      "step": 2510
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9722609329461705e-06,
+      "loss": 0.131,
+      "step": 2511
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.964704689659818e-06,
+      "loss": 0.1445,
+      "step": 2512
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.95715229592581e-06,
+      "loss": 0.1223,
+      "step": 2513
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.949603757518059e-06,
+      "loss": 0.1214,
+      "step": 2514
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.942059080207518e-06,
+      "loss": 0.1301,
+      "step": 2515
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9345182697622086e-06,
+      "loss": 0.1209,
+      "step": 2516
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.926981331947175e-06,
+      "loss": 0.1297,
+      "step": 2517
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.919448272524509e-06,
+      "loss": 0.1143,
+      "step": 2518
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9119190972533396e-06,
+      "loss": 0.1127,
+      "step": 2519
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.904393811889829e-06,
+      "loss": 0.1334,
+      "step": 2520
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.896872422187158e-06,
+      "loss": 0.1339,
+      "step": 2521
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8893549338955316e-06,
+      "loss": 0.132,
+      "step": 2522
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.881841352762169e-06,
+      "loss": 0.1326,
+      "step": 2523
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.87433168453131e-06,
+      "loss": 0.1192,
+      "step": 2524
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.866825934944203e-06,
+      "loss": 0.1197,
+      "step": 2525
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.859324109739092e-06,
+      "loss": 0.1376,
+      "step": 2526
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.851826214651223e-06,
+      "loss": 0.1046,
+      "step": 2527
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.844332255412845e-06,
+      "loss": 0.1206,
+      "step": 2528
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.836842237753187e-06,
+      "loss": 0.11,
+      "step": 2529
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.829356167398477e-06,
+      "loss": 0.1243,
+      "step": 2530
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.821874050071911e-06,
+      "loss": 0.1202,
+      "step": 2531
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8143958914936775e-06,
+      "loss": 0.132,
+      "step": 2532
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.806921697380928e-06,
+      "loss": 0.1341,
+      "step": 2533
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.799451473447787e-06,
+      "loss": 0.1235,
+      "step": 2534
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.791985225405339e-06,
+      "loss": 0.1274,
+      "step": 2535
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.784522958961636e-06,
+      "loss": 0.1317,
+      "step": 2536
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.777064679821688e-06,
+      "loss": 0.1394,
+      "step": 2537
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.769610393687447e-06,
+      "loss": 0.1196,
+      "step": 2538
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.762160106257815e-06,
+      "loss": 0.1181,
+      "step": 2539
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.754713823228644e-06,
+      "loss": 0.1246,
+      "step": 2540
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.747271550292713e-06,
+      "loss": 0.135,
+      "step": 2541
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.739833293139746e-06,
+      "loss": 0.1313,
+      "step": 2542
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7323990574563975e-06,
+      "loss": 0.1208,
+      "step": 2543
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.724968848926236e-06,
+      "loss": 0.1309,
+      "step": 2544
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7175426732297605e-06,
+      "loss": 0.1245,
+      "step": 2545
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.710120536044378e-06,
+      "loss": 0.1292,
+      "step": 2546
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7027024430444205e-06,
+      "loss": 0.1268,
+      "step": 2547
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.695288399901123e-06,
+      "loss": 0.1159,
+      "step": 2548
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.687878412282622e-06,
+      "loss": 0.1217,
+      "step": 2549
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.680472485853951e-06,
+      "loss": 0.1318,
+      "step": 2550
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.673070626277047e-06,
+      "loss": 0.1282,
+      "step": 2551
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.665672839210728e-06,
+      "loss": 0.1301,
+      "step": 2552
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.65827913031071e-06,
+      "loss": 0.136,
+      "step": 2553
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6508895052295774e-06,
+      "loss": 0.1175,
+      "step": 2554
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.643503969616809e-06,
+      "loss": 0.1195,
+      "step": 2555
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.636122529118744e-06,
+      "loss": 0.1141,
+      "step": 2556
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.628745189378594e-06,
+      "loss": 0.1204,
+      "step": 2557
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.621371956036432e-06,
+      "loss": 0.1365,
+      "step": 2558
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.614002834729202e-06,
+      "loss": 0.1274,
+      "step": 2559
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.606637831090701e-06,
+      "loss": 0.1362,
+      "step": 2560
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.599276950751571e-06,
+      "loss": 0.1393,
+      "step": 2561
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.591920199339303e-06,
+      "loss": 0.117,
+      "step": 2562
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.584567582478241e-06,
+      "loss": 0.133,
+      "step": 2563
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.577219105789554e-06,
+      "loss": 0.1336,
+      "step": 2564
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.569874774891263e-06,
+      "loss": 0.115,
+      "step": 2565
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.562534595398199e-06,
+      "loss": 0.1266,
+      "step": 2566
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.555198572922041e-06,
+      "loss": 0.122,
+      "step": 2567
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.547866713071271e-06,
+      "loss": 0.1317,
+      "step": 2568
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.540539021451196e-06,
+      "loss": 0.1272,
+      "step": 2569
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.533215503663938e-06,
+      "loss": 0.1303,
+      "step": 2570
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.525896165308433e-06,
+      "loss": 0.1193,
+      "step": 2571
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.518581011980411e-06,
+      "loss": 0.1295,
+      "step": 2572
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.511270049272406e-06,
+      "loss": 0.1191,
+      "step": 2573
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.503963282773746e-06,
+      "loss": 0.116,
+      "step": 2574
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.496660718070559e-06,
+      "loss": 0.1222,
+      "step": 2575
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.489362360745758e-06,
+      "loss": 0.1231,
+      "step": 2576
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.482068216379035e-06,
+      "loss": 0.1286,
+      "step": 2577
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4747782905468594e-06,
+      "loss": 0.1263,
+      "step": 2578
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.467492588822487e-06,
+      "loss": 0.1269,
+      "step": 2579
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.460211116775932e-06,
+      "loss": 0.1143,
+      "step": 2580
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.452933879973978e-06,
+      "loss": 0.1326,
+      "step": 2581
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.445660883980174e-06,
+      "loss": 0.1093,
+      "step": 2582
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.43839213435483e-06,
+      "loss": 0.1262,
+      "step": 2583
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.431127636655e-06,
+      "loss": 0.1399,
+      "step": 2584
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.423867396434493e-06,
+      "loss": 0.1292,
+      "step": 2585
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.41661141924386e-06,
+      "loss": 0.1216,
+      "step": 2586
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.409359710630396e-06,
+      "loss": 0.1259,
+      "step": 2587
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.402112276138137e-06,
+      "loss": 0.1332,
+      "step": 2588
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.394869121307843e-06,
+      "loss": 0.1229,
+      "step": 2589
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.387630251677001e-06,
+      "loss": 0.1261,
+      "step": 2590
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.380395672779833e-06,
+      "loss": 0.1294,
+      "step": 2591
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.373165390147266e-06,
+      "loss": 0.1209,
+      "step": 2592
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.365939409306954e-06,
+      "loss": 0.1109,
+      "step": 2593
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.358717735783264e-06,
+      "loss": 0.1284,
+      "step": 2594
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.35150037509726e-06,
+      "loss": 0.1257,
+      "step": 2595
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.344287332766712e-06,
+      "loss": 0.1274,
+      "step": 2596
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.337078614306087e-06,
+      "loss": 0.1177,
+      "step": 2597
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.329874225226553e-06,
+      "loss": 0.1323,
+      "step": 2598
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.322674171035969e-06,
+      "loss": 0.1365,
+      "step": 2599
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.315478457238869e-06,
+      "loss": 0.1218,
+      "step": 2600
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.308287089336473e-06,
+      "loss": 0.1215,
+      "step": 2601
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.301100072826687e-06,
+      "loss": 0.1444,
+      "step": 2602
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.293917413204081e-06,
+      "loss": 0.1293,
+      "step": 2603
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.286739115959893e-06,
+      "loss": 0.1268,
+      "step": 2604
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.279565186582032e-06,
+      "loss": 0.127,
+      "step": 2605
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.27239563055507e-06,
+      "loss": 0.1255,
+      "step": 2606
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.265230453360228e-06,
+      "loss": 0.1176,
+      "step": 2607
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.258069660475382e-06,
+      "loss": 0.1117,
+      "step": 2608
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.250913257375052e-06,
+      "loss": 0.1275,
+      "step": 2609
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.243761249530411e-06,
+      "loss": 0.1242,
+      "step": 2610
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.236613642409269e-06,
+      "loss": 0.1223,
+      "step": 2611
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.229470441476067e-06,
+      "loss": 0.1296,
+      "step": 2612
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.222331652191878e-06,
+      "loss": 0.1216,
+      "step": 2613
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.215197280014409e-06,
+      "loss": 0.1238,
+      "step": 2614
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.20806733039798e-06,
+      "loss": 0.125,
+      "step": 2615
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.200941808793543e-06,
+      "loss": 0.1293,
+      "step": 2616
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.193820720648647e-06,
+      "loss": 0.1134,
+      "step": 2617
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.186704071407471e-06,
+      "loss": 0.1154,
+      "step": 2618
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1795918665107885e-06,
+      "loss": 0.1245,
+      "step": 2619
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.172484111395972e-06,
+      "loss": 0.1296,
+      "step": 2620
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.165380811497001e-06,
+      "loss": 0.1259,
+      "step": 2621
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.15828197224445e-06,
+      "loss": 0.1198,
+      "step": 2622
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.151187599065477e-06,
+      "loss": 0.1314,
+      "step": 2623
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.144097697383826e-06,
+      "loss": 0.1221,
+      "step": 2624
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.137012272619821e-06,
+      "loss": 0.1146,
+      "step": 2625
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.129931330190373e-06,
+      "loss": 0.1259,
+      "step": 2626
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.122854875508952e-06,
+      "loss": 0.129,
+      "step": 2627
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.115782913985614e-06,
+      "loss": 0.1298,
+      "step": 2628
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.108715451026962e-06,
+      "loss": 0.1193,
+      "step": 2629
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.101652492036177e-06,
+      "loss": 0.1248,
+      "step": 2630
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.094594042412982e-06,
+      "loss": 0.131,
+      "step": 2631
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.087540107553656e-06,
+      "loss": 0.1284,
+      "step": 2632
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.080490692851033e-06,
+      "loss": 0.1282,
+      "step": 2633
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.073445803694492e-06,
+      "loss": 0.1269,
+      "step": 2634
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.06640544546994e-06,
+      "loss": 0.1163,
+      "step": 2635
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.059369623559831e-06,
+      "loss": 0.1296,
+      "step": 2636
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.052338343343143e-06,
+      "loss": 0.1319,
+      "step": 2637
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.045311610195389e-06,
+      "loss": 0.134,
+      "step": 2638
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.038289429488608e-06,
+      "loss": 0.1281,
+      "step": 2639
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.031271806591348e-06,
+      "loss": 0.1051,
+      "step": 2640
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.024258746868675e-06,
+      "loss": 0.1231,
+      "step": 2641
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.017250255682177e-06,
+      "loss": 0.1258,
+      "step": 2642
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0102463383899345e-06,
+      "loss": 0.1418,
+      "step": 2643
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.003247000346542e-06,
+      "loss": 0.113,
+      "step": 2644
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.996252246903093e-06,
+      "loss": 0.1256,
+      "step": 2645
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.989262083407169e-06,
+      "loss": 0.1144,
+      "step": 2646
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.982276515202843e-06,
+      "loss": 0.1316,
+      "step": 2647
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.975295547630679e-06,
+      "loss": 0.1395,
+      "step": 2648
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.968319186027727e-06,
+      "loss": 0.1219,
+      "step": 2649
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.961347435727504e-06,
+      "loss": 0.1244,
+      "step": 2650
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.954380302060017e-06,
+      "loss": 0.1279,
+      "step": 2651
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.947417790351726e-06,
+      "loss": 0.1158,
+      "step": 2652
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.940459905925577e-06,
+      "loss": 0.1195,
+      "step": 2653
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9335066541009616e-06,
+      "loss": 0.1211,
+      "step": 2654
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9265580401937334e-06,
+      "loss": 0.1254,
+      "step": 2655
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.919614069516208e-06,
+      "loss": 0.1129,
+      "step": 2656
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.912674747377149e-06,
+      "loss": 0.1176,
+      "step": 2657
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.905740079081759e-06,
+      "loss": 0.1305,
+      "step": 2658
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.898810069931688e-06,
+      "loss": 0.1206,
+      "step": 2659
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.891884725225019e-06,
+      "loss": 0.1328,
+      "step": 2660
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.8849640502562765e-06,
+      "loss": 0.1289,
+      "step": 2661
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.878048050316416e-06,
+      "loss": 0.1213,
+      "step": 2662
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.87113673069281e-06,
+      "loss": 0.1259,
+      "step": 2663
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.864230096669254e-06,
+      "loss": 0.1246,
+      "step": 2664
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.857328153525972e-06,
+      "loss": 0.1257,
+      "step": 2665
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.850430906539586e-06,
+      "loss": 0.1219,
+      "step": 2666
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8435383609831435e-06,
+      "loss": 0.1273,
+      "step": 2667
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.836650522126082e-06,
+      "loss": 0.1114,
+      "step": 2668
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.829767395234258e-06,
+      "loss": 0.1293,
+      "step": 2669
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.822888985569911e-06,
+      "loss": 0.1305,
+      "step": 2670
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.816015298391681e-06,
+      "loss": 0.1256,
+      "step": 2671
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8091463389545878e-06,
+      "loss": 0.1307,
+      "step": 2672
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8022821125100607e-06,
+      "loss": 0.1257,
+      "step": 2673
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7954226243058878e-06,
+      "loss": 0.1154,
+      "step": 2674
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.78856787958624e-06,
+      "loss": 0.1109,
+      "step": 2675
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.781717883591662e-06,
+      "loss": 0.119,
+      "step": 2676
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7748726415590763e-06,
+      "loss": 0.1285,
+      "step": 2677
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.768032158721756e-06,
+      "loss": 0.1304,
+      "step": 2678
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7611964403093537e-06,
+      "loss": 0.1219,
+      "step": 2679
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.754365491547861e-06,
+      "loss": 0.1343,
+      "step": 2680
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.74753931765964e-06,
+      "loss": 0.1217,
+      "step": 2681
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7407179238633895e-06,
+      "loss": 0.1116,
+      "step": 2682
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7339013153741566e-06,
+      "loss": 0.1187,
+      "step": 2683
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7270894974033344e-06,
+      "loss": 0.1108,
+      "step": 2684
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7202824751586555e-06,
+      "loss": 0.1152,
+      "step": 2685
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.713480253844177e-06,
+      "loss": 0.1194,
+      "step": 2686
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7066828386602925e-06,
+      "loss": 0.1204,
+      "step": 2687
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6998902348037134e-06,
+      "loss": 0.1272,
+      "step": 2688
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6931024474674838e-06,
+      "loss": 0.1156,
+      "step": 2689
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.686319481840963e-06,
+      "loss": 0.1281,
+      "step": 2690
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6795413431098172e-06,
+      "loss": 0.1258,
+      "step": 2691
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.672768036456024e-06,
+      "loss": 0.1221,
+      "step": 2692
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6659995670578764e-06,
+      "loss": 0.1271,
+      "step": 2693
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6592359400899536e-06,
+      "loss": 0.1206,
+      "step": 2694
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.652477160723146e-06,
+      "loss": 0.1236,
+      "step": 2695
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.645723234124636e-06,
+      "loss": 0.1157,
+      "step": 2696
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.638974165457891e-06,
+      "loss": 0.125,
+      "step": 2697
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6322299598826647e-06,
+      "loss": 0.1223,
+      "step": 2698
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6254906225549924e-06,
+      "loss": 0.1227,
+      "step": 2699
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.618756158627197e-06,
+      "loss": 0.117,
+      "step": 2700
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.612026573247861e-06,
+      "loss": 0.1251,
+      "step": 2701
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6053018715618515e-06,
+      "loss": 0.1193,
+      "step": 2702
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5985820587102893e-06,
+      "loss": 0.1185,
+      "step": 2703
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5918671398305694e-06,
+      "loss": 0.1345,
+      "step": 2704
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5851571200563373e-06,
+      "loss": 0.1197,
+      "step": 2705
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5784520045174898e-06,
+      "loss": 0.1374,
+      "step": 2706
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.571751798340185e-06,
+      "loss": 0.1199,
+      "step": 2707
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5650565066468244e-06,
+      "loss": 0.1097,
+      "step": 2708
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.558366134556048e-06,
+      "loss": 0.122,
+      "step": 2709
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.551680687182737e-06,
+      "loss": 0.1177,
+      "step": 2710
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.545000169638003e-06,
+      "loss": 0.1366,
+      "step": 2711
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.538324587029198e-06,
+      "loss": 0.1186,
+      "step": 2712
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5316539444599006e-06,
+      "loss": 0.1185,
+      "step": 2713
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5249882470299023e-06,
+      "loss": 0.128,
+      "step": 2714
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.51832749983522e-06,
+      "loss": 0.1219,
+      "step": 2715
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5116717079680906e-06,
+      "loss": 0.1307,
+      "step": 2716
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5050208765169533e-06,
+      "loss": 0.1178,
+      "step": 2717
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.498375010566466e-06,
+      "loss": 0.1198,
+      "step": 2718
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4917341151974783e-06,
+      "loss": 0.1191,
+      "step": 2719
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.485098195487051e-06,
+      "loss": 0.1128,
+      "step": 2720
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.478467256508434e-06,
+      "loss": 0.1411,
+      "step": 2721
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4718413033310704e-06,
+      "loss": 0.1264,
+      "step": 2722
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4652203410205887e-06,
+      "loss": 0.121,
+      "step": 2723
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4586043746388087e-06,
+      "loss": 0.1286,
+      "step": 2724
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4519934092437323e-06,
+      "loss": 0.1275,
+      "step": 2725
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.445387449889528e-06,
+      "loss": 0.1233,
+      "step": 2726
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.438786501626541e-06,
+      "loss": 0.1284,
+      "step": 2727
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4321905695012914e-06,
+      "loss": 0.1141,
+      "step": 2728
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4255996585564546e-06,
+      "loss": 0.1231,
+      "step": 2729
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.419013773830877e-06,
+      "loss": 0.1246,
+      "step": 2730
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.412432920359554e-06,
+      "loss": 0.1202,
+      "step": 2731
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.405857103173644e-06,
+      "loss": 0.1291,
+      "step": 2732
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.399286327300445e-06,
+      "loss": 0.1243,
+      "step": 2733
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3927205977634035e-06,
+      "loss": 0.1193,
+      "step": 2734
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3861599195821127e-06,
+      "loss": 0.1308,
+      "step": 2735
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3796042977723043e-06,
+      "loss": 0.1219,
+      "step": 2736
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3730537373458374e-06,
+      "loss": 0.1281,
+      "step": 2737
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3665082433107055e-06,
+      "loss": 0.1377,
+      "step": 2738
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.359967820671025e-06,
+      "loss": 0.1111,
+      "step": 2739
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.353432474427043e-06,
+      "loss": 0.1263,
+      "step": 2740
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3469022095751235e-06,
+      "loss": 0.1201,
+      "step": 2741
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.340377031107741e-06,
+      "loss": 0.1189,
+      "step": 2742
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3338569440134806e-06,
+      "loss": 0.1326,
+      "step": 2743
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3273419532770458e-06,
+      "loss": 0.1275,
+      "step": 2744
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.320832063879231e-06,
+      "loss": 0.1267,
+      "step": 2745
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.314327280796934e-06,
+      "loss": 0.1245,
+      "step": 2746
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3078276090031535e-06,
+      "loss": 0.1211,
+      "step": 2747
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.301333053466982e-06,
+      "loss": 0.1229,
+      "step": 2748
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2948436191535926e-06,
+      "loss": 0.1362,
+      "step": 2749
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.288359311024243e-06,
+      "loss": 0.1229,
+      "step": 2750
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2818801340362827e-06,
+      "loss": 0.1213,
+      "step": 2751
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.275406093143124e-06,
+      "loss": 0.1134,
+      "step": 2752
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.268937193294268e-06,
+      "loss": 0.1216,
+      "step": 2753
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2624734394352707e-06,
+      "loss": 0.1324,
+      "step": 2754
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.256014836507766e-06,
+      "loss": 0.1192,
+      "step": 2755
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2495613894494414e-06,
+      "loss": 0.1146,
+      "step": 2756
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.243113103194043e-06,
+      "loss": 0.1117,
+      "step": 2757
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2366699826713787e-06,
+      "loss": 0.12,
+      "step": 2758
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.230232032807302e-06,
+      "loss": 0.1208,
+      "step": 2759
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.223799258523714e-06,
+      "loss": 0.1304,
+      "step": 2760
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.217371664738559e-06,
+      "loss": 0.1307,
+      "step": 2761
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2109492563658153e-06,
+      "loss": 0.1305,
+      "step": 2762
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2045320383155053e-06,
+      "loss": 0.1138,
+      "step": 2763
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1981200154936853e-06,
+      "loss": 0.1229,
+      "step": 2764
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1917131928024314e-06,
+      "loss": 0.1214,
+      "step": 2765
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.185311575139842e-06,
+      "loss": 0.1436,
+      "step": 2766
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1789151674000518e-06,
+      "loss": 0.1207,
+      "step": 2767
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.172523974473195e-06,
+      "loss": 0.1199,
+      "step": 2768
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.166138001245426e-06,
+      "loss": 0.1288,
+      "step": 2769
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1597572525989105e-06,
+      "loss": 0.1161,
+      "step": 2770
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.153381733411822e-06,
+      "loss": 0.1234,
+      "step": 2771
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1470114485583304e-06,
+      "loss": 0.1295,
+      "step": 2772
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.140646402908604e-06,
+      "loss": 0.1212,
+      "step": 2773
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1342866013288053e-06,
+      "loss": 0.1241,
+      "step": 2774
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1279320486810928e-06,
+      "loss": 0.1149,
+      "step": 2775
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.121582749823613e-06,
+      "loss": 0.1294,
+      "step": 2776
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1152387096104875e-06,
+      "loss": 0.1271,
+      "step": 2777
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.10889993289182e-06,
+      "loss": 0.1216,
+      "step": 2778
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.102566424513698e-06,
+      "loss": 0.1156,
+      "step": 2779
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.09623818931817e-06,
+      "loss": 0.1127,
+      "step": 2780
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0899152321432647e-06,
+      "loss": 0.1182,
+      "step": 2781
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0835975578229627e-06,
+      "loss": 0.1199,
+      "step": 2782
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0772851711872187e-06,
+      "loss": 0.1228,
+      "step": 2783
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0709780770619366e-06,
+      "loss": 0.1161,
+      "step": 2784
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.064676280268973e-06,
+      "loss": 0.1113,
+      "step": 2785
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.058379785626142e-06,
+      "loss": 0.1212,
+      "step": 2786
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0520885979472026e-06,
+      "loss": 0.1292,
+      "step": 2787
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0458027220418494e-06,
+      "loss": 0.1162,
+      "step": 2788
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0395221627157234e-06,
+      "loss": 0.1179,
+      "step": 2789
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0332469247703934e-06,
+      "loss": 0.1367,
+      "step": 2790
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0269770130033715e-06,
+      "loss": 0.1278,
+      "step": 2791
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0207124322080863e-06,
+      "loss": 0.1203,
+      "step": 2792
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0144531871739003e-06,
+      "loss": 0.1172,
+      "step": 2793
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.008199282686086e-06,
+      "loss": 0.1189,
+      "step": 2794
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0019507235258473e-06,
+      "loss": 0.1166,
+      "step": 2795
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9957075144702887e-06,
+      "loss": 0.1225,
+      "step": 2796
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9894696602924254e-06,
+      "loss": 0.1184,
+      "step": 2797
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.983237165761188e-06,
+      "loss": 0.1148,
+      "step": 2798
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9770100356414044e-06,
+      "loss": 0.1285,
+      "step": 2799
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.970788274693801e-06,
+      "loss": 0.1203,
+      "step": 2800
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.964571887674993e-06,
+      "loss": 0.1263,
+      "step": 2801
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.958360879337503e-06,
+      "loss": 0.1268,
+      "step": 2802
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9521552544297227e-06,
+      "loss": 0.1204,
+      "step": 2803
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.945955017695946e-06,
+      "loss": 0.1293,
+      "step": 2804
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9397601738763314e-06,
+      "loss": 0.1208,
+      "step": 2805
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.933570727706929e-06,
+      "loss": 0.1163,
+      "step": 2806
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.927386683919653e-06,
+      "loss": 0.1196,
+      "step": 2807
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.921208047242284e-06,
+      "loss": 0.1413,
+      "step": 2808
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9150348223984802e-06,
+      "loss": 0.125,
+      "step": 2809
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.908867014107759e-06,
+      "loss": 0.1176,
+      "step": 2810
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9027046270854907e-06,
+      "loss": 0.1063,
+      "step": 2811
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.896547666042906e-06,
+      "loss": 0.1087,
+      "step": 2812
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.89039613568708e-06,
+      "loss": 0.1235,
+      "step": 2813
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8842500407209526e-06,
+      "loss": 0.1102,
+      "step": 2814
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8781093858432883e-06,
+      "loss": 0.114,
+      "step": 2815
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8719741757487096e-06,
+      "loss": 0.1236,
+      "step": 2816
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8658444151276623e-06,
+      "loss": 0.1197,
+      "step": 2817
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8597201086664383e-06,
+      "loss": 0.118,
+      "step": 2818
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.853601261047151e-06,
+      "loss": 0.12,
+      "step": 2819
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.84748787694774e-06,
+      "loss": 0.1254,
+      "step": 2820
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8413799610419756e-06,
+      "loss": 0.1286,
+      "step": 2821
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8352775179994454e-06,
+      "loss": 0.1179,
+      "step": 2822
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8291805524855474e-06,
+      "loss": 0.117,
+      "step": 2823
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8230890691614963e-06,
+      "loss": 0.1278,
+      "step": 2824
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.817003072684311e-06,
+      "loss": 0.1081,
+      "step": 2825
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.810922567706823e-06,
+      "loss": 0.1159,
+      "step": 2826
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.804847558877665e-06,
+      "loss": 0.1288,
+      "step": 2827
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.79877805084126e-06,
+      "loss": 0.1218,
+      "step": 2828
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.792714048237828e-06,
+      "loss": 0.1227,
+      "step": 2829
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.786655555703389e-06,
+      "loss": 0.11,
+      "step": 2830
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7806025778697364e-06,
+      "loss": 0.1119,
+      "step": 2831
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7745551193644613e-06,
+      "loss": 0.1147,
+      "step": 2832
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7685131848109225e-06,
+      "loss": 0.1213,
+      "step": 2833
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.762476778828269e-06,
+      "loss": 0.1266,
+      "step": 2834
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.756445906031411e-06,
+      "loss": 0.1288,
+      "step": 2835
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7504205710310363e-06,
+      "loss": 0.1152,
+      "step": 2836
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.744400778433587e-06,
+      "loss": 0.1259,
+      "step": 2837
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7383865328412907e-06,
+      "loss": 0.1254,
+      "step": 2838
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.732377838852115e-06,
+      "loss": 0.1216,
+      "step": 2839
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7263747010597873e-06,
+      "loss": 0.1204,
+      "step": 2840
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.720377124053788e-06,
+      "loss": 0.1369,
+      "step": 2841
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.714385112419352e-06,
+      "loss": 0.1139,
+      "step": 2842
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.708398670737448e-06,
+      "loss": 0.1194,
+      "step": 2843
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7024178035847992e-06,
+      "loss": 0.1043,
+      "step": 2844
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6964425155338547e-06,
+      "loss": 0.1238,
+      "step": 2845
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.69047281115281e-06,
+      "loss": 0.1239,
+      "step": 2846
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6845086950055834e-06,
+      "loss": 0.1219,
+      "step": 2847
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6785501716518192e-06,
+      "loss": 0.1309,
+      "step": 2848
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6725972456468953e-06,
+      "loss": 0.1312,
+      "step": 2849
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.666649921541906e-06,
+      "loss": 0.1116,
+      "step": 2850
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6607082038836616e-06,
+      "loss": 0.1277,
+      "step": 2851
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6547720972146816e-06,
+      "loss": 0.1197,
+      "step": 2852
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.648841606073208e-06,
+      "loss": 0.1156,
+      "step": 2853
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.642916734993177e-06,
+      "loss": 0.1303,
+      "step": 2854
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6369974885042393e-06,
+      "loss": 0.1181,
+      "step": 2855
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6310838711317345e-06,
+      "loss": 0.1279,
+      "step": 2856
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6251758873967093e-06,
+      "loss": 0.1154,
+      "step": 2857
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6192735418158953e-06,
+      "loss": 0.132,
+      "step": 2858
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6133768389017177e-06,
+      "loss": 0.1149,
+      "step": 2859
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.607485783162281e-06,
+      "loss": 0.1154,
+      "step": 2860
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.601600379101389e-06,
+      "loss": 0.1268,
+      "step": 2861
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5957206312185078e-06,
+      "loss": 0.1316,
+      "step": 2862
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.589846544008787e-06,
+      "loss": 0.1268,
+      "step": 2863
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.583978121963042e-06,
+      "loss": 0.1287,
+      "step": 2864
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5781153695677684e-06,
+      "loss": 0.1266,
+      "step": 2865
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.572258291305115e-06,
+      "loss": 0.1216,
+      "step": 2866
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.566406891652905e-06,
+      "loss": 0.1313,
+      "step": 2867
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.560561175084607e-06,
+      "loss": 0.1234,
+      "step": 2868
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.554721146069359e-06,
+      "loss": 0.1112,
+      "step": 2869
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.548886809071938e-06,
+      "loss": 0.1377,
+      "step": 2870
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.543058168552772e-06,
+      "loss": 0.1191,
+      "step": 2871
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5372352289679413e-06,
+      "loss": 0.1361,
+      "step": 2872
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.531417994769165e-06,
+      "loss": 0.1095,
+      "step": 2873
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5256064704037963e-06,
+      "loss": 0.106,
+      "step": 2874
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5198006603148238e-06,
+      "loss": 0.1256,
+      "step": 2875
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5140005689408676e-06,
+      "loss": 0.1299,
+      "step": 2876
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.50820620071618e-06,
+      "loss": 0.1238,
+      "step": 2877
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.502417560070636e-06,
+      "loss": 0.1175,
+      "step": 2878
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4966346514297303e-06,
+      "loss": 0.1296,
+      "step": 2879
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.49085747921457e-06,
+      "loss": 0.1176,
+      "step": 2880
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4850860478418928e-06,
+      "loss": 0.1209,
+      "step": 2881
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.479320361724027e-06,
+      "loss": 0.1242,
+      "step": 2882
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.473560425268926e-06,
+      "loss": 0.1207,
+      "step": 2883
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4678062428801353e-06,
+      "loss": 0.1294,
+      "step": 2884
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4620578189568124e-06,
+      "loss": 0.1164,
+      "step": 2885
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4563151578937006e-06,
+      "loss": 0.1166,
+      "step": 2886
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.450578264081146e-06,
+      "loss": 0.1264,
+      "step": 2887
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.444847141905079e-06,
+      "loss": 0.1261,
+      "step": 2888
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.439121795747024e-06,
+      "loss": 0.1243,
+      "step": 2889
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4334022299840878e-06,
+      "loss": 0.1294,
+      "step": 2890
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4276884489889564e-06,
+      "loss": 0.1292,
+      "step": 2891
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.42198045712989e-06,
+      "loss": 0.1321,
+      "step": 2892
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4162782587707324e-06,
+      "loss": 0.1089,
+      "step": 2893
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.410581858270886e-06,
+      "loss": 0.1216,
+      "step": 2894
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4048912599853337e-06,
+      "loss": 0.1374,
+      "step": 2895
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.399206468264611e-06,
+      "loss": 0.1263,
+      "step": 2896
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.393527487454823e-06,
+      "loss": 0.1254,
+      "step": 2897
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.387854321897627e-06,
+      "loss": 0.1071,
+      "step": 2898
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.382186975930233e-06,
+      "loss": 0.1264,
+      "step": 2899
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3765254538854064e-06,
+      "loss": 0.1271,
+      "step": 2900
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.370869760091463e-06,
+      "loss": 0.1159,
+      "step": 2901
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3652198988722553e-06,
+      "loss": 0.1193,
+      "step": 2902
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.359575874547174e-06,
+      "loss": 0.133,
+      "step": 2903
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3539376914311627e-06,
+      "loss": 0.1226,
+      "step": 2904
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3483053538346822e-06,
+      "loss": 0.1213,
+      "step": 2905
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.342678866063737e-06,
+      "loss": 0.1193,
+      "step": 2906
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.33705823241985e-06,
+      "loss": 0.1228,
+      "step": 2907
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3314434572000766e-06,
+      "loss": 0.1217,
+      "step": 2908
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.325834544696989e-06,
+      "loss": 0.1187,
+      "step": 2909
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3202314991986752e-06,
+      "loss": 0.1285,
+      "step": 2910
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3146343249887393e-06,
+      "loss": 0.1129,
+      "step": 2911
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3090430263463004e-06,
+      "loss": 0.119,
+      "step": 2912
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.303457607545985e-06,
+      "loss": 0.1267,
+      "step": 2913
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2978780728579196e-06,
+      "loss": 0.1175,
+      "step": 2914
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2923044265477332e-06,
+      "loss": 0.1206,
+      "step": 2915
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2867366728765606e-06,
+      "loss": 0.1297,
+      "step": 2916
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2811748161010184e-06,
+      "loss": 0.1297,
+      "step": 2917
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2756188604732298e-06,
+      "loss": 0.1116,
+      "step": 2918
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.270068810240793e-06,
+      "loss": 0.1222,
+      "step": 2919
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2645246696468038e-06,
+      "loss": 0.132,
+      "step": 2920
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.258986442929829e-06,
+      "loss": 0.1223,
+      "step": 2921
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.253454134323918e-06,
+      "loss": 0.1323,
+      "step": 2922
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.247927748058598e-06,
+      "loss": 0.1232,
+      "step": 2923
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2424072883588723e-06,
+      "loss": 0.114,
+      "step": 2924
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2368927594452016e-06,
+      "loss": 0.1189,
+      "step": 2925
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2313841655335223e-06,
+      "loss": 0.1152,
+      "step": 2926
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.225881510835226e-06,
+      "loss": 0.1176,
+      "step": 2927
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.22038479955717e-06,
+      "loss": 0.1031,
+      "step": 2928
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.214894035901669e-06,
+      "loss": 0.1251,
+      "step": 2929
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.209409224066483e-06,
+      "loss": 0.1261,
+      "step": 2930
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2039303682448254e-06,
+      "loss": 0.1366,
+      "step": 2931
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1984574726253603e-06,
+      "loss": 0.118,
+      "step": 2932
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1929905413921906e-06,
+      "loss": 0.1266,
+      "step": 2933
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.187529578724854e-06,
+      "loss": 0.1115,
+      "step": 2934
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1820745887983374e-06,
+      "loss": 0.1322,
+      "step": 2935
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.176625575783058e-06,
+      "loss": 0.1249,
+      "step": 2936
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1711825438448564e-06,
+      "loss": 0.116,
+      "step": 2937
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1657454971450066e-06,
+      "loss": 0.1424,
+      "step": 2938
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.160314439840202e-06,
+      "loss": 0.1288,
+      "step": 2939
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.154889376082564e-06,
+      "loss": 0.1164,
+      "step": 2940
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.149470310019629e-06,
+      "loss": 0.1026,
+      "step": 2941
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1440572457943477e-06,
+      "loss": 0.1241,
+      "step": 2942
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1386501875450773e-06,
+      "loss": 0.1292,
+      "step": 2943
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.133249139405593e-06,
+      "loss": 0.1237,
+      "step": 2944
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.127854105505066e-06,
+      "loss": 0.1143,
+      "step": 2945
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1224650899680755e-06,
+      "loss": 0.1128,
+      "step": 2946
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.117082096914601e-06,
+      "loss": 0.1203,
+      "step": 2947
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1117051304600124e-06,
+      "loss": 0.1169,
+      "step": 2948
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1063341947150726e-06,
+      "loss": 0.1138,
+      "step": 2949
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.100969293785935e-06,
+      "loss": 0.1271,
+      "step": 2950
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.095610431774142e-06,
+      "loss": 0.106,
+      "step": 2951
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0902576127766196e-06,
+      "loss": 0.1305,
+      "step": 2952
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.084910840885668e-06,
+      "loss": 0.1183,
+      "step": 2953
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.079570120188965e-06,
+      "loss": 0.1276,
+      "step": 2954
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.074235454769572e-06,
+      "loss": 0.1235,
+      "step": 2955
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0689068487059093e-06,
+      "loss": 0.1047,
+      "step": 2956
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0635843060717687e-06,
+      "loss": 0.1303,
+      "step": 2957
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.058267830936307e-06,
+      "loss": 0.1277,
+      "step": 2958
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0529574273640475e-06,
+      "loss": 0.1212,
+      "step": 2959
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0476530994148625e-06,
+      "loss": 0.1127,
+      "step": 2960
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.042354851143984e-06,
+      "loss": 0.1182,
+      "step": 2961
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0370626866019926e-06,
+      "loss": 0.1235,
+      "step": 2962
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0317766098348233e-06,
+      "loss": 0.1174,
+      "step": 2963
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.026496624883758e-06,
+      "loss": 0.1239,
+      "step": 2964
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0212227357854132e-06,
+      "loss": 0.1304,
+      "step": 2965
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0159549465717478e-06,
+      "loss": 0.1229,
+      "step": 2966
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0106932612700647e-06,
+      "loss": 0.1022,
+      "step": 2967
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0054376839029887e-06,
+      "loss": 0.1211,
+      "step": 2968
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0001882184884858e-06,
+      "loss": 0.1293,
+      "step": 2969
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.99494486903984e-06,
+      "loss": 0.1165,
+      "step": 2970
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9897076395656677e-06,
+      "loss": 0.1235,
+      "step": 2971
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9844765340699e-06,
+      "loss": 0.1256,
+      "step": 2972
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.979251556551789e-06,
+      "loss": 0.1185,
+      "step": 2973
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.974032711005901e-06,
+      "loss": 0.1148,
+      "step": 2974
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9688200014221194e-06,
+      "loss": 0.1176,
+      "step": 2975
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9636134317856292e-06,
+      "loss": 0.1204,
+      "step": 2976
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9584130060769247e-06,
+      "loss": 0.1196,
+      "step": 2977
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9532187282717984e-06,
+      "loss": 0.1192,
+      "step": 2978
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9480306023413544e-06,
+      "loss": 0.1159,
+      "step": 2979
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.94284863225198e-06,
+      "loss": 0.1323,
+      "step": 2980
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9376728219653673e-06,
+      "loss": 0.1245,
+      "step": 2981
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9325031754384894e-06,
+      "loss": 0.1239,
+      "step": 2982
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.927339696623618e-06,
+      "loss": 0.1248,
+      "step": 2983
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.922182389468299e-06,
+      "loss": 0.1218,
+      "step": 2984
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9170312579153637e-06,
+      "loss": 0.1132,
+      "step": 2985
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9118863059029236e-06,
+      "loss": 0.1306,
+      "step": 2986
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9067475373643695e-06,
+      "loss": 0.1212,
+      "step": 2987
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.901614956228357e-06,
+      "loss": 0.1209,
+      "step": 2988
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8964885664188148e-06,
+      "loss": 0.1253,
+      "step": 2989
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.891368371854937e-06,
+      "loss": 0.1234,
+      "step": 2990
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.886254376451182e-06,
+      "loss": 0.1182,
+      "step": 2991
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.881146584117275e-06,
+      "loss": 0.1277,
+      "step": 2992
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8760449987581897e-06,
+      "loss": 0.1139,
+      "step": 2993
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8709496242741553e-06,
+      "loss": 0.1142,
+      "step": 2994
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8658604645606614e-06,
+      "loss": 0.1194,
+      "step": 2995
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8607775235084346e-06,
+      "loss": 0.1212,
+      "step": 2996
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8557008050034552e-06,
+      "loss": 0.1248,
+      "step": 2997
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8506303129269477e-06,
+      "loss": 0.1262,
+      "step": 2998
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8455660511553697e-06,
+      "loss": 0.1197,
+      "step": 2999
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.84050802356042e-06,
+      "loss": 0.1309,
+      "step": 3000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8354562340090255e-06,
+      "loss": 0.1213,
+      "step": 3001
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8304106863633552e-06,
+      "loss": 0.1179,
+      "step": 3002
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.825371384480793e-06,
+      "loss": 0.1185,
+      "step": 3003
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8203383322139613e-06,
+      "loss": 0.1182,
+      "step": 3004
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8153115334106918e-06,
+      "loss": 0.1284,
+      "step": 3005
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8102909919140454e-06,
+      "loss": 0.1173,
+      "step": 3006
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8052767115622937e-06,
+      "loss": 0.1259,
+      "step": 3007
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8002686961889194e-06,
+      "loss": 0.1154,
+      "step": 3008
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7952669496226216e-06,
+      "loss": 0.1157,
+      "step": 3009
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7902714756873063e-06,
+      "loss": 0.1246,
+      "step": 3010
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7852822782020806e-06,
+      "loss": 0.1231,
+      "step": 3011
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7802993609812535e-06,
+      "loss": 0.1277,
+      "step": 3012
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.775322727834331e-06,
+      "loss": 0.1222,
+      "step": 3013
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7703523825660207e-06,
+      "loss": 0.1152,
+      "step": 3014
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7653883289762208e-06,
+      "loss": 0.1248,
+      "step": 3015
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7604305708600178e-06,
+      "loss": 0.1218,
+      "step": 3016
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.755479112007682e-06,
+      "loss": 0.1299,
+      "step": 3017
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7505339562046763e-06,
+      "loss": 0.12,
+      "step": 3018
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7455951072316368e-06,
+      "loss": 0.1197,
+      "step": 3019
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.740662568864384e-06,
+      "loss": 0.119,
+      "step": 3020
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7357363448739083e-06,
+      "loss": 0.1206,
+      "step": 3021
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7308164390263794e-06,
+      "loss": 0.1168,
+      "step": 3022
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7259028550831314e-06,
+      "loss": 0.1231,
+      "step": 3023
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7209955968006653e-06,
+      "loss": 0.1158,
+      "step": 3024
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7160946679306423e-06,
+      "loss": 0.1201,
+      "step": 3025
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7112000722199006e-06,
+      "loss": 0.1116,
+      "step": 3026
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7063118134104205e-06,
+      "loss": 0.1165,
+      "step": 3027
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7014298952393426e-06,
+      "loss": 0.1183,
+      "step": 3028
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6965543214389591e-06,
+      "loss": 0.1318,
+      "step": 3029
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.691685095736717e-06,
+      "loss": 0.1122,
+      "step": 3030
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6868222218552e-06,
+      "loss": 0.1316,
+      "step": 3031
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6819657035121495e-06,
+      "loss": 0.1279,
+      "step": 3032
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6771155444204335e-06,
+      "loss": 0.1247,
+      "step": 3033
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6722717482880713e-06,
+      "loss": 0.1243,
+      "step": 3034
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6674343188182075e-06,
+      "loss": 0.1163,
+      "step": 3035
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6626032597091225e-06,
+      "loss": 0.126,
+      "step": 3036
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6577785746542286e-06,
+      "loss": 0.1207,
+      "step": 3037
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6529602673420663e-06,
+      "loss": 0.1225,
+      "step": 3038
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.648148341456296e-06,
+      "loss": 0.1306,
+      "step": 3039
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6433428006757002e-06,
+      "loss": 0.1232,
+      "step": 3040
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.638543648674179e-06,
+      "loss": 0.124,
+      "step": 3041
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6337508891207521e-06,
+      "loss": 0.1208,
+      "step": 3042
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6289645256795516e-06,
+      "loss": 0.1247,
+      "step": 3043
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6241845620098162e-06,
+      "loss": 0.116,
+      "step": 3044
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6194110017658926e-06,
+      "loss": 0.1222,
+      "step": 3045
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.614643848597236e-06,
+      "loss": 0.1131,
+      "step": 3046
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6098831061483978e-06,
+      "loss": 0.131,
+      "step": 3047
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6051287780590275e-06,
+      "loss": 0.1308,
+      "step": 3048
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6003808679638832e-06,
+      "loss": 0.1235,
+      "step": 3049
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5956393794928027e-06,
+      "loss": 0.1221,
+      "step": 3050
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5909043162707194e-06,
+      "loss": 0.1316,
+      "step": 3051
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.58617568191765e-06,
+      "loss": 0.1223,
+      "step": 3052
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.581453480048707e-06,
+      "loss": 0.1223,
+      "step": 3053
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5767377142740724e-06,
+      "loss": 0.1255,
+      "step": 3054
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5720283881990206e-06,
+      "loss": 0.1127,
+      "step": 3055
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5673255054238878e-06,
+      "loss": 0.1177,
+      "step": 3056
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5626290695441005e-06,
+      "loss": 0.1134,
+      "step": 3057
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.557939084150144e-06,
+      "loss": 0.1068,
+      "step": 3058
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.553255552827575e-06,
+      "loss": 0.1175,
+      "step": 3059
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5485784791570203e-06,
+      "loss": 0.1147,
+      "step": 3060
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.543907866714167e-06,
+      "loss": 0.1197,
+      "step": 3061
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5392437190697628e-06,
+      "loss": 0.1165,
+      "step": 3062
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5345860397896117e-06,
+      "loss": 0.1175,
+      "step": 3063
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5299348324345697e-06,
+      "loss": 0.1241,
+      "step": 3064
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5252901005605513e-06,
+      "loss": 0.1127,
+      "step": 3065
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5206518477185218e-06,
+      "loss": 0.1258,
+      "step": 3066
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5160200774544864e-06,
+      "loss": 0.1112,
+      "step": 3067
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.511394793309493e-06,
+      "loss": 0.1167,
+      "step": 3068
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5067759988196406e-06,
+      "loss": 0.1163,
+      "step": 3069
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5021636975160558e-06,
+      "loss": 0.1206,
+      "step": 3070
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4975578929249123e-06,
+      "loss": 0.1225,
+      "step": 3071
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4929585885674035e-06,
+      "loss": 0.1167,
+      "step": 3072
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4883657879597691e-06,
+      "loss": 0.1187,
+      "step": 3073
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4837794946132623e-06,
+      "loss": 0.1145,
+      "step": 3074
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4791997120341695e-06,
+      "loss": 0.1226,
+      "step": 3075
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4746264437237934e-06,
+      "loss": 0.1189,
+      "step": 3076
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4700596931784638e-06,
+      "loss": 0.1225,
+      "step": 3077
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.465499463889527e-06,
+      "loss": 0.1183,
+      "step": 3078
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4609457593433373e-06,
+      "loss": 0.1256,
+      "step": 3079
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4563985830212624e-06,
+      "loss": 0.1124,
+      "step": 3080
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4518579383996855e-06,
+      "loss": 0.1166,
+      "step": 3081
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.447323828949987e-06,
+      "loss": 0.119,
+      "step": 3082
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4427962581385612e-06,
+      "loss": 0.1155,
+      "step": 3083
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.438275229426792e-06,
+      "loss": 0.1133,
+      "step": 3084
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4337607462710745e-06,
+      "loss": 0.1183,
+      "step": 3085
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4292528121227888e-06,
+      "loss": 0.1183,
+      "step": 3086
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4247514304283106e-06,
+      "loss": 0.1159,
+      "step": 3087
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4202566046290112e-06,
+      "loss": 0.1201,
+      "step": 3088
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4157683381612486e-06,
+      "loss": 0.1108,
+      "step": 3089
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4112866344563603e-06,
+      "loss": 0.125,
+      "step": 3090
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.406811496940672e-06,
+      "loss": 0.1213,
+      "step": 3091
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.402342929035485e-06,
+      "loss": 0.1111,
+      "step": 3092
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3978809341570821e-06,
+      "loss": 0.1057,
+      "step": 3093
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3934255157167243e-06,
+      "loss": 0.1074,
+      "step": 3094
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3889766771206348e-06,
+      "loss": 0.1176,
+      "step": 3095
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3845344217700108e-06,
+      "loss": 0.1237,
+      "step": 3096
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3800987530610232e-06,
+      "loss": 0.1058,
+      "step": 3097
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3756696743847975e-06,
+      "loss": 0.1122,
+      "step": 3098
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3712471891274237e-06,
+      "loss": 0.1191,
+      "step": 3099
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3668313006699542e-06,
+      "loss": 0.118,
+      "step": 3100
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.362422012388399e-06,
+      "loss": 0.1194,
+      "step": 3101
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3580193276537179e-06,
+      "loss": 0.1224,
+      "step": 3102
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3536232498318202e-06,
+      "loss": 0.1196,
+      "step": 3103
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.349233782283572e-06,
+      "loss": 0.1332,
+      "step": 3104
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3448509283647793e-06,
+      "loss": 0.1161,
+      "step": 3105
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3404746914261958e-06,
+      "loss": 0.1151,
+      "step": 3106
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3361050748135118e-06,
+      "loss": 0.1199,
+      "step": 3107
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3317420818673632e-06,
+      "loss": 0.1098,
+      "step": 3108
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.327385715923316e-06,
+      "loss": 0.1281,
+      "step": 3109
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3230359803118699e-06,
+      "loss": 0.1201,
+      "step": 3110
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3186928783584573e-06,
+      "loss": 0.1144,
+      "step": 3111
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.314356413383444e-06,
+      "loss": 0.1186,
+      "step": 3112
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3100265887021146e-06,
+      "loss": 0.1273,
+      "step": 3113
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3057034076246788e-06,
+      "loss": 0.1087,
+      "step": 3114
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3013868734562652e-06,
+      "loss": 0.1185,
+      "step": 3115
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.297076989496925e-06,
+      "loss": 0.1188,
+      "step": 3116
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2927737590416278e-06,
+      "loss": 0.1153,
+      "step": 3117
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2884771853802491e-06,
+      "loss": 0.1262,
+      "step": 3118
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2841872717975768e-06,
+      "loss": 0.1129,
+      "step": 3119
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2799040215733116e-06,
+      "loss": 0.1224,
+      "step": 3120
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2756274379820566e-06,
+      "loss": 0.1305,
+      "step": 3121
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2713575242933151e-06,
+      "loss": 0.1192,
+      "step": 3122
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2670942837714984e-06,
+      "loss": 0.1225,
+      "step": 3123
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2628377196759123e-06,
+      "loss": 0.1143,
+      "step": 3124
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2585878352607584e-06,
+      "loss": 0.1033,
+      "step": 3125
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2543446337751309e-06,
+      "loss": 0.1295,
+      "step": 3126
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2501081184630127e-06,
+      "loss": 0.1234,
+      "step": 3127
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2458782925632805e-06,
+      "loss": 0.1362,
+      "step": 3128
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.241655159309697e-06,
+      "loss": 0.1223,
+      "step": 3129
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2374387219309025e-06,
+      "loss": 0.1231,
+      "step": 3130
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2332289836504208e-06,
+      "loss": 0.115,
+      "step": 3131
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2290259476866562e-06,
+      "loss": 0.1219,
+      "step": 3132
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2248296172528839e-06,
+      "loss": 0.1268,
+      "step": 3133
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2206399955572623e-06,
+      "loss": 0.1111,
+      "step": 3134
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2164570858028079e-06,
+      "loss": 0.1244,
+      "step": 3135
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2122808911874184e-06,
+      "loss": 0.1203,
+      "step": 3136
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2081114149038486e-06,
+      "loss": 0.1164,
+      "step": 3137
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.20394866013972e-06,
+      "loss": 0.1174,
+      "step": 3138
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1997926300775165e-06,
+      "loss": 0.128,
+      "step": 3139
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1956433278945811e-06,
+      "loss": 0.111,
+      "step": 3140
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.191500756763112e-06,
+      "loss": 0.1208,
+      "step": 3141
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1873649198501625e-06,
+      "loss": 0.1079,
+      "step": 3142
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.183235820317632e-06,
+      "loss": 0.1131,
+      "step": 3143
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1791134613222788e-06,
+      "loss": 0.1257,
+      "step": 3144
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1749978460156998e-06,
+      "loss": 0.1274,
+      "step": 3145
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1708889775443421e-06,
+      "loss": 0.1233,
+      "step": 3146
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.166786859049489e-06,
+      "loss": 0.1211,
+      "step": 3147
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1626914936672695e-06,
+      "loss": 0.1158,
+      "step": 3148
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1586028845286456e-06,
+      "loss": 0.12,
+      "step": 3149
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1545210347594127e-06,
+      "loss": 0.1119,
+      "step": 3150
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1504459474802022e-06,
+      "loss": 0.1139,
+      "step": 3151
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1463776258064774e-06,
+      "loss": 0.1235,
+      "step": 3152
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1423160728485228e-06,
+      "loss": 0.1296,
+      "step": 3153
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.138261291711451e-06,
+      "loss": 0.1141,
+      "step": 3154
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1342132854952003e-06,
+      "loss": 0.1258,
+      "step": 3155
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.130172057294523e-06,
+      "loss": 0.1268,
+      "step": 3156
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1261376101989985e-06,
+      "loss": 0.1303,
+      "step": 3157
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.122109947293013e-06,
+      "loss": 0.1241,
+      "step": 3158
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1180890716557735e-06,
+      "loss": 0.1269,
+      "step": 3159
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1140749863612921e-06,
+      "loss": 0.1309,
+      "step": 3160
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1100676944783905e-06,
+      "loss": 0.1186,
+      "step": 3161
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1060671990707006e-06,
+      "loss": 0.116,
+      "step": 3162
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1020735031966579e-06,
+      "loss": 0.1257,
+      "step": 3163
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0980866099094956e-06,
+      "loss": 0.1216,
+      "step": 3164
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.094106522257249e-06,
+      "loss": 0.1191,
+      "step": 3165
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.090133243282746e-06,
+      "loss": 0.1236,
+      "step": 3166
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0861667760236194e-06,
+      "loss": 0.1204,
+      "step": 3167
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0822071235122812e-06,
+      "loss": 0.111,
+      "step": 3168
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.078254288775945e-06,
+      "loss": 0.1112,
+      "step": 3169
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.074308274836604e-06,
+      "loss": 0.1248,
+      "step": 3170
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0703690847110437e-06,
+      "loss": 0.1136,
+      "step": 3171
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0664367214108262e-06,
+      "loss": 0.1287,
+      "step": 3172
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0625111879422966e-06,
+      "loss": 0.1135,
+      "step": 3173
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.05859248730658e-06,
+      "loss": 0.1175,
+      "step": 3174
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0546806224995786e-06,
+      "loss": 0.11,
+      "step": 3175
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0507755965119649e-06,
+      "loss": 0.1104,
+      "step": 3176
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.046877412329186e-06,
+      "loss": 0.1127,
+      "step": 3177
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0429860729314534e-06,
+      "loss": 0.125,
+      "step": 3178
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0391015812937521e-06,
+      "loss": 0.1208,
+      "step": 3179
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0352239403858323e-06,
+      "loss": 0.1234,
+      "step": 3180
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0313531531721999e-06,
+      "loss": 0.1251,
+      "step": 3181
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0274892226121246e-06,
+      "loss": 0.1205,
+      "step": 3182
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.023632151659637e-06,
+      "loss": 0.1211,
+      "step": 3183
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.019781943263518e-06,
+      "loss": 0.1238,
+      "step": 3184
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0159386003673077e-06,
+      "loss": 0.1104,
+      "step": 3185
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.01210212590929e-06,
+      "loss": 0.1214,
+      "step": 3186
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0082725228225077e-06,
+      "loss": 0.1217,
+      "step": 3187
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.004449794034742e-06,
+      "loss": 0.1098,
+      "step": 3188
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0006339424685218e-06,
+      "loss": 0.1165,
+      "step": 3189
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.96824971041116e-07,
+      "loss": 0.1217,
+      "step": 3190
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.930228826645383e-07,
+      "loss": 0.1175,
+      "step": 3191
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.892276802455369e-07,
+      "loss": 0.1224,
+      "step": 3192
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.854393666855965e-07,
+      "loss": 0.1095,
+      "step": 3193
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.816579448809327e-07,
+      "loss": 0.1293,
+      "step": 3194
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.77883417722496e-07,
+      "loss": 0.1047,
+      "step": 3195
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.74115788095963e-07,
+      "loss": 0.1261,
+      "step": 3196
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.703550588817413e-07,
+      "loss": 0.1225,
+      "step": 3197
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.666012329549558e-07,
+      "loss": 0.1224,
+      "step": 3198
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.628543131854618e-07,
+      "loss": 0.1328,
+      "step": 3199
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.59114302437829e-07,
+      "loss": 0.1165,
+      "step": 3200
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.55381203571345e-07,
+      "loss": 0.1196,
+      "step": 3201
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.516550194400165e-07,
+      "loss": 0.129,
+      "step": 3202
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.479357528925648e-07,
+      "loss": 0.1235,
+      "step": 3203
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.442234067724188e-07,
+      "loss": 0.1116,
+      "step": 3204
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.405179839177159e-07,
+      "loss": 0.1264,
+      "step": 3205
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.368194871613057e-07,
+      "loss": 0.1244,
+      "step": 3206
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.331279193307374e-07,
+      "loss": 0.1155,
+      "step": 3207
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.29443283248267e-07,
+      "loss": 0.125,
+      "step": 3208
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.257655817308475e-07,
+      "loss": 0.1158,
+      "step": 3209
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.220948175901345e-07,
+      "loss": 0.1267,
+      "step": 3210
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.184309936324765e-07,
+      "loss": 0.1227,
+      "step": 3211
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.147741126589171e-07,
+      "loss": 0.1155,
+      "step": 3212
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.11124177465188e-07,
+      "loss": 0.1162,
+      "step": 3213
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.074811908417225e-07,
+      "loss": 0.119,
+      "step": 3214
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.038451555736294e-07,
+      "loss": 0.1124,
+      "step": 3215
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.002160744407084e-07,
+      "loss": 0.1029,
+      "step": 3216
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.965939502174392e-07,
+      "loss": 0.1117,
+      "step": 3217
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.929787856729899e-07,
+      "loss": 0.1095,
+      "step": 3218
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.893705835711997e-07,
+      "loss": 0.1179,
+      "step": 3219
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.857693466705919e-07,
+      "loss": 0.1085,
+      "step": 3220
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.821750777243576e-07,
+      "loss": 0.1193,
+      "step": 3221
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.785877794803688e-07,
+      "loss": 0.1096,
+      "step": 3222
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.750074546811626e-07,
+      "loss": 0.115,
+      "step": 3223
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.714341060639442e-07,
+      "loss": 0.105,
+      "step": 3224
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.678677363605892e-07,
+      "loss": 0.1188,
+      "step": 3225
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.643083482976378e-07,
+      "loss": 0.1293,
+      "step": 3226
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.60755944596291e-07,
+      "loss": 0.1339,
+      "step": 3227
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.572105279724085e-07,
+      "loss": 0.1154,
+      "step": 3228
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.536721011365079e-07,
+      "loss": 0.131,
+      "step": 3229
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.501406667937684e-07,
+      "loss": 0.1144,
+      "step": 3230
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.466162276440204e-07,
+      "loss": 0.1268,
+      "step": 3231
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.430987863817452e-07,
+      "loss": 0.1155,
+      "step": 3232
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.39588345696073e-07,
+      "loss": 0.1191,
+      "step": 3233
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.360849082707856e-07,
+      "loss": 0.1288,
+      "step": 3234
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.325884767843095e-07,
+      "loss": 0.1165,
+      "step": 3235
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.290990539097121e-07,
+      "loss": 0.1107,
+      "step": 3236
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.256166423147061e-07,
+      "loss": 0.1409,
+      "step": 3237
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.221412446616462e-07,
+      "loss": 0.1359,
+      "step": 3238
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.186728636075181e-07,
+      "loss": 0.109,
+      "step": 3239
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.152115018039475e-07,
+      "loss": 0.1062,
+      "step": 3240
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.117571618971909e-07,
+      "loss": 0.1132,
+      "step": 3241
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.083098465281413e-07,
+      "loss": 0.1149,
+      "step": 3242
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.048695583323196e-07,
+      "loss": 0.1259,
+      "step": 3243
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.014362999398717e-07,
+      "loss": 0.1243,
+      "step": 3244
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.980100739755681e-07,
+      "loss": 0.1103,
+      "step": 3245
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.945908830588101e-07,
+      "loss": 0.1324,
+      "step": 3246
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.911787298036134e-07,
+      "loss": 0.1241,
+      "step": 3247
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.877736168186179e-07,
+      "loss": 0.1159,
+      "step": 3248
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.843755467070758e-07,
+      "loss": 0.1117,
+      "step": 3249
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.809845220668621e-07,
+      "loss": 0.1326,
+      "step": 3250
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.776005454904601e-07,
+      "loss": 0.1247,
+      "step": 3251
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.742236195649655e-07,
+      "loss": 0.1193,
+      "step": 3252
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.708537468720845e-07,
+      "loss": 0.1048,
+      "step": 3253
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.674909299881339e-07,
+      "loss": 0.1195,
+      "step": 3254
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.641351714840306e-07,
+      "loss": 0.1081,
+      "step": 3255
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.607864739252968e-07,
+      "loss": 0.1148,
+      "step": 3256
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.574448398720613e-07,
+      "loss": 0.1138,
+      "step": 3257
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.541102718790472e-07,
+      "loss": 0.1166,
+      "step": 3258
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.507827724955774e-07,
+      "loss": 0.122,
+      "step": 3259
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.474623442655704e-07,
+      "loss": 0.1135,
+      "step": 3260
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.441489897275423e-07,
+      "loss": 0.1082,
+      "step": 3261
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.408427114145944e-07,
+      "loss": 0.1126,
+      "step": 3262
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.375435118544238e-07,
+      "loss": 0.12,
+      "step": 3263
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.342513935693107e-07,
+      "loss": 0.1002,
+      "step": 3264
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.309663590761285e-07,
+      "loss": 0.1148,
+      "step": 3265
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.276884108863302e-07,
+      "loss": 0.1096,
+      "step": 3266
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.244175515059514e-07,
+      "loss": 0.112,
+      "step": 3267
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.211537834356086e-07,
+      "loss": 0.12,
+      "step": 3268
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.178971091705e-07,
+      "loss": 0.1084,
+      "step": 3269
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.146475312003931e-07,
+      "loss": 0.1231,
+      "step": 3270
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.114050520096394e-07,
+      "loss": 0.1124,
+      "step": 3271
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.081696740771549e-07,
+      "loss": 0.1169,
+      "step": 3272
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.049413998764343e-07,
+      "loss": 0.1138,
+      "step": 3273
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.017202318755356e-07,
+      "loss": 0.1151,
+      "step": 3274
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.985061725370845e-07,
+      "loss": 0.1122,
+      "step": 3275
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.952992243182743e-07,
+      "loss": 0.1143,
+      "step": 3276
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.920993896708628e-07,
+      "loss": 0.1067,
+      "step": 3277
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.889066710411663e-07,
+      "loss": 0.1167,
+      "step": 3278
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.857210708700623e-07,
+      "loss": 0.1133,
+      "step": 3279
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.825425915929828e-07,
+      "loss": 0.1141,
+      "step": 3280
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.793712356399229e-07,
+      "loss": 0.1076,
+      "step": 3281
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.762070054354264e-07,
+      "loss": 0.1185,
+      "step": 3282
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.730499033985904e-07,
+      "loss": 0.1075,
+      "step": 3283
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.698999319430621e-07,
+      "loss": 0.1158,
+      "step": 3284
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.667570934770395e-07,
+      "loss": 0.1267,
+      "step": 3285
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.636213904032652e-07,
+      "loss": 0.118,
+      "step": 3286
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.604928251190257e-07,
+      "loss": 0.1115,
+      "step": 3287
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.573714000161525e-07,
+      "loss": 0.1127,
+      "step": 3288
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.542571174810208e-07,
+      "loss": 0.1139,
+      "step": 3289
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.511499798945398e-07,
+      "loss": 0.1204,
+      "step": 3290
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.480499896321591e-07,
+      "loss": 0.1301,
+      "step": 3291
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.449571490638618e-07,
+      "loss": 0.1213,
+      "step": 3292
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.418714605541698e-07,
+      "loss": 0.117,
+      "step": 3293
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.387929264621351e-07,
+      "loss": 0.1199,
+      "step": 3294
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.357215491413371e-07,
+      "loss": 0.1076,
+      "step": 3295
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.326573309398853e-07,
+      "loss": 0.129,
+      "step": 3296
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.296002742004192e-07,
+      "loss": 0.1237,
+      "step": 3297
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.26550381260097e-07,
+      "loss": 0.1233,
+      "step": 3298
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.23507654450608e-07,
+      "loss": 0.1132,
+      "step": 3299
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.204720960981547e-07,
+      "loss": 0.1317,
+      "step": 3300
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.174437085234663e-07,
+      "loss": 0.1255,
+      "step": 3301
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.144224940417854e-07,
+      "loss": 0.1118,
+      "step": 3302
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.114084549628707e-07,
+      "loss": 0.1182,
+      "step": 3303
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.084015935909959e-07,
+      "loss": 0.1293,
+      "step": 3304
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.054019122249522e-07,
+      "loss": 0.1123,
+      "step": 3305
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.024094131580338e-07,
+      "loss": 0.1199,
+      "step": 3306
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.994240986780464e-07,
+      "loss": 0.1089,
+      "step": 3307
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.964459710673065e-07,
+      "loss": 0.1151,
+      "step": 3308
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.934750326026317e-07,
+      "loss": 0.1209,
+      "step": 3309
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.905112855553453e-07,
+      "loss": 0.1163,
+      "step": 3310
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.875547321912722e-07,
+      "loss": 0.1312,
+      "step": 3311
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.846053747707403e-07,
+      "loss": 0.1004,
+      "step": 3312
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.816632155485724e-07,
+      "loss": 0.1363,
+      "step": 3313
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.787282567740904e-07,
+      "loss": 0.1162,
+      "step": 3314
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.758005006911072e-07,
+      "loss": 0.123,
+      "step": 3315
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.728799495379355e-07,
+      "loss": 0.1098,
+      "step": 3316
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.699666055473774e-07,
+      "loss": 0.1082,
+      "step": 3317
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.670604709467231e-07,
+      "loss": 0.1175,
+      "step": 3318
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.641615479577512e-07,
+      "loss": 0.1213,
+      "step": 3319
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.612698387967308e-07,
+      "loss": 0.1136,
+      "step": 3320
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.583853456744104e-07,
+      "loss": 0.1117,
+      "step": 3321
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.55508070796027e-07,
+      "loss": 0.1198,
+      "step": 3322
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.526380163612943e-07,
+      "loss": 0.1378,
+      "step": 3323
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.497751845644105e-07,
+      "loss": 0.1247,
+      "step": 3324
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.469195775940495e-07,
+      "loss": 0.1304,
+      "step": 3325
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.440711976333601e-07,
+      "loss": 0.1159,
+      "step": 3326
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.412300468599696e-07,
+      "loss": 0.1204,
+      "step": 3327
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.383961274459771e-07,
+      "loss": 0.1233,
+      "step": 3328
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.355694415579515e-07,
+      "loss": 0.1142,
+      "step": 3329
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.327499913569345e-07,
+      "loss": 0.1195,
+      "step": 3330
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.299377789984316e-07,
+      "loss": 0.1051,
+      "step": 3331
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.271328066324211e-07,
+      "loss": 0.1142,
+      "step": 3332
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.243350764033405e-07,
+      "loss": 0.1089,
+      "step": 3333
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.215445904500949e-07,
+      "loss": 0.1163,
+      "step": 3334
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.187613509060463e-07,
+      "loss": 0.1108,
+      "step": 3335
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.159853598990227e-07,
+      "loss": 0.1187,
+      "step": 3336
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.132166195513055e-07,
+      "loss": 0.1214,
+      "step": 3337
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.104551319796347e-07,
+      "loss": 0.1209,
+      "step": 3338
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.077008992952049e-07,
+      "loss": 0.1068,
+      "step": 3339
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.049539236036671e-07,
+      "loss": 0.1262,
+      "step": 3340
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.022142070051205e-07,
+      "loss": 0.1102,
+      "step": 3341
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.994817515941153e-07,
+      "loss": 0.1061,
+      "step": 3342
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.967565594596512e-07,
+      "loss": 0.1366,
+      "step": 3343
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.940386326851753e-07,
+      "loss": 0.1222,
+      "step": 3344
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.913279733485821e-07,
+      "loss": 0.1248,
+      "step": 3345
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.886245835222059e-07,
+      "loss": 0.1148,
+      "step": 3346
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.859284652728236e-07,
+      "loss": 0.129,
+      "step": 3347
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.832396206616574e-07,
+      "loss": 0.1061,
+      "step": 3348
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.805580517443642e-07,
+      "loss": 0.1161,
+      "step": 3349
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.778837605710418e-07,
+      "loss": 0.1165,
+      "step": 3350
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.752167491862203e-07,
+      "loss": 0.1128,
+      "step": 3351
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.725570196288687e-07,
+      "loss": 0.1172,
+      "step": 3352
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.69904573932386e-07,
+      "loss": 0.1222,
+      "step": 3353
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.672594141246023e-07,
+      "loss": 0.1272,
+      "step": 3354
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6462154222778e-07,
+      "loss": 0.1111,
+      "step": 3355
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.619909602586059e-07,
+      "loss": 0.1094,
+      "step": 3356
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.593676702282002e-07,
+      "loss": 0.1191,
+      "step": 3357
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5675167414210073e-07,
+      "loss": 0.1122,
+      "step": 3358
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.541429740002734e-07,
+      "loss": 0.1097,
+      "step": 3359
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5154157179710614e-07,
+      "loss": 0.1261,
+      "step": 3360
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4894746952140467e-07,
+      "loss": 0.113,
+      "step": 3361
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.46360669156396e-07,
+      "loss": 0.1232,
+      "step": 3362
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.43781172679727e-07,
+      "loss": 0.1121,
+      "step": 3363
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.412089820634569e-07,
+      "loss": 0.1289,
+      "step": 3364
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3864409927405927e-07,
+      "loss": 0.1222,
+      "step": 3365
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.360865262724223e-07,
+      "loss": 0.1086,
+      "step": 3366
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.335362650138475e-07,
+      "loss": 0.1195,
+      "step": 3367
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3099331744804404e-07,
+      "loss": 0.1198,
+      "step": 3368
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2845768551912916e-07,
+      "loss": 0.1195,
+      "step": 3369
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.259293711656287e-07,
+      "loss": 0.1137,
+      "step": 3370
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.234083763204755e-07,
+      "loss": 0.1103,
+      "step": 3371
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.208947029110011e-07,
+      "loss": 0.1121,
+      "step": 3372
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.183883528589461e-07,
+      "loss": 0.1193,
+      "step": 3373
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.158893280804477e-07,
+      "loss": 0.1105,
+      "step": 3374
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1339763048604764e-07,
+      "loss": 0.121,
+      "step": 3375
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.109132619806799e-07,
+      "loss": 0.1188,
+      "step": 3376
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.084362244636797e-07,
+      "loss": 0.1054,
+      "step": 3377
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.059665198287743e-07,
+      "loss": 0.1139,
+      "step": 3378
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.035041499640879e-07,
+      "loss": 0.1206,
+      "step": 3379
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.010491167521391e-07,
+      "loss": 0.1094,
+      "step": 3380
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.986014220698309e-07,
+      "loss": 0.1134,
+      "step": 3381
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9616106778845863e-07,
+      "loss": 0.1213,
+      "step": 3382
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9372805577370997e-07,
+      "loss": 0.1218,
+      "step": 3383
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9130238788565256e-07,
+      "loss": 0.1163,
+      "step": 3384
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.888840659787463e-07,
+      "loss": 0.1218,
+      "step": 3385
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.864730919018289e-07,
+      "loss": 0.119,
+      "step": 3386
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.840694674981249e-07,
+      "loss": 0.1121,
+      "step": 3387
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8167319460523653e-07,
+      "loss": 0.1227,
+      "step": 3388
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.792842750551473e-07,
+      "loss": 0.1239,
+      "step": 3389
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7690271067421957e-07,
+      "loss": 0.1219,
+      "step": 3390
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.745285032831936e-07,
+      "loss": 0.1115,
+      "step": 3391
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.721616546971818e-07,
+      "loss": 0.1118,
+      "step": 3392
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6980216672567235e-07,
+      "loss": 0.117,
+      "step": 3393
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.674500411725246e-07,
+      "loss": 0.1118,
+      "step": 3394
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6510527983597334e-07,
+      "loss": 0.1211,
+      "step": 3395
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6276788450862133e-07,
+      "loss": 0.1037,
+      "step": 3396
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.604378569774369e-07,
+      "loss": 0.1283,
+      "step": 3397
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5811519902375857e-07,
+      "loss": 0.1121,
+      "step": 3398
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.557999124232925e-07,
+      "loss": 0.1058,
+      "step": 3399
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5349199894610496e-07,
+      "loss": 0.1166,
+      "step": 3400
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.511914603566269e-07,
+      "loss": 0.1296,
+      "step": 3401
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4889829841365243e-07,
+      "loss": 0.1158,
+      "step": 3402
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4661251487033696e-07,
+      "loss": 0.1153,
+      "step": 3403
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.443341114741916e-07,
+      "loss": 0.1251,
+      "step": 3404
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.420630899670863e-07,
+      "loss": 0.1111,
+      "step": 3405
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.39799452085251e-07,
+      "loss": 0.1213,
+      "step": 3406
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.375431995592637e-07,
+      "loss": 0.1163,
+      "step": 3407
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3529433411406466e-07,
+      "loss": 0.1227,
+      "step": 3408
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3305285746893954e-07,
+      "loss": 0.118,
+      "step": 3409
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.308187713375299e-07,
+      "loss": 0.1194,
+      "step": 3410
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.28592077427824e-07,
+      "loss": 0.1201,
+      "step": 3411
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2637277744215987e-07,
+      "loss": 0.129,
+      "step": 3412
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.241608730772228e-07,
+      "loss": 0.1295,
+      "step": 3413
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2195636602404547e-07,
+      "loss": 0.1192,
+      "step": 3414
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.19759257968002e-07,
+      "loss": 0.1226,
+      "step": 3415
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.175695505888132e-07,
+      "loss": 0.1181,
+      "step": 3416
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.153872455605378e-07,
+      "loss": 0.1213,
+      "step": 3417
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1321234455158e-07,
+      "loss": 0.1136,
+      "step": 3418
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.11044849224682e-07,
+      "loss": 0.1121,
+      "step": 3419
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.088847612369217e-07,
+      "loss": 0.1158,
+      "step": 3420
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.06732082239718e-07,
+      "loss": 0.0989,
+      "step": 3421
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.045868138788222e-07,
+      "loss": 0.1148,
+      "step": 3422
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.024489577943224e-07,
+      "loss": 0.1147,
+      "step": 3423
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0031851562063676e-07,
+      "loss": 0.1114,
+      "step": 3424
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9819548898652016e-07,
+      "loss": 0.1162,
+      "step": 3425
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9607987951505434e-07,
+      "loss": 0.1117,
+      "step": 3426
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.939716888236532e-07,
+      "loss": 0.1297,
+      "step": 3427
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9187091852405534e-07,
+      "loss": 0.1233,
+      "step": 3428
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.897775702223282e-07,
+      "loss": 0.1107,
+      "step": 3429
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8769164551886717e-07,
+      "loss": 0.1118,
+      "step": 3430
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8561314600838997e-07,
+      "loss": 0.1117,
+      "step": 3431
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.835420732799377e-07,
+      "loss": 0.1144,
+      "step": 3432
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.814784289168737e-07,
+      "loss": 0.0984,
+      "step": 3433
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.794222144968828e-07,
+      "loss": 0.1181,
+      "step": 3434
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.773734315919674e-07,
+      "loss": 0.1175,
+      "step": 3435
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7533208176845237e-07,
+      "loss": 0.1168,
+      "step": 3436
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7329816658697605e-07,
+      "loss": 0.1192,
+      "step": 3437
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7127168760249676e-07,
+      "loss": 0.1127,
+      "step": 3438
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6925264636428306e-07,
+      "loss": 0.1225,
+      "step": 3439
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6724104441592015e-07,
+      "loss": 0.1102,
+      "step": 3440
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6523688329530563e-07,
+      "loss": 0.1156,
+      "step": 3441
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.632401645346494e-07,
+      "loss": 0.1171,
+      "step": 3442
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.612508896604693e-07,
+      "loss": 0.1162,
+      "step": 3443
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.592690601935932e-07,
+      "loss": 0.1156,
+      "step": 3444
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5729467764915585e-07,
+      "loss": 0.1149,
+      "step": 3445
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5532774353660084e-07,
+      "loss": 0.1193,
+      "step": 3446
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.533682593596776e-07,
+      "loss": 0.1143,
+      "step": 3447
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5141622661643774e-07,
+      "loss": 0.1321,
+      "step": 3448
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.494716467992353e-07,
+      "loss": 0.1235,
+      "step": 3449
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4753452139473e-07,
+      "loss": 0.1054,
+      "step": 3450
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4560485188388163e-07,
+      "loss": 0.1139,
+      "step": 3451
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.436826397419456e-07,
+      "loss": 0.1326,
+      "step": 3452
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4176788643848205e-07,
+      "loss": 0.1135,
+      "step": 3453
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.398605934373477e-07,
+      "loss": 0.1188,
+      "step": 3454
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.379607621966906e-07,
+      "loss": 0.1221,
+      "step": 3455
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.360683941689601e-07,
+      "loss": 0.1166,
+      "step": 3456
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3418349080089663e-07,
+      "loss": 0.1075,
+      "step": 3457
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.323060535335353e-07,
+      "loss": 0.1108,
+      "step": 3458
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.304360838022024e-07,
+      "loss": 0.1068,
+      "step": 3459
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2857358303651433e-07,
+      "loss": 0.1144,
+      "step": 3460
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.26718552660381e-07,
+      "loss": 0.1135,
+      "step": 3461
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2487099409199686e-07,
+      "loss": 0.1273,
+      "step": 3462
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2303090874384426e-07,
+      "loss": 0.1254,
+      "step": 3463
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2119829802269677e-07,
+      "loss": 0.1165,
+      "step": 3464
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1937316332960924e-07,
+      "loss": 0.119,
+      "step": 3465
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1755550605992216e-07,
+      "loss": 0.1148,
+      "step": 3466
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.157453276032595e-07,
+      "loss": 0.1262,
+      "step": 3467
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.139426293435265e-07,
+      "loss": 0.1123,
+      "step": 3468
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.121474126589118e-07,
+      "loss": 0.118,
+      "step": 3469
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1035967892188314e-07,
+      "loss": 0.1191,
+      "step": 3470
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.085794294991872e-07,
+      "loss": 0.1068,
+      "step": 3471
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.068066657518486e-07,
+      "loss": 0.12,
+      "step": 3472
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.05041389035171e-07,
+      "loss": 0.1294,
+      "step": 3473
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0328360069873266e-07,
+      "loss": 0.1211,
+      "step": 3474
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0153330208638412e-07,
+      "loss": 0.1155,
+      "step": 3475
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9979049453625498e-07,
+      "loss": 0.1134,
+      "step": 3476
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9805517938074615e-07,
+      "loss": 0.1144,
+      "step": 3477
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9632735794652857e-07,
+      "loss": 0.1176,
+      "step": 3478
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9460703155454564e-07,
+      "loss": 0.121,
+      "step": 3479
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.928942015200097e-07,
+      "loss": 0.1169,
+      "step": 3480
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9118886915240442e-07,
+      "loss": 0.115,
+      "step": 3481
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8949103575547912e-07,
+      "loss": 0.1041,
+      "step": 3482
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8780070262724993e-07,
+      "loss": 0.1256,
+      "step": 3483
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.861178710600009e-07,
+      "loss": 0.1207,
+      "step": 3484
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.844425423402796e-07,
+      "loss": 0.123,
+      "step": 3485
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8277471774889698e-07,
+      "loss": 0.1224,
+      "step": 3486
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8111439856092984e-07,
+      "loss": 0.1187,
+      "step": 3487
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7946158604571162e-07,
+      "loss": 0.1299,
+      "step": 3488
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7781628146684382e-07,
+      "loss": 0.1096,
+      "step": 3489
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7617848608218247e-07,
+      "loss": 0.1141,
+      "step": 3490
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7454820114384374e-07,
+      "loss": 0.1109,
+      "step": 3491
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.72925427898204e-07,
+      "loss": 0.1213,
+      "step": 3492
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7131016758589636e-07,
+      "loss": 0.1186,
+      "step": 3493
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6970242144180859e-07,
+      "loss": 0.1165,
+      "step": 3494
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6810219069508527e-07,
+      "loss": 0.1028,
+      "step": 3495
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.665094765691222e-07,
+      "loss": 0.1238,
+      "step": 3496
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6492428028157536e-07,
+      "loss": 0.1103,
+      "step": 3497
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6334660304434535e-07,
+      "loss": 0.1288,
+      "step": 3498
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6177644606359066e-07,
+      "loss": 0.1207,
+      "step": 3499
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6021381053971664e-07,
+      "loss": 0.1245,
+      "step": 3500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5865869766737986e-07,
+      "loss": 0.1303,
+      "step": 3501
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5711110863548594e-07,
+      "loss": 0.1159,
+      "step": 3502
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.555710446271863e-07,
+      "loss": 0.112,
+      "step": 3503
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5403850681988242e-07,
+      "loss": 0.1279,
+      "step": 3504
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.525134963852204e-07,
+      "loss": 0.1253,
+      "step": 3505
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5099601448909097e-07,
+      "loss": 0.1212,
+      "step": 3506
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4948606229163054e-07,
+      "loss": 0.1355,
+      "step": 3507
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4798364094721797e-07,
+      "loss": 0.1146,
+      "step": 3508
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.464887516044744e-07,
+      "loss": 0.1266,
+      "step": 3509
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4500139540626458e-07,
+      "loss": 0.1175,
+      "step": 3510
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.435215734896922e-07,
+      "loss": 0.1119,
+      "step": 3511
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.420492869861012e-07,
+      "loss": 0.1207,
+      "step": 3512
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4058453702107455e-07,
+      "loss": 0.1045,
+      "step": 3513
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3912732471443313e-07,
+      "loss": 0.1178,
+      "step": 3514
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.376776511802369e-07,
+      "loss": 0.1148,
+      "step": 3515
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3623551752678154e-07,
+      "loss": 0.1297,
+      "step": 3516
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3480092485659736e-07,
+      "loss": 0.1145,
+      "step": 3517
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3337387426645031e-07,
+      "loss": 0.1233,
+      "step": 3518
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3195436684733885e-07,
+      "loss": 0.121,
+      "step": 3519
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3054240368449821e-07,
+      "loss": 0.126,
+      "step": 3520
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2913798585739158e-07,
+      "loss": 0.1173,
+      "step": 3521
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2774111443971672e-07,
+      "loss": 0.1148,
+      "step": 3522
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2635179049940049e-07,
+      "loss": 0.1089,
+      "step": 3523
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2497001509859997e-07,
+      "loss": 0.1217,
+      "step": 3524
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2359578929370231e-07,
+      "loss": 0.1227,
+      "step": 3525
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2222911413532158e-07,
+      "loss": 0.1234,
+      "step": 3526
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2086999066829863e-07,
+      "loss": 0.1164,
+      "step": 3527
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1951841993170343e-07,
+      "loss": 0.1061,
+      "step": 3528
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1817440295883054e-07,
+      "loss": 0.1098,
+      "step": 3529
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1683794077719802e-07,
+      "loss": 0.1263,
+      "step": 3530
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1550903440854966e-07,
+      "loss": 0.1087,
+      "step": 3531
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.141876848688539e-07,
+      "loss": 0.1218,
+      "step": 3532
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1287389316829934e-07,
+      "loss": 0.1276,
+      "step": 3533
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1156766031129917e-07,
+      "loss": 0.1254,
+      "step": 3534
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1026898729648461e-07,
+      "loss": 0.1099,
+      "step": 3535
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0897787511671032e-07,
+      "loss": 0.1211,
+      "step": 3536
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0769432475904785e-07,
+      "loss": 0.1149,
+      "step": 3537
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0641833720479e-07,
+      "loss": 0.1153,
+      "step": 3538
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0514991342944758e-07,
+      "loss": 0.1262,
+      "step": 3539
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.038890544027471e-07,
+      "loss": 0.1376,
+      "step": 3540
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0263576108863194e-07,
+      "loss": 0.1127,
+      "step": 3541
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0139003444526341e-07,
+      "loss": 0.1139,
+      "step": 3542
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0015187542501525e-07,
+      "loss": 0.1128,
+      "step": 3543
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.892128497447696e-08,
+      "loss": 0.1222,
+      "step": 3544
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.769826403445371e-08,
+      "loss": 0.1189,
+      "step": 3545
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.648281353996091e-08,
+      "loss": 0.1207,
+      "step": 3546
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.52749344202264e-08,
+      "loss": 0.1215,
+      "step": 3547
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.407462759869035e-08,
+      "loss": 0.1095,
+      "step": 3548
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.288189399300429e-08,
+      "loss": 0.1298,
+      "step": 3549
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.169673451502992e-08,
+      "loss": 0.1194,
+      "step": 3550
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.05191500708369e-08,
+      "loss": 0.107,
+      "step": 3551
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.934914156070506e-08,
+      "loss": 0.1222,
+      "step": 3552
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.818670987912226e-08,
+      "loss": 0.132,
+      "step": 3553
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.703185591478314e-08,
+      "loss": 0.1146,
+      "step": 3554
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.588458055058923e-08,
+      "loss": 0.1207,
+      "step": 3555
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.474488466364894e-08,
+      "loss": 0.1185,
+      "step": 3556
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.361276912527528e-08,
+      "loss": 0.1052,
+      "step": 3557
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.248823480098477e-08,
+      "loss": 0.1145,
+      "step": 3558
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.137128255050197e-08,
+      "loss": 0.1305,
+      "step": 3559
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.026191322774934e-08,
+      "loss": 0.1144,
+      "step": 3560
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.916012768085735e-08,
+      "loss": 0.1182,
+      "step": 3561
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.80659267521533e-08,
+      "loss": 0.121,
+      "step": 3562
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.697931127817248e-08,
+      "loss": 0.1224,
+      "step": 3563
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.590028208964483e-08,
+      "loss": 0.1233,
+      "step": 3564
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.482884001150381e-08,
+      "loss": 0.1132,
+      "step": 3565
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.376498586288084e-08,
+      "loss": 0.1219,
+      "step": 3566
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.270872045710753e-08,
+      "loss": 0.1101,
+      "step": 3567
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.16600446017135e-08,
+      "loss": 0.1135,
+      "step": 3568
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.061895909842632e-08,
+      "loss": 0.116,
+      "step": 3569
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.958546474316818e-08,
+      "loss": 0.1206,
+      "step": 3570
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.855956232606153e-08,
+      "loss": 0.1225,
+      "step": 3571
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.75412526314212e-08,
+      "loss": 0.1118,
+      "step": 3572
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.653053643776108e-08,
+      "loss": 0.1197,
+      "step": 3573
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.552741451778643e-08,
+      "loss": 0.1178,
+      "step": 3574
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.453188763839824e-08,
+      "loss": 0.1211,
+      "step": 3575
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.354395656068879e-08,
+      "loss": 0.1037,
+      "step": 3576
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.256362203994615e-08,
+      "loss": 0.1138,
+      "step": 3577
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.159088482564968e-08,
+      "loss": 0.1119,
+      "step": 3578
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.062574566147117e-08,
+      "loss": 0.1013,
+      "step": 3579
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.96682052852715e-08,
+      "loss": 0.1178,
+      "step": 3580
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.871826442910289e-08,
+      "loss": 0.117,
+      "step": 3581
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.777592381920882e-08,
+      "loss": 0.1119,
+      "step": 3582
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6841184176021915e-08,
+      "loss": 0.1287,
+      "step": 3583
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.591404621416274e-08,
+      "loss": 0.1043,
+      "step": 3584
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.499451064244321e-08,
+      "loss": 0.1136,
+      "step": 3585
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.408257816385764e-08,
+      "loss": 0.115,
+      "step": 3586
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3178249475592805e-08,
+      "loss": 0.1051,
+      "step": 3587
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.228152526902008e-08,
+      "loss": 0.1152,
+      "step": 3588
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.139240622969666e-08,
+      "loss": 0.1093,
+      "step": 3589
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.051089303736767e-08,
+      "loss": 0.1221,
+      "step": 3590
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9636986365961814e-08,
+      "loss": 0.1093,
+      "step": 3591
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8770686883591326e-08,
+      "loss": 0.1152,
+      "step": 3592
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.79119952525553e-08,
+      "loss": 0.108,
+      "step": 3593
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.706091212933528e-08,
+      "loss": 0.1301,
+      "step": 3594
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.621743816459523e-08,
+      "loss": 0.109,
+      "step": 3595
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.538157400318266e-08,
+      "loss": 0.1198,
+      "step": 3596
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.455332028412862e-08,
+      "loss": 0.1287,
+      "step": 3597
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.373267764064215e-08,
+      "loss": 0.1171,
+      "step": 3598
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2919646700116945e-08,
+      "loss": 0.1184,
+      "step": 3599
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2114228084126904e-08,
+      "loss": 0.1133,
+      "step": 3600
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.131642240842504e-08,
+      "loss": 0.1111,
+      "step": 3601
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0526230282944556e-08,
+      "loss": 0.1151,
+      "step": 3602
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9743652311798886e-08,
+      "loss": 0.1173,
+      "step": 3603
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8968689093279444e-08,
+      "loss": 0.1259,
+      "step": 3604
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.820134121985453e-08,
+      "loss": 0.1172,
+      "step": 3605
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.744160927817486e-08,
+      "loss": 0.1291,
+      "step": 3606
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6689493849063616e-08,
+      "loss": 0.1124,
+      "step": 3607
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5944995507525284e-08,
+      "loss": 0.1125,
+      "step": 3608
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.520811482273678e-08,
+      "loss": 0.1031,
+      "step": 3609
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.447885235805637e-08,
+      "loss": 0.108,
+      "step": 3610
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.375720867101251e-08,
+      "loss": 0.1292,
+      "step": 3611
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.304318431331277e-08,
+      "loss": 0.1267,
+      "step": 3612
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.2336779830839384e-08,
+      "loss": 0.1139,
+      "step": 3613
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.163799576364701e-08,
+      "loss": 0.1216,
+      "step": 3614
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.09468326459661e-08,
+      "loss": 0.111,
+      "step": 3615
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.026329100620173e-08,
+      "loss": 0.1136,
+      "step": 3616
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.958737136692813e-08,
+      "loss": 0.1174,
+      "step": 3617
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8919074244896372e-08,
+      "loss": 0.1182,
+      "step": 3618
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8258400151029987e-08,
+      "loss": 0.1204,
+      "step": 3619
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7605349590421603e-08,
+      "loss": 0.1248,
+      "step": 3620
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.695992306233852e-08,
+      "loss": 0.1103,
+      "step": 3621
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6322121060217142e-08,
+      "loss": 0.1157,
+      "step": 3622
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.569194407166742e-08,
+      "loss": 0.1095,
+      "step": 3623
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.506939257846619e-08,
+      "loss": 0.1208,
+      "step": 3624
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4454467056564958e-08,
+      "loss": 0.1215,
+      "step": 3625
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.3847167976081e-08,
+      "loss": 0.1155,
+      "step": 3626
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.324749580130292e-08,
+      "loss": 0.1131,
+      "step": 3627
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2655450990688443e-08,
+      "loss": 0.1221,
+      "step": 3628
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2071033996864387e-08,
+      "loss": 0.1186,
+      "step": 3629
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.149424526662669e-08,
+      "loss": 0.1264,
+      "step": 3630
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.092508524093595e-08,
+      "loss": 0.111,
+      "step": 3631
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.03635543549241e-08,
+      "loss": 0.1144,
+      "step": 3632
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9809653037888844e-08,
+      "loss": 0.1242,
+      "step": 3633
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.926338171329478e-08,
+      "loss": 0.101,
+      "step": 3634
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.872474079877451e-08,
+      "loss": 0.1058,
+      "step": 3635
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.819373070612751e-08,
+      "loss": 0.1201,
+      "step": 3636
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7670351841315715e-08,
+      "loss": 0.1154,
+      "step": 3637
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.715460460447238e-08,
+      "loss": 0.1196,
+      "step": 3638
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6646489389890995e-08,
+      "loss": 0.1126,
+      "step": 3639
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.614600658603305e-08,
+      "loss": 0.1291,
+      "step": 3640
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5653156575525795e-08,
+      "loss": 0.1122,
+      "step": 3641
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5167939735157842e-08,
+      "loss": 0.1083,
+      "step": 3642
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4690356435886899e-08,
+      "loss": 0.1184,
+      "step": 3643
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4220407042828677e-08,
+      "loss": 0.1158,
+      "step": 3644
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3758091915268002e-08,
+      "loss": 0.137,
+      "step": 3645
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3303411406651034e-08,
+      "loss": 0.1348,
+      "step": 3646
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2856365864586385e-08,
+      "loss": 0.1139,
+      "step": 3647
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2416955630847327e-08,
+      "loss": 0.1292,
+      "step": 3648
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1985181041368476e-08,
+      "loss": 0.1265,
+      "step": 3649
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1561042426248003e-08,
+      "loss": 0.1167,
+      "step": 3650
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1144540109744306e-08,
+      "loss": 0.1205,
+      "step": 3651
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.073567441028156e-08,
+      "loss": 0.1289,
+      "step": 3652
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0334445640440839e-08,
+      "loss": 0.1251,
+      "step": 3653
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.94085410696899e-09,
+      "loss": 0.1105,
+      "step": 3654
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.55490011076976e-09,
+      "loss": 0.1087,
+      "step": 3655
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.176583946912676e-09,
+      "loss": 0.1166,
+      "step": 3656
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.805905904624157e-09,
+      "loss": 0.1206,
+      "step": 3657
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.44286626729307e-09,
+      "loss": 0.1101,
+      "step": 3658
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.087465312468513e-09,
+      "loss": 0.122,
+      "step": 3659
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.739703311859803e-09,
+      "loss": 0.1083,
+      "step": 3660
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.399580531334272e-09,
+      "loss": 0.1105,
+      "step": 3661
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.067097230920583e-09,
+      "loss": 0.1147,
+      "step": 3662
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.74225366480874e-09,
+      "loss": 0.1251,
+      "step": 3663
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.42505008134453e-09,
+      "loss": 0.1301,
+      "step": 3664
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.11548672303397e-09,
+      "loss": 0.1209,
+      "step": 3665
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.81356382654441e-09,
+      "loss": 0.1176,
+      "step": 3666
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.519281622697881e-09,
+      "loss": 0.124,
+      "step": 3667
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.232640336478856e-09,
+      "loss": 0.1126,
+      "step": 3668
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.953640187027597e-09,
+      "loss": 0.1134,
+      "step": 3669
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.682281387643484e-09,
+      "loss": 0.1366,
+      "step": 3670
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.418564145785009e-09,
+      "loss": 0.1077,
+      "step": 3671
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.162488663066455e-09,
+      "loss": 0.1017,
+      "step": 3672
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.914055135261219e-09,
+      "loss": 0.1174,
+      "step": 3673
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.6732637523018143e-09,
+      "loss": 0.1122,
+      "step": 3674
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.4401146982743216e-09,
+      "loss": 0.1093,
+      "step": 3675
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.214608151425047e-09,
+      "loss": 0.1073,
+      "step": 3676
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9967442841571936e-09,
+      "loss": 0.1181,
+      "step": 3677
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.78652326303086e-09,
+      "loss": 0.1087,
+      "step": 3678
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5839452487630425e-09,
+      "loss": 0.1218,
+      "step": 3679
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.389010396226521e-09,
+      "loss": 0.1203,
+      "step": 3680
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.201718854452084e-09,
+      "loss": 0.1111,
+      "step": 3681
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.022070766626305e-09,
+      "loss": 0.1063,
+      "step": 3682
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8500662700937644e-09,
+      "loss": 0.1121,
+      "step": 3683
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6857054963537177e-09,
+      "loss": 0.1256,
+      "step": 3684
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5289885710612073e-09,
+      "loss": 0.1255,
+      "step": 3685
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3799156140292814e-09,
+      "loss": 0.11,
+      "step": 3686
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.2384867392267742e-09,
+      "loss": 0.1304,
+      "step": 3687
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1047020547771959e-09,
+      "loss": 0.1091,
+      "step": 3688
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.785616629620632e-10,
+      "loss": 0.1193,
+      "step": 3689
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.600656602153479e-10,
+      "loss": 0.1157,
+      "step": 3690
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.492141371301387e-10,
+      "loss": 0.1388,
+      "step": 3691
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.460071784541999e-10,
+      "loss": 0.1226,
+      "step": 3692
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.504448630899717e-10,
+      "loss": 0.1125,
+      "step": 3693
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.625272640967904e-10,
+      "loss": 0.1267,
+      "step": 3694
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.822544486886681e-10,
+      "loss": 0.1144,
+      "step": 3695
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.096264782342928e-10,
+      "loss": 0.1284,
+      "step": 3696
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.446434082603588e-10,
+      "loss": 0.1288,
+      "step": 3697
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.873052884460158e-10,
+      "loss": 0.1124,
+      "step": 3698
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3761216262841991e-10,
+      "loss": 0.122,
+      "step": 3699
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.556406879718261e-11,
+      "loss": 0.1175,
+      "step": 3700
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.116103909881154e-11,
+      "loss": 0.1248,
+      "step": 3701
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.440309983560042e-11,
+      "loss": 0.1195,
+      "step": 3702
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.529027146451867e-11,
+      "loss": 0.1144,
+      "step": 3703
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.822568597211529e-12,
+      "loss": 0.1133,
+      "step": 3704
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.1252,
+      "step": 3705
+    },
+    {
+      "epoch": 3.0,
+      "step": 3705,
+      "total_flos": 2.499052997015765e+18,
+      "train_loss": 0.25374489042118653,
+      "train_runtime": 24443.2256,
+      "train_samples_per_second": 19.406,
+      "train_steps_per_second": 0.152
+    }
+  ],
+  "max_steps": 3705,
+  "num_train_epochs": 3,
+  "total_flos": 2.499052997015765e+18,
+  "trial_name": null,
+  "trial_params": null
+}